Thermal stratification and the breakup of nocturnal inversions in valleys can strongly vary with the seasonal snow-cover change. When snow is absent or patchy, nocturnal inversions usually form and break up during the day, while they may persist in snow-covered valleys during wintertime when surface heating is not sufficient. The East River Valley is a high-altitude valley in the Colorado Rocky Mountains. Due to its importance for the watershed hydrology of the Colorado River Basin, this valley was chosen as the study area for the Surface-Atmosphere Integrated field Laboratory (SAIL) and Study of Precipitation, the Lower Atmosphere and Surface for Hydrometeorology (SPLASH) campaigns. Several passive remote sensing instruments, including three infrared spectrometers (two AERIs and one ASSIST) were simultaneously deployed along the valley axis during a three-month period starting in October 2021. Using an optimal estimation physical retrieval framework (TROPoe), temperature profiles were retrieved with sub-hourly temporal resolution. This instrument combination provides a unique opportunity to study the spatio-temporal responses of the boundary layer thermal structure in a high-altitude valley to the seasonal snow-cover transition. In a second step, we use this data set to evaluate how well NOAA’s operational High-resolution-Rapid-Refresh (HRRR) forecast model captures the main boundary layer characteristics. In 2021, the initial change from patchy to homogeneous snow cover occurred after a multi-day snow fall event at the beginning of December and was associated with a jump of surface albedo values from around 0.2 to around 0.9 and a change from positive to negative daily mean net radiation and sensible heat flux. Despite a drop in maximum near-surface air temperatures to near-freezing values during the snow-covered regime, a shallow daytime convective boundary layer (CBL) of a few hundred meters still developed. Strong nocturnal surface inversions formed independent of snow cover. Over patchy snow, the inversion was mixed out by a CBL of around 1000 m depth, while under snow-covered conditions a low-level inversion persisted during daytime above a shallow CBL. The HRRR model failed to capture the nocturnal surface inversions during both regimes and the persistent low-level inversion during the snow-covered regime, leading to very large warm biases (nearly 15 °C). We attribute this to the failure of the model to correctly simulate the thermally driven circulations, in particularly the nocturnal cold drainage flows, which can likely be explained by the model’s relatively coarse horizontal resolution (3 km) and its inability to fully represent the valley’s topographic structure.

The planetary boundary layer height (PBLH) is a key parameter for air quality control, visibility forecasting, and to understand turbulent exchange between the surface and the atmosphere. A large number of methods have been developed to estimate the PBLH, largely based on profiles of thermodynamic or aerosol variables. Although such methods show good agreement for convective boundary layers, estimations of the PBLH for stable boundary layers show lower consistency. The performances of common PBLH algorithms were evaluated using profiles acquired during the CLOUDLAB field campaign, with a focus on stable boundary layers at night and in winter. The algorithms based on thermodynamic profiles (potential temperature gradient, relative humidity gradient, parcel method, bulk Richardson method, temperature inversions) were applied to measurements from radiosondes, drones and a microwave radiometer. The success of these algorithms is strongly dependent on the accuracy and resolution of the profiles to which they are applied. Profiles and PBLH estimations from the microwave radiometer were compared to those of the unmanned aerial vehicles and radiosondes during intense observation periods in the winters (December to February) of 2021—2022 and 2022—2023. Algorithms based on aerosol backscatter profiles (attenuated backscatter gradient, STRATfinder) were applied to a ceilometer. In addition, a wind profiler was used to investigate the presence of low–level jets within the boundary layer and the influence of such jets on the PBLH. The field site (Eriswil Rapierplatz, 47.071◦N, 7.874◦E) is situated at 920 m above sea level in rolling hills. This study therefore provides a test of PBLH algorithms in terrain which is intermediate between ideal horizontally–homogeneous and truly ‘complex’ mountainous.

The Cold Fog Amongst Complex Terrain (CFACT) Project, sponsored by the US National Science Foundation (NSF), investigates the life cycle of cold fog in an elevated mountain basin. The CFACT field campaign was conducted in Heber Valley, Utah, during January and February 2022 with support from the NSF Lower Atmospheric Observing Facilities (LAOF), managed by NCAR’s Earth Observing Laboratory (EOL). A network of ground-based in-situ instruments and a suite of remote sensing platforms complemented by a tethered balloon system were deployed to observe the vertical structure as well as the spatial variability of the cold fog conditions in the basin’s atmosphere. The observations were designed to capture the complex thermodynamic, dynamic, and microphysical processes that interact to affect fog formation, maintenance, and dissipation. Comprehensive data analysis and NWP model prediction verification are currently being performed. This presentation will first give a brief overview over the CFACT campaign and its overarching goals, including observational strategies as well as modeling and data assimilation endeavors. Then, mountain-meteorological processes modulating fog formation will be discussed. These processes include the formation of a basin cold pool, aerosol-microphysics processes, interacting thermally-drive flows, and gravity waves. Results from a novel method to directly measure longwave radiative cooling from a tethered balloon-based platform will be presented, which show the radiative cooling impact on fog occurrence. In summary, the importance of thermodynamic, dynamic, and microphysical processes on the life cycle of cold fog, and its numerical forecasting issues, will be highlighted for three fog events.

Three-dimensional turbulent processes in complex terrain are manifold and crucial for the exchange of energy in the mountain boundary layer. In preparation of the TEAMx observational campaign which is currently planned for the years 2024-2025, a site was identified in 2022 which can be well suited for the investigation of a multitude of thermodynamic processes in Alpine valley systems. The Nafingalm site with a valley base altitude of 1920 m can be reached through the Weer valley, a tributary to the Inn valley in Tirol (Austria). Its size of approximately 2x2 km with an altitude difference of 350-500 m and the North-South orientation with only low vegetation on its 30° slopes towards East, West and South make it a good site to investigate valley wind systems, as well as slope winds. Measuring all the relevant scales within the ABL in the valley requires distributed sensors at the ground, but also aloft, ideally up to the boundary layer height. For this purpose, within the ESTABLIS-UAS project, a fleet of quadrotor uncrewed aerial systems (UAS), is enabled to conduct simultaneous, distributed measurements. The SWUF-3D fleet has already been deployed at the FESSTVaL campaign for calibration and validation and to investigate correlation and coherence in a flat, but heterogeneous terrain. In the pre-campaign towards TEAMx (TEAMx-PC22), which is described here, a goal was to investigate if the systems as used in the SWUF-3D fleet are suitable for deployment in Alpine terrain and if the resolution and accuracy is sufficient to capture valley and slope winds at the site. From 20-28 June 2022, three UAS were operated simultaneously in different configurations and flight strategies for this purpose. Additionally, for a longer period from June through September, ground-based instrumentation was deployed to get a better statistical understanding of the processes in the valley. The flight strategies that were pursued with the UAS were simultaneous vertical profiles along the valley up to 120m above ground, cross-section flights across the valley and distributed hover position flights to capture turbulence statistics. We will show first results of statistics and specific events during the TEAMX-PC22 at the Nafingalm.

CACTI field campaign in Argentina (2018) are examined. Clear-air returns from the Department of Energy Atmospheric Radiation Measurement (ARM) radars (CSAPR2, XSACR, KASACR) are used to characterize the structure and variability of the ridge-normal (i.e., up/downslope) flow components (extracted from east-west radar scans), which transport mass to the SDC crest and contribute to convective initiation. Data are compiled for the entire project period, including days with clear skies, shallow cumuli, cumulus congestus, and deep convection. To examine shared variability amongst the 1000s of RHI scans we use (a) a principal component analysis (PCA) to isolate modes of variability in the upslope mass transport, and (b) hourly composite analysis based on convective outcomes, which are determined from GOES16 satellite observations and KASACR cloud radar reflectivity. These data are further contextualized with observed surface sensible heat fluxes and meteorological state variables from along-slope weather stations. Results indicate distinct diurnally-varying thermally-forced upslope flow modes and synoptically modulated (e.g., no diurnal preference) mechanically-forced modes. In some instances there is a super-position of thermal and mechanical forcing, yielding either deeper upslope flow (constructive superposition), or flow convergence over the eastern slopes of the SDC (destructive superposition). The composite analyses show that successively deeper convective outcomes are associated with successively deeper upslope flow layers that more readily transport mass to the ridge crest in conjunction with lower lifting condensation levels, facilitating convective initiation.

Turbulence is the main mechanism through which the surface and the atmosphere communicate, and is thus the key component of the climate system. Correctly representing this interaction is therefore of fundamental importance in weather, climate and air pollution models. Yet the parametrizations of this surface-atmosphere exchange still rely on Monin-Obukhov similarity theory (MOST) despite its assumptions being violated over majority of the Earth’s surface. In addition, turbulence is generally assumed to be isotropic in most applications, despite the known anisotropic nature of turbulence forcing. In this work we introduce why the directionality of turbulence exchange (anisotropy of the Reynolds stress tensor) is fundamental in understanding surface-layer atmospheric turbulence, and present how its inclusion into MOST can provide a first generalized extension of MOST encompassing a wide range of realistic surface and flow conditions. The novel theory shows that the constants in the classical MOST, are actually functions of anisotropy, allowing a seamless transition between classical MOST and the novel generalization. The new scaling relations, based on measurements from 14 well-known datasets ranging from flat to highly complex mountainous terrain, show substantial improvements to scaling under all stratifications. The results also highlight the role of anisotropy in explaining general characteristics of complex terrain and strongly unstable and stable turbulence, adding to the mounting evidence that anisotropy fully encodes the information on the complexity of the boundary conditions.

Flux-gradient relations from Monin Obukhov Similarity theory (MOST) are used in most numerical models to predict the turbulent fluxes from profiles of mean temperature and wind. These relations, derived under the assumption of homogeneous and flat terrain, are affected by significant error in very stable and very unstable regimes (|ζ|>1), where the literature disagrees on their formulations. This is especially true over complex terrain where MOST generally fails. In this work we explore the validity of MOST relations for mean temperature and wind speed profiles in a local-scaling sense, from five well-known turbulence datasets ranging from flat and weakly heterogeneous terrain to more complex terrain, and covering the entire stability range. The results show departures from the universal relations, as well as large scatter. We then show that turbulence anisotropy can explain the scatter and shed light on what formulation is correct. Finally, new parametrizations are provided, that account for turbulence anisotropy. The new scaling relations improve the accuracy of our representation of the surface layer up to 77 % depending on the regime and variable considered (percentage of median absolute deviation improvement from Högstrom, 1996).

Spectral scaling has been used to study the low frequency range of atmospheric spectra in the 70’s, for both spectra of velocity components and temperature as well as co-spectra, over a flat surface [1]. Stability parameter z/L (where z is the measurement height and L is the Obukhov length) has been shown to be the main factor responsible for the low frequency spreading of these spectra in stably stratified conditions (z/L > 0). However, it is well known that scaling of horizontal velocity variances is one of the biggest failure in Monin-Obukhov Similarity Theory, even over flat terrains. For z/L < 0, atmospheric stability thus fails in explaining the low frequency spreading of spectra, except for the vertical velocity component. Since then, some refinements have been proposed like the approach of [2] who derive laws for the scaled variances of velocity as a function of zi/L, zi being the mixed-layer height. More recently, [4] used a so-called ”Multi-point Monin-Obukhov similarity” where both the vertical and the horizontal length scales scaled normalized by L are important. However, these approaches are limited to flat terrains and are thus not applicable to the increasingly studied complex and mountainous terrains. In the present work, we use data from 14 field experiments, both over flat and complex terrains, in unstably stratified conditions (−2≤ z/L≤ 0). We show that turbulence anisotropy is the main factor able to characterize the spectral energy at low frequency for the horizontal velocity components and the temperature spectra, as well as the momentum and heat flux co-spectra. Given that there exists a link between turbulence anisotropy and atmospheric stability, one might think that there is a bias linked to the repartition of turbulence anisotropy intensity within the stability range. Based on [3], that relates scaled variances and dissipation rate with both turbulence anisotropy and atmospheric stability, we were able to separate the contribution of turbulence anisotropy from the one of stability alone. While clustering of spectra by stability conditions in [1] was not able to explain the spreading of low frequency, z/L is still a second order factor not negligible and is able to refine the analysis inside a cluster of turbulence anisotropy. Its impact increases as anisotropy decreases. [1] Kaimal et al. (1972), QJRMS, 98, 563-589 (1972) [2] Panofsky et al. (1977), BLM, 11(3), 355-361 [3] Stiperski and Calaf (2023), PRL [4] Tong and Nguyen (2015), JAS, 72(11), 4337-4348

A uniform approximation of flow over gentle hills with a turbulent closure based on mixing length theory is derived. It permits to describe the transition from neutral to stratified flow in the production of mountain drag. Our results corroborate previous studies showing that the transition from the form drag associated to the mountain induced changes in boundary layer friction to the mountain gravity waves drag can be captured by theory. We also confirm that the first is associated with downstream sheltering with relative acceleration at the hill top, the second with upstream blocking with strong downslope winds. We also show that the downslope winds penetrate well into the inner layer. The theory show that the altitude at which the incident flow need to be taken to calculate the drag is related to the inner layer depth at which dissipative effects equilibrate disturbance advection. We also show that the parameter that capture the transition, which in our case is a Richardson number, is directly related to the altitude of the turning levels of the gravity waves with respect to the mountain length. Our uniform solutions are also used to describe the wave field aloft and the distribution of the Reynolds stress in the vertical. Some directions to combine neutral and stratified effects in the parameterization of subgrid scale orographies in large scale models are given.

The essential dynamics responsible for orographic precipitation during warm-sector passage of cyclones is often thought to be captured by models in which a horizontally uniform unidirectional flow of moist air is forced to ascend the windward slope of a ridge. In this paradigm, the role of the large-scale storm system is to set the upstream environmental conditions. We compare the integrated vapor transport (IVT) and resulting precipitation in published observations of real-world events with the same quantities in previous studies of idealized moist airstreams impinging on a ridge. We find that observed events produce substantially more orographic precipitation than idealized numerical studies of shear flow cases with the same incoming IVT. These shear-flow cases use idealized 2D, 3D, or real topography, and are driven by moist horizontally uniform environmental winds specified by initial and upstream boundary conditions. We reproduced this discrepancy in a pair of more complex yet still idealized simulations. We simulated a realistic, prototypical mid-latitude cyclone encountering an isolated mountain ridge in a baroclinically unstable westerly flow. The cyclone is initialized with a PV anomaly upstream of a 1-km high north-south ridge and passes north of the terrain when the cyclone is mature. Warm-sector conditions occur over the ridge for more than 10 hours. A pre-frontal warm-sector sounding taken from this simulation upstream of the center of the ridge was used to initialize a second horizontally uniform shear flow simulation, in which the airstream encounters the same 1-km ridge. Although the shear-flow experiment had slightly larger sustained IVT than the cyclone-plus-mountain warm-sector environment, it produced only 1/4 as much precipitation over the terrain. Possible sources of the discrepancy between this pair of simulations arise from microphysics or dynamics. On the microphysical side, the seeder-feeder mechanism is not active in our basic shear-flow simulation, but by artificially adding ice cloud aloft, we show seeder-feeder does not produce nearly enough enhancement to account for the difference in precipitation. Instead, it is dynamically driven by a substantial contrast in the water-vapor-flux convergence over the upstream slope in the two simulations. This contrast arises from differences in the cross-mountain wind speed at the ridge crest that occur because the shear-flow experiment cannot reproduce the mountain-wave amplitude and downstream environment present in the full mountain+cyclone simulation. Our results illustrate how care must be taken when formulating the mathematical problem statement for simulations of isolated meteorological processes.

Observations have shown that shallow orographic convection can organize into either scattered cells or quasi-stationary bands. The latter morphology, which is far less frequent than the former, tends to enhance flash-flooding risk by focusing rainfall over narrow mountain watersheds. While previous studies have explained the mechanisms by which such bands form, they have not determined the environmental conditions that distinguish bands from cells. This deficiency limits the ability of local forecasters to warn their communities of potentially hazardous meteorological events. To address this knowledge gap, this study performs a five-year radar-based climatology of cool-season convection over the Oregon Coastal Range. All convection events are categorized as cellular or banded, and a composite upstream sounding is created for each type. To assess whether differences in the two soundings explain the differences in cloud morphology, these soundings are used to initialize high-resolution quasi-idealized simulations of moist flows over the Coastal Range. When run over a free-slip surface, both the cellular and banded soundings give rise to stationary cloud bands, suggesting that the soundings alone do not dictate the cloud morphology. A more important factor, suggested by buoy observations, is the upstream ocean—air temperature contrast: cellular events tend to have warmer ocean SSTs than surface air temperatures, while banded events tend to have the opposite. Additional numerical experiments reveal that positive ocean—air temperature differences give rise to stronger turbulence in the impinging boundary layer, which readily seeds cellular moist convection initiation over the windward slope. The large amplitudes of these cells dominate the weaker standing lee-wave circulations nominally responsible for band formation. In contrast, when the ocean is cooler than the air, the boundary-layer turbulence is much weaker and does not interfere with the lee-wave convection initiation, allowing convective bands to develop and persist.

Several studies have employed models and observations to examine the connection between mountain waves and orographic precipitation. The vast majority of these studies have focused on single, quasi two-dimensional ridges. However, most terrain configurations are more complex. One form of this complexity is seen in topography characterized by multiple adjacent ridges that are each relatively narrow compared to larger mountain barriers such as the Sierra Nevada, Cascades and Alps. A comparatively small number of studies have explored the linkages between mountain waves and orographic precipitation in association with narrow, multi-ridge orography. Some recent studies have investigated these linkages over the multi-ridge orography of western Idaho (USA) with data from the 2017 SNOWIE field campaign. The terrain associated with SNOWIE sampling is characterized by several steep and narrow ridgelines separated by 10-20 km that extend along an incline from the Snake River Plain up to the Idaho Central Mountains. Another category of multi-ridge orography involves alternating sequences of mountain ranges and broad valleys (i.e., basins) where the approximately parallel ridges are separated by ~20-40 km and not superimposed on larger mountain barriers. This “basin and range topography” exists across a large swath of the semi-arid western USA, where the mountains of the region can receive >1 m of liquid equivalent precipitation annually while the adjacent basins receive significantly less. The present study examines mountain waves and orographic precipitation for a winter storm passing over basin and range topography near Salt Lake City, Utah (USA) on 22 March 2019. Specifically, observations from the W-band Wyoming Cloud Radar onboard the University of Wyoming King Air (UWKA) research aircraft are used to document horizontal and vertical motions along with precipitation structure in two-dimensional, zonally oriented vertical planes that extend 60-80 km from west of the Oquirrh Range to east of the central Wasatch. Vertically propagating mountain waves are evident above and downstream of the Oquirrh’s and central Wasatch. Observed profiles of the Scorer parameter do not indicate a layer where wave energy can be trapped. Therefore, it is hypothesized that waves in the lee of both barriers are of the dispersive, non-hydrostatic variety, which is consistent with the observed decrease of wave amplitude downstream of the barriers. Clear correlations between these mountain waves and orographic precipitation patterns are not evident. A mesoscale precipitation band passing through the region during airborne sampling may have masked this correlation.

Global climate and numerical weather prediction models are unable to explicitly represent convection due to their insufficient horizontal resolution. They therefore rely on sub-grid parameterisation schemes to represent the mean effects of convection within each grid-box. Such schemes were typically developed and tuned for horizontally homogeneous and flat surfaces. This work investigates the need for convection schemes to account for unresolved orography, using MetUM simulations of the Kerala floods case of August 2018. This was a case of significant orographic rainfall where moist, westerly low-level monsoon flow ascended the western Ghats. The global model failed to capture the most extreme rainfall, and underestimated the amount of rain over land for this case. Global MetUM simulations were compared against a high resolution (330m) resolution Limited Area Model (LAM) simulation of the event, which explicitly represents the convection. While the microphysics scheme does produce some rain over the Western Ghats, rain produced by the convection scheme dominates over the mountains, and is the only source of rain over the sea upstream of India. Rain produced by both the microphysics and convection scheme increases with model resolution, but none of the global models manage to capture the full magnitude of the orographic rainfall produced by the LAM. This was true for a number of Global Atmosphere configurations, including a simulation with the new COMORPH scheme. In order to test the sensitivity to orography, high resolution simulations were also performed for this case using a degraded-resolution orography similar to that found in the global models. This significantly reduced the convective precipitation over land in the simulation. A comparison of vertical velocity cross-sections from simulations of varying orography resolution show that steep fine-scale slopes can act as a source of initiation of convective updraughts, and this source is missing over the degraded orography. In this way, fine scale orography focuses the rainfall, with greatly enhanced rain over the first windward peak and reduced rainfall either side. The overall effect is an increase in mean rainfall over land in the presence of the fine-scale orography. Now that the need for a sub-grid orographic convection scheme has been demonstrated, possible methods for doing this are being investigated.

Observations show the occurrence of two flow regimes in monsoonal rainfall over the Western Ghats - an onshore and an offshore phase. Using observations, a Froude number classification is shown to successfully explain the two phases. Under low Froude number conditions onshore winds are weak, with a strong diurnal cycle of thermally-driven flow over the orography leading to a diurnal cycle of conection. Under these conditions a nocturnal offshore propagation of rainfall is observed, with weaker rain over the Western Ghats due to flow blocking. At high Froude numbers the flow blocking is weak leading to the onshore phase with rainfall enhanced over the Western Ghats and a much weaker diurnal cycle in precipitation. A two-week WRF simulation was conducted to explore the representation of the onshore and offshore phases in the model at various resolutions - 1.33 and 4km (explicit convection) and 12km (parametrised convection and explicit convection). The offshore and onshore periods differed in their large scale synoptic conditions. At all resolutions the model overestimates the rainfall over the Western Ghats as it consistently simulates too warm a boundary layer at the coast which weakens the flow blocking by the mountains. The higher resolution simulations with explicit convection capture the onshore-offshore transition, although the contrast between rainfall over the land and sea is not strong enough. The 12km simulation with parametrised convection does not capture the regime transition, producing light rainfall along the coast throughout the simulation. The 12km simulation with explicit convection does capture the transition, however the rainfall intensity was too high. These results suggest that a km-scale simulation with explicit convection is needed to correctly represent the rainfall modes, however further work is needed to improve the representation of the boundary layer to accurately capture the land-sea differences in rainfall.

Landfalling lake- and sea-effect (hereafter lake-effect) systems often interact with complex coastlines and downstream orography, altering the distribution and intensity of precipitation. In this presentation we synthesize findings from observational and modeling studies of lake-effect systems over the Sea of Japan and the formidable downstream orography of the Japanese Archipelago and Lake Ontario and the 500-m high Tug Hill Plateau. Japanese meteorologists distinguish between Satoyuki storms with heavy lowland (coastal) snowfall and Yamayuki storms with heavy mountain snowfall. In central Honshu, snowfall is frequently produced by broad-coverage lake-effect systems with longitudinal-mode precipitation bands associated with horizontal roll convection. An analysis of nine winters of radar data from the Japanese Meteorological Agency radar on the central coast of Honshu and real-data simulations of a major multi-day event illustrate that the inland penetration and orographic enhancement of lake-effect precipitation are strongly dependent on the boundary layer wind speed and sea-induced Convective Available Potential Energy. Higher values favor inland lake-effect penetration and a large ratio of upland to lowland precipitation. These storm characteristics are well described by the non-dimensional mountain height, H, with H < 1 associated with heavy mountain snowfall and H > 1 associated with flow blocking, coastal-front development, and lowland snowfall. Oval-shaped Lake Ontario frequently produces intense long-lake-axis-parallel (LLAP) precipitation bands during westerly flow. Real-data simulations show that bulges in the south shoreline can produce two land-breeze fronts, one that serves as the locus for LLAP-band formation and another that forms downstream, cuts obliquely across the LLAP band, and enhances precipitation over the Tug Hill Plateau. Sensitivity studies show that an inland precipitation maximum would exist without orography, but the Tug Hill Plateau enhances precipitation through orographic ascent and reduced sub-cloud sublimation. Idealized numerical simulations indicate that broad-coverage lake-effect systems experience considerable orographic enhancement during flow over large topographic features due to increased convective vigor. In contrast, an inland precipitation maximum appears ubiquitous to LLAP bands even in the absence of orography as unmodified continental airstreams circumscribe the oval lake and form a cold pool downstream of the lake. Ascent at the coastal front at the leading edge of this cold pool enhances hydrometeor growth within the LLAP band, with hydrometeor transport and fallout creating the inland maximum. The addition of a plateau only weakly enhances this inland precipitation, whereas larger topographic obstacles shift the coastal front and associated precipitation offshore with little mountain precipitation.

A new lightning-flash and convective initiation climatology is developed over the Alpine area, one of the hotspots for lightning activity in Europe. The climatology uses cloud-to-ground (CG) data from the European Cooperation for LIghtning Detection (EUCLID) network, occurring from 2005 to 2019. The CG lightning data are gridded at a resolution of approximately 2km and 10min. A new and simple method of identifying convective initiation (CI) events applies a spatiotemporal mask to the CG data to determine CI timing and location. Although the method depends on a few empirical thresholds, sensitivity tests show the results to be robust. The maximum activity for both CG flashes and CI events is observed from mid-May to mid-September, with a peak at the end of July; the peak in the diurnal cycle occurs in the afternoon. CI is mainly concentrated over and around the Alps, particularly in northern and northeastern Italy. Since many thunderstorms follow the prevailing mid-latitude westerly flow, a peak of CG flashes extends from the mountains into the plains and coastal areas of northeastern Italy and Slovenia. CG flashes and CI events over the sea/coast occur less frequently than in plains and mountains, have a weaker diurnal cycle, and have a seasonal maximum in autumn instead of summer.

A recurrent weather situation in alpine meteorology is the spillover of precipitation to the lee of an orographic barrier. This study presents a climatology of precipitation spillover events on the southern side of the Swiss Alps associated to northerly synoptic flows. The mesoscale atmospheric conditions influencing the spillover events are also investigated. The findings are based on CombiPrecip data (Quantitative precipitation estimation in Switzerland combining rain gauge measurements and radar estimates) over the period 2012-2022 and ground stations data. Results show that the transport of precipitation is more frequent and widespread during north-westerly events while it is more rare when the wind blows from north and northeast. Spillover events are more frequent during winter and their diurnal distribution presents a minimum around noon, more accentuated during spring. Spillover is also enhanced during stronger flows and warm temperature advection cases. This study aims at providing a better knowledge of the mechanisms controlling spillover events during northerly flows over the Alps, in order to improve forecasting capability and to provide a reference for numerical model verification and development.

Clouds and precipitation over mountainous terrain are a challenge for models and observations alike. In this contribution, we present a first analysis of the unique, nearly one-decade long dataset of collocated microwave radiometer, cloud radar, ceilometer, and auxiliary observations collected at the Environmental Research Station Schneefernerhaus (UFS). Located at 2650 m a.s.l. just 300 m below the summit of Zugspitze, Germany's highest mountain, this dataset allows a combined view on water vapor, clouds, and precipitation. Annual and diurnal cycles of water vapor, cloud liquid water, cloud ice, rainfall, and snowfall rate are investigated. Strong diurnal cycles during summer in several observables indicate a strong coupling with the surface and convective transport of air from the surrounding valleys to the level of UFS resulting in maximum amounts in integrated water vapor, cloud liquid water path (LWP) and rain during the afternoon. In contrast, no diurnal cycle is found during winter, which points to the predominance of advection of cloud systems associated with large scale dynamics during winter. Daily precipitation estimates for snowfall and rainfall derived from a vertically pointing, low-cost micro rain radar (MRR) are found to be in surprisingly good agreement with manual observations from the German Weather Service at the summit. The comparison strongly underlines the usefulness of active remote sensing methods for estimating precipitation in remote locations with harsh weather conditions. The analysis revealed strong vertical gradients of radar reflectivity in the lowest few hundred meters above ground which has strong impacts on retrieved precipitation quantities. The synergy of MRR and microwave radiometer measurements also revealed that almost 90 % of the snow clouds contained significant amounts of super-cooled LWP which highlights the importance of mixed-phase microphysical processes for orographic precipitation. The still growing data set at this very particular location, also in combination with further observations, such as trace gases and aerosols, has a unique potential for many applications, e.g. to investigate cloud processes, evaluate high resolution models, and to validate satellite products.

An extreme precipitation event, associated with the extra-tropical storm Alex, caused flash-floods and widespread damages on the western Alps on 2-3 October 2020. Precipitation records were broken in some places and for different accumulation periods, and were in the order of 600 mm/24h. Both the synoptic and the mesoscale features were typical of previous heavy precipitation phenomena and major floods on the southern side of the Alps: an upper-level trough over the western Mediterranean basin inducing a meridional transport of warm and moist air towards the orography over the Mediterranean and the Adriatic Sea, also in the form of a pre-frontal low-level jet. However, in the present case, the scene was made more complex by the contribution of an atmospheric river (AR), a narrow corridor of strong water vapour transport from the North Atlantic (even from the Gulf of Mexico) to the Mediterranean, which dominated the moisture supply in the middle troposphere and determined the distribution and the intensity of the rainfall. The contribution of the intense AR superimposed on the well-known dynamics-thermodynamics of heavy precipitation over the Alps and contributed critically to turning the heavy rainfall event into a devastating flood. High resolution numerical experiments have been exploited to investigate the dynamics and to improve our understanding of the complex interaction between large-scale flows and mesoscale mechanisms during this heavy Alpine rainfall event. Moreover, this study aims to eventually assess AR characteristics and their contribution to the rainfall, as well as to determine how often they are associated with synoptic upper-troposphere disturbance affecting the Mediterranean responsible for extreme precipitation episodes over Italy.

The Wasatch Range is a narrow, meridionally oriented mountain range in the North American interior that rises 2000 m locally to 3500 m MSL. The central Wasatch Range broadens southeast of Salt Lake City, Utah and contains a series of zonally oriented alpine ridges, creating a region of high topography exposed to flow from many directions. Snowfall and liquid precipitation equivalent (LPE) maximize in the high terrain near Little Cottonwood Canyon (LCC) where mean annual snowfall exceeds 1250 cm and a major highway ascends >1000 m in 11 km to Alta Ski Area, traversing 50 avalanche paths. Due to heavy traffic and a lack of snowsheds and other avalanche protection structures, the highway has the highest uncontrolled avalanche hazard index of any major road in the world. Using a 23-year record of 12-hour manual snow and LPE observations from Alta, ERA5 reanalyses, and operational radar data from a US National Weather Service radar, we analyze the synoptic and mesoscale characteristics of cool-season (October–April) precipitation extremes in upper LCC, where most of the cool-season precipitation falls as snow. Major precipitation events occur over a wide range of 700-hPa (near-crest-level) flow directions. Major snowfall events are most frequent, however, in northwesterly flow, whereas major LPE events are most frequent in southwesterly flow. Precipitation efficiency, defined as the fraction of vertically integrated water-vapor flux converted to precipitation during storm periods, is highest during northwesterly flow, with high snow-to-liquid ratios yielding low-density snow and large snowfall accumulations. Southwesterly flow contains larger vertically integrated water-vapor fluxes and LPE, but snowfall is limited by smaller snow-to-liquid ratios. Regionally, there are two pathways for water vapor penetration to the central Wasatch. One is across the relatively low northern Sierra Nevada of northern California and southern Cascades of southern Oregon, the other through the lower Colorado River Basin. Both avoid the high terrain of the Sierra Nevada in southern California. During southwesterly flow, precipitation systems have broad coverage with a high frequency of radar echoes in the central Wasatch and surrounding mountains and valleys. In contrast, during periods of northwesterly flow, precipitation enhancement in the central Wasatch and LCC can be highly localized due to orographic forcing or lake-effect processes. These results illustrate the large variety of storm types that produce heavy precipitation in the central Wasatch and LCC, including local forcing mechanisms and key vapor-transport pathways enabling heavy precipitation over an interior, continental mountain range.

Last year’s extraordinary warm and dry summer caused all across Europe droughts to an extent barely ever registered before, impacting drastically the everyday life of the citizens and their health. This event was no exception; people around the world face water scarcity, and even long-lasting droughts. For decades, the private and public sector tried to counteract these issues by employing a weather modification method to increase the precipitation in the affected areas. The applied method relies on so-called cryogenic seeding particles, that are injected into a supercooled liquid cloud at sub-zero temperatues to initiate the formation of ice crystals, thus the glaciation of a cloud, and subsequently precipitation. However, unknowns regarding the cloud characteristics and the dynamical variability render any clear statistical response next to impossible. In the CLOUDLAB project, we aim to obtain statistics from field experiments carried out at Eriswil on the impacts of cryogenic cloud seeding for ice crystal nucleation and growth and the evolution of wintertime supercooled low stratus clouds over the Swiss Plateau. The field experiments will be combined with numerical weather prediction (NWP). By targeting stable high fog situations, a controlled lab environment can be mimicked in the field. CLOUDLAB uses the ICON-NWP (Zängl et al., 2014) model in limited-area and large-eddy mode. We implemented a seeding parameterization in the two-moment microphysics scheme (Seifert and Beheng, 2006) and are simulating the seeding experiments conducted in the CLOUDLAB field campaign with varying horizontal resolutions ranging from 1 km down to 65 m. While it remains a challenge to realistically represent low stratus in ICON, we were able to inject seeding particles in an existing cloud comparable to observations. We conducted a series of sensitivity test to assess the significance of the changes in micro- and macrophysical properties of the targeted clouds and will present them at the ICAM conference.

Flow past mountains or mountain ranges is often associated with cloud formation, typically on the windward side. However, in case of an isolated steep mountain a cloud is sometimes observed to occur on the leeward rather than the windward side. Such a cloud is often referred to as a “banner cloud”. The presentation starts with a definition of “banner cloud” and presents observations taken some time ago at Mount Zugspitze. We proceed by looking at numerical simulations with idealized model configurations, performed to uncover the basic mechanism of lee-side cloud formation and to shed light on the flow regimes that are conducive to this type of clouds. In addition, we explore to what extent the strength of the ambient wind as well as its vertical shear have an impact on the formation of banner clouds. Finally, we proceed from simple model configurations with idealized mountains to more realistic mountains such as Mount Zugspitze or the Matterhorn.

The Laseyer windstorm is a local and strong wind phenomenon in the narrow Schwende valley in northeastern Switzerland. The phenomenon has raised the interest of meteorologists as it has - in the past - led to derailments of the local train. It is characterised by easterly to southeasterly winds during strong northwesterly ambient wind conditions, thus the wind in the valley reverses compared to the flow further aloft. Previously, this flow reversion during the Laseyer has been associated with rotor circulations - however without direct scientific evidence. Here, we study the Laseyer phenomenon with a combination of local observations, reanalysis data, and high-resolution numerical simulations. In the first part of our work, we show an updated climatology of the Laseyer and the synoptic conditions during Laseyer events. Our analysis is based on (i) hourly measurements from the weather station in the Schwende valley and other nearby stations, and (ii) measured wind data with 10 minute temporal resolution at the same stations. These data update the previously published climatology of the Laseyer and extend it by illustrating additional characteristics of the flow during Laseyer events, as for instance the cold air advection. Using the resulting Laseyer climatology of more than 20 years and ERA5 reanalysis data, we identify the synoptic conditions necessary for the occurrence of the Laseyer. We classify the slightly different synoptic conditions and show how they lead to the occurrence of Laseyer events. Moreover, we try to identify synoptic predictors for the strength and duration of Laseyer events. In the second part of our work, we use idealised large-eddy simulations (LES) to gain a more coherent depiction of the local flow in the complex terrain. The LES are performed with the Finite-Volume Module of the IFS using idealised initial and boundary conditions with real orography at a horizontal resolution of up to 50 m. Using the LES, we can study the sensitivity to ambient flow parameters in a controlled environment. As a result, we are able to shed light on the mechanism of the Laseyer by revealing the flow structure and variability in the narrow Schwende valley. A set of idealised LES with carefully selected ambient flow parameters allows us to identify the wind direction and wind speed conditions needed for the occurrence of the Laseyer and its response to changing ambient flow conditions.

Terrain smoothing is common practice in NWP models with terrain following grids, to avoid numerical errors and model instability associated with steeper slopes. Increasingly steep slopes become resolved at higher horizontal resolutions, and the compromise required for models to run efficiently, stably, and retain the terrain detail introduced by high resolution, can become difficult. In order to avoid this in simulations at 100 m horizontal grid spacing over steep mountains in the Tibetan Plateau, a method to spatially modulate smoothing, targeting areas where slopes are steepest, was developed. In the simulations in question, this allows a drastic reduction in the smoothing employed in the majority of the domain, compared to more conventional uniform smoothing. As a result, flow features associated with funnelling down narrow lee slope channels are introduced, among other qualitative changes. Temporal variability around and downwind of flow separation zones quantitatively increases. Spatial variability at small scales visibly increases in flow depictions, and point observations suggest leeside variability and flow effects are better reproduced when the targeted smoothing approach is used. The new method proves equally useful for coarse (e.g. global model) resolutions, and is in the process of integration into the central MetUM code base.

We revealed effects of mountain width on occurrence of a downslope wind. Firstly, we conducted idealized, three-dimensional, non-hydrostatic, free-slip numerical simulations. The simulations showed a mountain with narrow width suppressed occurrence of downslope wind. Specifically, for h^=0.7 (where h^ is non-dimensional mountain height; h^=Nh/U; h is mountain height, and; N and U are upstream basic flow speed and Brunt-Väisälä frequency, respectively), the narrow mountain (a^=2.5, where a^ is non-dimensional mountain width; a is mountain width) did not induce downslope winds, while the wide mountain (a^=10.0) induced downslope winds. Both mountains induced downslope winds for h^=0.8. We analytically solved Long (1955)’s equations considering non-hydrostatic effects of mountain waves theoretically to discuss the simulation results. The solution exhibited that a narrow mountain suppressed the downslope wind occurrence by extending vertical wavelengths. Long vertical wavelengths resulted in relatively small vertical displacement of flow and mountain height. The present solution is similar to Smith (1985)’s solution when we consider flow over a wide mountain.

Two of the most archetypal characteristics of Foehn winds are their distinct warming and their descending motion in the lee of the Alpine barrier. In the course of the last century, both aspects triggered fierce and long-standing scientific disputes. Here, we revisit the warming mechanisms and the descent of Foehn air with state-of-the-art NWP datasets. We employ a palette of 15 COSMO hindcasts at kilometer-scale resolution together with online trajectories calculated during model integration. On the one hand, the online trajectories are combined with a Lagrangian heat budget in order to quantify the warming contributions by adiabatic and diabatic processes. On the other hand, the explicitly resolved descending motion along online trajectories enables us to identify and characterize hotspot regions of strong descent. In previous studies, the warming was often solely explained by upstream latent heating in clouds, while recently it has been primarily attributed to adiabatic descent (isentropic drawdown). Employing the Lagrangian heat budget, we confirm adiabatic descent as the overall most important mechanism, when all events and regions are conjointly analyzed. However, we also denote a substantial regional variability: While the adiabatic descent often dominates the heat budget in the Eastern Alps, diabatic processes provide a substantial contribution to the warming in the Western Alps. Adding to the complexity, a pronounced case-to-case variability is likewise diagnosed. In a detailed case study of a particularly long-lasting South Foehn event, it is revealed that the different warming contributions are tightly linked to the formation of coherent airflows on the Alpine south side. Overall, our results corroborate a nuanced view on Foehn air warming, as the governing mechanism depends upon the case studied, the considered region and the time period in focus. With respect to the descent, the online trajectories reveal that distinct descent hotspots exist. These regions of strongly descending motion are spatially confined to the immediate lee of local mountain peaks and ranges. The Foehn air parcels descend rapidly (5-10 min) and cover a vertical distance that is related to the elevation difference of the underlying terrain. A particularly pronounced descent hotspot is situated downwind of the Raetikon, a local mountain chain adjacent to the Rhine Valley. Local peaks along this chain excite propagating gravity waves and therefore induce the descent within the waves. Besides, the local terrain north of the Raetikon favorably redirects descending air parcels into the Rhine Valley.

Thermally driven valley winds are efficient in transporting pollutants, moisture and trace gases out from the valley atmosphere under favorable conditions. The strength and spatial structure of these valley winds are sensitive to the valley shape, synoptic-scale forcing, and surface properties of the valley. The effect of the valley shape has been studied by numerous measurement campaigns and modelling studies, using both real and idealised topographies. A recent case study compared the winds in four Himalayan valleys and found the valleys with steeper inclination had weaker and shallower daytime up-valley winds. In this work we further investigate the effect of the valley floor inclination on the daytime valley wind strength and transport of passive tracers out from the valley by means of idealised WRF-simulations (Weather Research and Forecasting model). The analysis consists of four experiments with different idealised topographies. The valley topographies are defined with a sinusoidal shape in the cross-valley direction and the along-valley inclination is constant throughout the valley. Three valleys, each 100-km-long and 2-km-deep, are simulated. One of the valleys has a flat valley floor and in two of the valleys the floors reach 2 km (1.15 degrees) and 4 km (2.30 degrees) height along the 100-km valley length. In addition, a flat slope inclined at 1.15 degrees is simulated. Compared to previous studies, here the valley ridges are inclined to the same degree as the valley floor and hence the valley floor inclination does not lead to differences in the valley volume. The inclinations are representative of the Himalayan valleys and much steeper than in previous idealised valley simulations. The simulation is forced by a sinusoidal daily cycle in the surface sensible heat flux and synoptic-scale winds are excluded. Passive tracers are released near the surface at different locations in along-valley and cross-valley direction and at different times of the day to simulate the effect of different emission sources. The daytime up-slope winds are mostly responsible for the ventilation of the tracers whereas the along-valley winds transport the tracers towards the valley top within the valley atmosphere. The inclined valley topographies have a layered daytime along-valley circulation, with 2-3 stacked cells, that lift some of the tracer up from the valley center. The tracers released at the slopes are transported more efficiently out of the valley atmosphere compared to those released in the valley center-axis. These initial results, and others, will be presented.

During the final and most intense phase of the Adrian storm (also known as Vaia), on the evening of 29 October 2018, quasi-stationary convective rainbands were observed over the eastern Italian Alps. Rainbands oriented from southeast to northwest, driven by the strong Sirocco wind, caused floods and landslides in several locations. Their development is investigated in the present study through semi-idealized numerical simulations, with the Weather Research and Forecasting (WRF) model, in order to identify the role of the thermodynamic conditions and of small-scale topographic features. Simulations are initialized using the vertical profile measured at Udine-Rivolto radio-sounding station at 18:00 UTC, 29 October 2018, representative of the conditions upstream of the eastern Italian Alps. Either background thermal fluctuations embedded in the low-level flow or random perturbations added to the topographic field provide the small-scale energy needed to develop convection. Simulations show that small-scale topographic features are not necessary for the generation of rainbands since the latter develop even using a smooth ridge. Non-stationary rainbands result from flow-parallel roll-type circulations with tilted updrafts extending up to 6-7 km altitude. Sensitivity experiments with 1, 0.5, and 0.2 km grid spacing highlight that such features are independent of the model resolution. Additional numerical experiments with different sounding profiles are performed to investigate the influence of stability, wind speed and wind shear on the development of rainbands. Akin to previous studies, banded convection is favored in intense vertically sheared flows without directional shear and weakly unstable cap clouds. Also, moderate values of convective inhibition are fundamental for constraining the release of convection over the ridge. Conversely, the presence of either saturated layers or strong potential instability within the mid-upper part of the statically unstable cap cloud may disrupt the convective organization.

Gravity waves generated by mountains are multi-scale and three-dimensional. Current orographic gravity wave drag parametrization schemes assume that the waves are two-dimensional, varying only in the vertical and along one direction. These schemes, therefore, do not represent the process of partial critical level filtering, whereby a portion of the wave spectrum is saturated when the winds parallel to its wavevectors become small. This results in an unrealistic vertical distribution of the momentum flux and forcing of the waves on the mean flow. In this work, a method of accounting for partial critical level filtering in the orographic gravity wave drag parametrization using the full spectrum of realistic topography is presented. This is achieved through binning of the expression for the linear hydrostatic surface stress, computed using Fourier transforms of the subgrid orographic heights within model gridboxes, into wavevector directions. The parametrization is compared with idealised simulations of flow over complex topography and is shown to perform well as the number of wavevector direction bins is increased. Implementation of the scheme into an operational forecasting model is tested using short-range 5-day forecasts. As is found from idealised simulations, the binned scheme leads to less forcing in the troposphere and increased forcing in the stratosphere within the model. The binned scheme is shown to alleviate biases in the upper stratosphere, between 45 km and 65 km, as well as having significant local effects in the troposphere.

It has been well known that the Tibetan Plateau (TP) exerts great impacts on the spring persistent rainfall (SPR) in East Asia. However, the underlying dynamics are incompletely understood, with previous work focusing only on the mechanical and thermal effects of the TP’s large-scale orography. Based upon numerical experiments using the Weather Research and Forecasting model, this work reveals that orographic gravity waves (OGWs) triggered by small-scale orography over the TP also play a vital role in the formation of SPR in East Asia. The breaking of OGWs produces pronounced orographic gravity wave drag (OGWD) in the middle troposphere over the TP, which drives a meridional circulation across the plateau. The rising branch of the meridional circulation to the south of the TP dynamically pumps the low-level air upward and thus lowers the pressure, with the meridional pressure gradient increased. In consequence, the low-level southwesterlies on the southeastern flank of the TP are strengthened, which tends to enhance the water vapor transport and hence SPR in East Asia. Our study improves the understanding of the TP’s multiscale complex terrain in affecting the rainfall in East Asia and provides new insights into the westerly-monsoon synergy in the Asian monsoon region.

A linear analysis of the trapped mountain waves that can develop in the presence of a turbulent boundary layer is presented. The theory uses a mixing length turbulence model based on Monin-Obukhov similarity theory. First, the backward reflection of a stationary gravity wave (GW) propagating toward the ground is examined. Three parameters are investigated systematically: the Monin-Obukhov length (Lmo), the roughness length (zo) and the limit value of the mixing length (λ) aloft the "inner" layer (roughly the layer over which dissipations affect the dynamics). The reflection coefficient appears to strongly depends on the Richardson number aloft the inner layer (J=λ/Lmo): the reflection decreasing when the stability (J) increases. As a large ground reflection favors the downstream development of trapped lee waves, the fact that it decreases when J increases explains why the more unstable boundary layers favor the onset of trapped lee waves. The influence of the roughness and mixing lengths on the reflection coefficient is less significant but still present. It tends to decrease when zo increases but increases with λ. Their influence is indirect, but can be explained using a combination that measures the depth of a "pseudo"-critical level located below the surface: the reflection increases when the "pseudo" critical level goes away below the surface. The formalism is then extended to identify the preferential modes of oscillation that exist when there are turning levels in the inviscid part of the flow. The results are then tested including a mountain ridge which stays entirely located within the "inner" layer.

While it is known that trapped lee waves propagating at low levels in a stratified atmosphere exert a drag on the mountains that generate them, the distribution of the corresponding reaction force exerted on the atmospheric mean circulation, defined by the wave momentum flux profiles, has not been established, because for inviscid trapped lee waves these profiles oscillate indefinitely downstream. A framework is developed here for the unambiguous calculation of momentum flux profiles produced by trapped lee waves, which circumvents the difficulties plaguing the inviscid trapped lee wave theory. Using linear theory, and taking Scorer’s two-layer atmosphere as an example, the waves are assumed to be subject to a small dissipation, expressed as a Rayleigh damping. The resulting wave pattern decays downstream, so the momentum flux profile integrated over the area occupied by the waves converges to a well-defined form. Remarkably, for weak dissipation, this form is independent of the value of Rayleigh damping coefficient, and the inviscid drag, determined in previous studies, is recovered as the momentum flux at the surface. The divergence of this momentum flux profile accounts for the areally integrated drag exerted by the waves on the atmosphere. The application of this framework to this and other types of trapped lee waves potentially enables the development of physically based parametrizations of the effects of trapped lee waves on the atmosphere.

In 2021 the Colorado River basin snowpack was 80% of average but only delivered 30% of average flows. This is concerning for the 40 million people in the western United States who depend on the river. Many are now asking, where did the snow water go? Is snow water likely to disappear like this again in the future? For winter 2022-2023, we deployed snow pillows, scanning lidars, blowing snow sensors, and the National Center for Atmospheric Research (NCAR) Earth Observing Laboratory (EOL)’s Integrated Surface Flux System (ISFS), in partnership with the Department of Energy (DOE)’s Surface Atmospheric Integrated Field Laboratory (SAIL) and NOAA’s Study of Precipitation, the Lower Atmosphere and Surface for Hydrometeorology (SPLASH), at the Rocky Mountain Biological Lab (RMBL) in the East River Watershed, Colorado to bring together observations essential to span scales from seconds to seasons, from turbulence to valley-circulation to mesoscale meteorology, to constrain this difficult and societally-relevant problem. We present an overview of the field campaign, as well as preliminary results on the dominant controls of mountain snow sublimation and where the mountain water really went.

In 2018, a fatal crash of a Junkers Ju-52 occurred in a small, steep Alpine side valley to the south-west of the Piz Segnas in the Canton of Grisons in Switzerland. During the subsequent safety investigation, simulations of complex local flow patterns were conducted. This was done by means of high-resolution large eddy simulations with the PALM model system, a turbulence resolving model for atmospheric boundary layer flows. PALM allows the accurate treatment of topography on a cartesian grid. The simulations were conducted for multiple days of summer 2019, focusing on days with comparable atmospheric conditions to the day of the accident. All simulations were performed by applying a large scale forcing from COSMO-1 reanalysis fields. A nested simulation approach was chosen to achieve high spatial resolution of ten meters within the area of interest. The simulation results for the vertical wind velocity were compared with measurements of a continuous-wave LIDAR system and horizontal wind speeds were compared against data from a measurement mast located on the upstream ridge of the valley. The LIDAR was positioned in a way to achieve the best accuracy for the vertical wind component along the critical phase of the flight, as the main interest was the magnitude and distribution of up- and downdrafts. After the validation of the simulations for 2019, the same modeling technique was applied for the conditions of the day of the accident, where observations by rescue crews of prevalent wind flows were available for a qualitative verification. The results of the simulation are showing good agreement with both the measured horizontal wind speed and direction at the ridge and with the vertical wind speed component as measured by the LIDAR. Yet, the simulated values for the horizontal wind speed and direction components differ from the measured values for increasing elevations above the LIDAR. This could be explained by the presence of a pronounced rotor flow that develops in the lee of the crest. Furthermore, given the measurement principle of the LIDAR, the data at this point could suffer from a 180° ambiguity for the wind direction. In general, the simulations provide valuable insights into the highly complex flow structures in the side valley under study. Especially the complex flow conditions observed at the time of the accident were reconstructed. The detailed report is available online (see STSB final report No. 2370, Annex A1.7).

Wind energy production at high latitudes is becoming increasingly common. Even in Svalbard (78° N), wind turbines are being investigated as components of a modern power system. However, the high Arctic poses special features relevant for wind power, such as low temperatures, midnight sun and polar night, or complex terrain with fjords, valleys, mountains, and glaciers. This environment gives rise to local, thermally-driven breezes such as valley and glacier winds or land/sea breezes. Research regarding these high-latitude winds has mostly concentrated on the effect of large glaciers on the local wind behaviour so far. However, the role of Arctic ice-free valleys should also be considered, especially since local up-valley breezes can develop there but not on glaciers. A field campaign conducted in the valley Adventdalen (Svalbard) in the summer of 2022 revealed that a combined sea breeze and up-valley wind develop frequently in the summertime. The breeze is driven by the strong heating of the valley compared to the fjord. The breeze can persist over several days due to the midnight sun. Its strength, extent, and depth vary diurnally due to the interaction of large-scale winds and local thermal forcing. Lidar scans of the wind speed show the influence of the breeze on the average wind speed profile. The fjord breeze was relatively strong and frequent despite low absolute levels of radiative forcing, probably due to the strengthening effect of the valley and plateau topography. Understanding the thermally-driven winds in summer will contribute to judging the effect of Arctic valleys on wind power potential, wind speed profiles, and spatial wind speed differences between exposed mountains and valleys.

Results from the simulation of a passive tracer dispersion by a thermally driven circulation along a simple slope are presented. The dispersion process associated with slope winds, including both the upslope (or anabatic) diurnal component and the downslope (katabatic) nocturnal one, is investigated exploring the sensitivity to the different combinations of characteristics of the emission source and of the wind profile. The dispersion properties of the vertical structure and the characteristic jet-shaped wind profile are obtained using the steady-state Prandtl (1942), both in the original version and in the modified solution proposedby Charrondiere et al. (2022). Different heights of steady pontiwise tracer source are tested. An Eulerian model is developed and validated with the analytical Gaussian solutions for known configurations of the wind profile and later used to prove the limits of the Gaussian approach for the slope wind layer. The role of the relative position of the source and the wind jet is analyzed and two different regimes are identified and described. Charrondière, Claudine, et al. "Mean flow structure of katabatic winds and turbulent mixing properties." Journal of Fluid Mechanics 941 (2022). Prandtl L (1942) Führer durch die strömungslehre. F Vieweg & Sohn, Braunschweig

Aerosol concentrations at high-altitude sites can be highly influenced by both the large-scale circulation and the local vertical transport induced by complex terrain. In this study we use the WRF-CHIMEREv2020r3 Chemical Transport Model (CTM) in a high-resolution configuration, i.e. down to 1 km, to investigate the transportation patterns of sulphate aerosols at the Mt Cimone Atmospheric Observatory, located in the Northern Apennines, Italy (i.e. CMN GAW/WMO Global Station, 44°11' N, 10°42' E, 2165 m a.s.l.). The performance of the model is evaluated against meteorological measurements from the operational weather stations in the surrounding area, as well as against online measurements of aerosol chemical composition during July 2017. Both the synoptic influence and local thermally driven flows developing in the adjacent costal and mountainous areas are very well captured by the model, with few discrepancies on ridges in the proximity of CMN. The model reproduces the monthly trend and absolute values of the total sulphate aerosols mass which exhibits two overlapping patterns: a first one linked to regional long-range transport of sulphate-enriched air masses from the western Mediterranean, occurring on a weekly timescale, and a second one due to the combined effect of the diurnal sea-breeze over the Tyrrhenian coast and upvalley circulations in the Northern Apennine. Divergence between model and observations is attributed to i) smoothing effects in the model topography, ii) uncertainties in the parameterization of the aqueous-phase sulphur chemistry, a very efficient sulphate aerosol formation mechanism. The analysis of the modelled wind patterns clearly identifies the south-western quadrant as the main source region of air masses reaching CMN in the study period. Moreover, a model-based source apportionment analysis indicates that periods with higher sulphate concentrations are linked to maritime sources (ship emissions, dimethyl sulfyde and sea salt emissions) located to the south-west of CMN, which contribute up to 40% of the total sulphate concentrations. On the other hand, when air masses come continental Europe, sulphate concentration modelled at CMN is lower and is linked to industrial combustion processes.

Atmospheric waves are known to affect transport of momentum, energy, and mass and thereby atmospheric composition. To date, many aspects of the wave lifecycle and wave effects are to a large extent unknown and not all wave types are fully resolved or properly parameterized in state-of-the-science atmospheric models. Model deficiencies emerge prominently in mountainous regions, which are key regions of wave sourcing and also hotspots of climate change and boundary layer pollution. In this presentation we introduce the research project UnrAvelLing MOuntain Wave EFfects in the Troposphere (ALOFT) that is planned in the frame of the TEAMx programme and show preliminary results. The over-arching goal of ALOFT is to study the impact of mountain waves (MWs) on tropospheric composition in the Alpine region. To this end we analyze dissipative and, for the first time, non-dissipative GW effects on tropospheric chemistry and transport processes revealed by high-resolution simulations with the interactive chemistry. Based on these results, we aim to modify GW parameterizations and coupling of individual modules of a well- established chemistry-climate model (CCM) to account for dissipative and non-dissipative GW effects over a complex terrain. Within ALOFT we perform a set of hindcast sensitivity simulations with the Weather Research and Forecasting Model with full online chemistry (WRF-Chem) and in different configurations for the Alpine domain with variable smoothing of the underlying orography. The unique architecture of sensitivity experiments allows for a clean attribution of MW effects on chemistry and transport in the troposphere, which we document with our first results. Ultimately, ALOFT will pave the way to modify GW parameterization schemes to achieve a more realistic spatio-temporal distribution of radiatively active trace gases and atmospheric pollutants, and their impacts, by including GW effects on chemistry and transport - to date not considered in CCMs.

In the summer of 2022, a pre-campaign of the "Multi-scale Transport and Exchange Processes in the Atmosphere over Mountains - Programme and Experiment" (TEAMx-PC22) took place in the Austrian and German part of the Inn Valley. The aim was to test new instruments, new instrument configurations and new measurement sites to support the planning of the main TEAMx observational campaign (TOC) in 2024/2025. Six German and Austrian research institutions and weather services took part in the pre-campaign and collected measurements in four sub-target areas in the Inn Valley and in one of its tributaries, the Weer Valley. Instrumentation included a new Raman lidar for the observation of tropospheric water vapor, temperature and aerosols, multiple scanning Doppler wind lidars, uncrewed aerial systems (UAS) and ceilometers, a cloud radar, a distributed temperature sensing (DTS) system and several surface stations. On a few selected days, radiosondes were launched at two locations. The presentation gives an overview of the campaign and presents highlights of the observation of thermally driven flows, boundary layer evolution and scale interaction.

TEAMx stands for ‘Multi-scale Transport and Exchange Processes in the Atmosphere over Mountains - Programme and Experiment’ and is an international endeavour to better understand atmospheric processes and their interaction in the ‘Mountain Boundary Layer’ (MoBL). The MoBL is much more than the Atmospheric Boundary Layer over flat and homogeneous terrain, as it comprises – depending on the spatial scales of the orography – atmospheric processes between micro- up to meso-𝛼-scales and especially the interaction between them. The TEAMx community combines a year-long observational campaign with coordinated numerical modelling activities to achieve the goal of better understanding the relevant exchange processes and their interactions, improving the numerical models where necessary and providing better input to weather and climate services in the mountains. This presentation gives a brief overview on the TEAMx state of affairs for those, who could not participate in the TEAMx workshop taking place in the week before the 36th ICAM.

Ice formation and growth processes play a crucial role for precipitation initiation. However, fundamental knowledge gaps in ice phase processes exist which fundamentally impact precipitation forecasts. Especially in mountainous terrain, the occurrence of orographic lifting may force the generation of clouds and subsequent precipitation. CLOUDLAB strives to better understand the processes of ice crystal formation and growth by using the wintertime low stratus clouds that regularly occur in Switzerland as a natural laboratory for targeted glaciogenic cloud seeding. Taking advantage of the rise in terrain towards the Alps, the CLOUDLAB field site is located at the lower center of the basin formed by Jura Mountains and Swiss alps at an elevation that allows probing the low stratus with in-situ and ground-based remote sensing devices. In our seeding experiments, an uncrewed aerial system (UAS) releases ice nucleating particles within supercooled stratus clouds. Related microphysical changes are measured downstream of the release location with an array of ground-based remote sensing devices (ceilometer, microwave radiometer and two or more radars), in-situ devices on a tethered balloon (holographic imager, optical particle counter, radiation), and a second UAS (optical particle counter). The flexibility of the UAS enables us to select the seeding height and location, which directly translates into the temperature at which the crystals grow and their growth time, or rather ice crystal size. This precise control of seeding locations via UAS together with the persistent stratus clouds means we can repeat our experiment at the same or similar, well-constrained initial conditions, ensuring statistical robustness of our findings. During the past two winters (2021/22, 2022/23), we successfully conducted several seeding experiments, sampling before, during, and after seeding at temperatures below -5° C. Seeding signals were detected either directly via increased radar reflectivity or alteration of linear depolarisation ratio (i.e., shape of particles) relative to the background of the undisturbed cloud. During a first seeding experiment, an updraft led to the creation of ice crystals whereas a downdraft during a second experiment within one hour later led only to an increase in cloud droplets based on the linear depolarisation ratio, suggesting that the seeding particles can activate as cloud condensation nuclei before nucleating ice crystals. Here, we will report on our field setup and give an overview of the conducted seeding experiments. Ultimately, we will use these field observations to enhance the representation of the ice formation and the potential subsequent precipitation in weather models.

During near-freezing surface conditions, diverse surface precipitation types (p-types) are possible: rain, drizzle, freezing rain, freezing drizzle, wet snow, ice pellets, and snow. In regions of complex terrain, orographic airflow dynamics can play an important role in determining p-type transitions by shaping the thermodynamic environment and the mesoscale pattern of vertical motion. This presentation provides an overview of the orographic influence on p-type observed during the Winter Precipitation Type Research Multi-scale Experiment (WINTRE-MIX), which was conducted from February - March 2022 to better understand how multi-scale processes influence the variability and predictability of p-type and amount under near-freezing surface conditions. WINTRE-MIX took place near the US / Canadian border, in northern New York and southern Quebec, focused on the Lake Champlain and Saint Lawrence River Valleys where complex terrain plays important roles in near-freezing precipitation events. During WINTRE-MIX, observations from existing advanced mesonets in New York and Quebec were complemented by deployment of additional surface instruments, the National Research Council Convair-580 research aircraft, three mobile University of Illinois Doppler on Wheels dual-polarization radars, and teams collecting manual hydrometeor observations and radiosonde measurements. Observations will be presented from WINTRE-MIX intensive observing periods that depict a range of terrain influences on p-type, including: (1) channeled cold northerly flow down the Champlain valley that supporting an extended period of ice pellet accumulation, and (2) trapped/channeled northeasterly flow in the Saint Lawrence Valley supporting the occurrence of freezing rain. Airborne, mobile radar, sounding, and surface observations will be synthesized to reveal the interplay between synoptic forcing, orographic airflow dynamics, boundary layer flows, and cloud/precipitation microphysics. The performance of operational high-resolution numerical weather prediction in capturing the observed orographic influences on p-type will be discussed.

Convection significantly contributes to the vertical overturning of heat, moisture, and momentum in the atmospheric boundary layer and is responsible for the formation of convective clouds and precipitation. Most numerical weather prediction models rely on parameterisation schemes that are often difficult to constrain, as observations of atmospheric convection are notoriously difficult to obtain due to the intermittent, localized, and turbulent character of convection. Convection thus remains a key source of model uncertainty in weather and climate models. However, the characteristic properties, dynamics, and processes that trigger and shape the development of atmospheric convection are still only sparsely sampled, hampering model validation and development. To address this shortcoming, we present an approach to probe and characterise atmospheric convection from both a Eulerian as well as a quasi-Lagrangian perspective. We utilise multiple doppler Lidars in combination to obtain observations over the mountainous terrain of southwestern Norway. While this setup can accurately characterise atmospheric convection in complex terrain by resolving its dynamic evolution, it is limited to the vertical cross-section spanned by the Lidars. However, on most convective days, hundreds of paragliding, hang gliding, and gliding pilots fly in convective plumes to seek lift, providing additional data way beyond the specific Lidar cross sections. We show how tracks of these engineless aircrafts can be used to sample atmospheric convection and elaborate on how this data can be used to characterise atmospheric convection and thereby to evaluate parameterisations as well as improve them through machine learning. In addition, we also equipped several paragliders and gliders with temperature and humidity sensors providing additional in situ observations at extremely low cost.

Starting in September 2021, the National Oceanic and Atmospheric Administration (NOAA) deployed instruments to the East River watershed near Crested Butte, Colorado. These instruments were deployed to capture observations of the lower atmosphere, surface energy budget, precipitation, and surface to advance weather and water prediction in complex terrain. Led by the NOAA Physical Sciences Laboratory, the SPLASH (Study of Precipitation, the Lower Atmosphere and Surface for Hydrometeorolgy) project leverages collaborations with several NOAA laboratories, the US Department of Energy, universities, NCAR and private industry. Together with the DOE-supported Surface-Atmosphere Integrated Laboratory (SAIL) and Watershed Function Science Focus Area, and NSF-supported work on sublimation of snow, SPLASH assets provide one of the most detailed perspectives ever captured of the atmosphere and its interactions with the surface in a single mountain watershed. To monitor clouds and precipitation, SPLASH includes two S-band snow level radars (SLRs), disdrometers, surface precipitation gauges, snow depth sensors, two ceilometers, a microwave radiometer, and a scanning X-band precipitation radar. Together, these systems provide information on precipitation intensity, cloud base height, water path, and precipitation size and accumulation. In addition, Atmospheric Sounder Spectrometer for Infrared Spectral Technology (ASSIST) and microwave radiometer data can be combined to support retrievals of temperature and moisture in the lower atmosphere. Remotely-sensed temperature, moisture and wind profiles were obtained at a second location using a CLAMPS (Collaborative Lower Atmospheric Mobile Profiling System) system. In addition to the profilers and hydrometeor sensors, NOAA deployed multiple systems capable of providing information on the evolution of the surface energy budget, including two remote Atmospheric Surface Flux Stations, a 10-meter flux tower, and broadband and spectral irradiance sensors. Additionally, there are systems deployed to measure soil moisture and temperature profiles to 50 cm. Finally, aircraft surveys of soil moisture, snow depth, and surface cover were conducted with piloted and uncrewed aircraft, and a network of snow stake and camera systems have been deployed at higher elevations to monitor the evolution of the snowpack throughout the SPLASH campaign. SPLASH is scheduled to continue into September of 2023. In this presentation, we will provide an overview of the SPLASH campaign, the observation systems deployed, some initial scientific highlights, and perspectives on future studies. This includes work to evaluate and improve numerical prediction tools supporting weather, hydrometeorology and climate studies. We will also discuss ongoing education and outreach work being conducted in conjunction with SPLASH.

The near-surface zero degree line (ZDL) is a key isotherm in mountain regions worldwide, but a detailed analysis of methods for the ZDL determination, their properties and applicability in a changing climate is missing. We here test different approaches to determine the near-surface ZDL on a monthly scale in the Swiss Alps. A non-linear profile yields more robust and more realistic ZDLs than a linear profile throughout the year especially in the winterhalf year when frequent inversions disqualify a linear assumption. In the period 1871–2019, the Swiss ZDL has risen significantly in every calendar month: In northern Switzerland, the monthly ZDL increases generally amount to 300–400 m with smaller values in April and September (200–250 m) and a larger value in October (almost 500 m). The largest increases of 600–700 m but also very large uncertainties (±400 m, 95% confidence interval) are found in December and January. The increases have accelerated in the last decades, especially in spring and summer. The ZDL is currently increasing by about 160 m/°C warming in the summer-half year and by up to 340 ± 45 m/°C warming in winter months. In southern Switzerland, ZDL trends and temperature scalings are somewhat smaller, especially in winter. Sensitivity analyses using a simple shift of the non-linear temperature profile suggest that the winter ZDL-temperature scalings are at a record high today or will reach it in the near future, and are expected to decrease with a strong future warming. Nevertheless, the cumulative ZDL increase for strong warming is considerably larger in winter than in summer. Based on a few key criteria, we also present best practises to determine the ZDL in mountain regions worldwide. The outlined methods lay a foundation for the analysis of further isotherms and to study the future ZDL evolution based on climate scenario data.

Valley winds are, inter alia, influenced by net radiation, Bowen ratio, and static stability in the valley. These parameters are likely to change under the current global warming conditions. We study the reaction of the valley wind system by investigating the trends in the past 30 yrs of parameters such as diurnal pressure and temperature amplitudes under fair-weather conditions at stations between Landeck and Kufstein in the Inn Valley (Tyrol, Austria) as well as observed winds and along-valley potential temperature and pressure gradients. The outcome of this study should be highly relevant for the validation of regional climate models in mountain regions.

The aim of this study is to evaluate whether extreme temperature events (heatwaves and cold spells) differ for plain versus mountain regions and for the four alpine sub-areas of the Greater Alpine Region (GAR) for past versus future; the average number of hot and cold days per year and the circulation patterns associated with these extremes are then investigated. ERA5 has been used to understand the observed past - 1951-2020 with 1951-1980 as the reference period - extreme temperature statistics and how future extremes will change by the middle of the century is estimated using high-resolution data from a CMIP6 model (EC-Earth 3P-HR). We find that during the 1991-2020 heatwaves occurrence, intensity and duration have increased (with respect to the 1951-1980), especially in the plains and the Southern region (mostly during summer). What we find as novel is that the Southern area is more affected by hot spells with respect to the Northern one. This is more evident in the future (2021-2050) when all the considered sub-regions show an increased number of hot spells as well as longer duration and higher intensity. Cold spells statistics show a decrease in the occurrence as well as their intensity and duration; even greater decreases are predicted for the future; there are no significant differences among seasons, plains, mountains and in the GAR. The Southern region again exhibits, mainly in the future, a higher (lower) number of hot (cold) days per year, with respect to the Northern area. Concerning the circulation patterns associated with temperature extremes, it is found that hot (cold) spells are associated with jet stream pattern modifications, bringing warm (cold) air from the African continent (North Pole) towards Italy. However, there are no differences in the composite of the circulation for heatwaves and cold spells occurring in plain, mountain and the four alpine sub-regions.

The field of extreme event attribution (EEA) seeks to quantify how recent extreme events are directly exacerbated by ongoing climate change. As this is a relatively new field in climate science, there is a noticeable knowledge gap in EEA analysis in mountain areas. Precisely, this work performs an attribution to climate change of the two greatest heatwaves (HWs) occurred during June and July 2022, both hitting the Iberian Peninsula and southern France, and therefore, the Pyrenees Mountain range. We used the analogues technique on 500 hPa geopotential height composites to identify the 30 days closer to the dynamical structure of both heatwaves for the counterfactual (1950-1985) and factual (1986-2021) period, using ERA5 daily data. Results showed that factual HWs analogues in the factual period have a spatial structure closer to the 2022 HWs events than those analogues extracted from the counterfactual period. At the Pyrenean scale, we observed that 2-meter air temperature differences consisted of a positive non-uniform pattern in a factual world, with a significant increase in the southern slope of the mountain range and in the nearby depressed areas. However, most of the mountain range exhibited a small increase of the HW air temperature in a factual world. We also provided an explanation of the physical process involving the abovementioned 2-meter air temperature differences. In this study, we revealed the complexity of conducting the attribution of extreme heatwaves to climate change in mountain areas, both because of the scarcity of in-situ data, as well as due to the physical processes involved during these extreme events in an area of complex terrain.

The C3S Arctic Regional Reanalysis (CARRA) dataset contains dynamically downscaled reanalysis and short term forecasts over 30 years at 2.5km resolution in an East and West domain. This is a new level of detail which makes it possible to resolve relatively small features. We have analyzed the wind data from the West domain which covers Greenland, Iceland and the surrounding sea. All main features of the flow over Greenland will be presented, including the main Greenland jets and several smaller features. These will be discussed with references to the mechanisms of orographic flow acceleration and deceleration.

Early numerical modelling efforts in atmospheric sciences and numerical weather prediction started in the 1950s. These early efforts were truly pioneering: At that time, decent programming languages were not yet available, storage of data and programs was largely in the form of punched cards and magnetic tapes, then available computers were large in size and marginal in performance, digital exchange of data and codes was not yet feasible, and the internet was unheard of. Nevertheless, the role of topography in atmospheric models quickly emerged as an important topic in atmospheric modeling. Already during the infancy of the field, different approaches emerged to represent topography, despite staggering limitations in computational resolution. Many of these early attempts left imprints in current weather and climate models. Tracing the history of topography in atmospheric models is thus a fascinating topic. In this presentation, an overview will be provided about this development. The lecture will detail some of the early ideas, address some of the more recent approaches, and provide some insights in the progress that has become possible in the last decades. The presentation will also discuss the role of Alpine field experiments – such as the Alpine Experiment (ALPEX) and the Mesoscale Alpine Programme (MAP) – in fostering some of these developments.

While elevation-dependent warming (EDW) has become a widely studied phenomenon referring to the stratification of warming rates with the elevation, elevation-dependent climate change (EDCC) is now gaining momentum. EDCC encompasses a broad range of responses to changing climate conditions along elevational gradients. These responses certainly include elevation-dependent changes in precipitation, a critical variable of the hydrological cycle, and in temperature and precipitation extremes. This talk will provide an overview of the state-of-the-art knowledge on EDCC, both at the global level and for specific mountains, including the Alps, Himalayas, Andes, and Rockies, both using observational data and model simulations, along with a discussion of the possible mechanisms involved in the uneven distribution of climate change across different elevations.

Climate change is a global phenomenon with regionally varying peculiarities and sensitivities. Mountainous regions have been found to be particularly sensitive and prone to severe impacts. At local scale the complex orography determines a range of climate change impacts that often depend on elevation which makes the future projection of the evolving mountain environment even more complex. Although regional climate models can solve the atmospheric equations on very fine scales, simulated climatic patterns can show large differences when compared with observations due to morphology and topography of the territory which, if complex, do not allow the models to explicitly resolve all the processes at that local scale. Studying the impact of climate change as a function of elevation is therefore unreliable if models fail to capture local physical mechanisms or large scale elevational patterns. In this study we aim to evaluate the representativeness of historical climate simulations in function of the elevation over the European Alps by using the EURO-CORDEX ensemble of regional climate models at 0.11° resolution. In addition to evaluating the standard EURO-CORDEX model output we assess the impact of bias-adjustment as represented by the CORDEX-Adjust ensemble in historical periods and for several climatic indices. The model data are compared to high-resolution observational datasets over the entire Alpine region and to national observational datasets. We identify potential advantages and weaknesses of bias-adjustment methods depending on elevation, region, and climatic index. Based on the results, we can put estimates of future climate change elevation patterns into context, evaluating which are the most appropriate indices for studies including climate change adaptation in mountain regions, and defining physical mechanisms at play at different elevations.

High-altitude regions have been identified as hot-spots of climate change. In particular, the dependence of warming rates on elevation has been largely discussed in the literature, focussing on the identification of positive Elevation-Dependent Warming (EDW), i.e. an enhancement of warming rates with elevation. Other key climate variables, e.g. precipitation or climate extremes, have been less investigated in terms of vertical stratification of their trends, although they are largely relevant in our understanding of the elevational dependence of climate change and related impacts. The aim of our study is to perform an analysis of elevation-dependent changes in climate extremes in mountain regions of the world. We have considered nine mountain regions: Tibetan Plateau, Loess Plateau, Yunnan-Guizhou Plateau, US Rockies, Alps, Appalachian Mountain, Andes and Mongolian Plateau. Using the ERA5 global reanalysis, we have analyzed temporal trends of mean temperature, mean precipitation and a selection of their extreme indices following the ETCCDI definitions (R10mm, R20mm, Rx1day, CDD, CWD, R95p, TX90p, TN10p) over the period 1951-2020. In our presentation, we show results for precipitation and precipitation extremes. Evidence of an altitude stratification of mean precipitation and precipitation extremes has been found in the majority of the mountain areas taken into consideration, with a general consistency among precipitation mean and its extremes. In the Tibetan Plateau, Loess Plateau, Alps and Andes (tropical and subtropical) the stratification shows an enhancement of precipitation and extreme precipitation rates at higher elevations, contrary to the Rockies Mountains where an inverse signal was found. Such patterns in precipitation are not well related to those in temperatures as one may expect. Preliminary results from an extension of the analysis to climate models from the new model ensemble CMIP6 are also discussed. At a first stage, the historical experiments were evaluated to see if models can represent the altitudinal stratification of precipitation change as the observations do. Different scenario projections are then analyzed to seek for a possible amplification of the signal of stratification in the future under different scenarios.

There is increasing evidence that temperature trends over recent decades are often dependent on elevation, with higher elevations often showing faster warming rates (elevation-dependent warming). This can be because of the influence of many drivers including the melting of snow and ice, and changes in atmospheric moisture. In the tropics most models suggest enhanced free-air warming in the upper troposphere (above 500 mb), associated with a moistening of the tropical atmosphere and an enhanced hydrological cycle. We analyse climate data obtained from a transect of 21 stations across Kilimanjaro in Tanzania, ranging from 990 m (Moshi) to 5803 m (Northern Ice Field). The transect covers both slopes from the wetter south-west to drier north-east, and to our knowledge has one of the largest elevation ranges in the world. Air temperature (2m) and relative humidity are measured using Hobo U23-001 dataloggers every hour with records extending from 2004 to 2022. Seasonal and diurnal cycles in temperature and vapour pressure are strong, particularly the latter, which is unsurprising for a tropical environment. Although both mean daily temperature and vapour pressure reduce with elevation, moisture moves upslope during the afternoon from the forested lower slopes towards the drier crater. This effect is stronger on the moister south-west slope. Seasonal and diurnal cycles were removed before trend analysis. Temperature anomalies show a significant warming trend over the 18 year period at all ten sites on the south-west slope with faster warming at higher elevations. Patterns are less distinct on the north-east slope with some cooling mid-slope. Vapour pressure anomalies show a significant increase at all sites except two, where there are possible microclimate changes which may confound the trends. The influence of ENSO is also examined, with El Nino/La Nina leading to significantly warmer/cooler temperature anomalies at most locations. The influence on moisture anomalies is more complex, with El Nino tending to lead to moister conditions on the south-west slope, but drier on the north east (i.e. an enhanced slope contrast). In the crater region increased moisture is dominant. La Nina effects are less consistent, but there is a predominance towards drier conditions spread over both slopes. Overall moistening of the mountain climate over time may appear somewhat inconsistent with observations of rapid ice ablation on the summit crater, but further investigation is required.

Distinct wet and dry seasons, weak easterly flow aloft, strong diurnal cycles and two north-south oriented mountain ranges modulate hydrometeological patterns in the northern Peruvian Andes. Herein we analyze trends in near-surface air temperature and water vapor from a 16-year-long (2006-2022) record of hourly measurements from satellite-linked HOBO weather stations and low-cost Lascar El-USB2 dataloggers hanging in trees along a road from the valley floor at 4000 to a ridge pass at 4600 m a.s.l. in the pro-glacial Llanganuco Valley. We also report on comparisons with a shorter overlapping time period in the Quilcayhuanca Valley about 60 km to the south. Data were physically downloaded from dataloggers during annual June-August field expeditions and through partnerships with the National Water Authority and Huascaran National Park offices in Huaraz, Peru. Although there are limitations and challenges common with mountain observations, we demonstrate the insights gained by maintaining these networks. We found strong temporal variability in freezing-level height (FLH) connected to ENSO cycles and greater warming rates at higher elevations driven in part by increased water vapor. Driven by marked reductions in near-surface lapse rate (LR) we projected an FLH rise of nearly 200 meters over the last decade with an average diurnal range of 150 meters during the wet and 400 meters during the dry seasons. While the observed rate of warming at 4000 m was less than 0.1 C/decade the warming at 4600 m was 0.8 C/decade. The linear trend of monthly LRs suggests negative LR feedback from 9 C/km in 2006 to 7.5 C/km in 2021. Averaging less than 2 m/s, wind speed at 4000m decreased from by 0.5 m/s over the last decade of the record. Absolute humidity rose less than 1% at 4000 m and by 10% at 4750 m. Our results and collaborations with climate modelers suggest that sustained, low-cost sensor networks provide data for hydroclimate studies and can help inform climate models about diurnal and seasonal meteorological forcing in alpine regions.

In the western United States, mountains substantially affect local and regional climate. Global Climate Models (GCM), which produce climate projections at a relatively coarse grid spacing (100-250 km) therefore must be downscaled for end-users to make informed decisions. At a grid spacing necessary to resolve topography (~6 km), many decisions have so far been guided using statistical downscaling methods. Statistical downscaling requires substantial bias corrections to the GCM’s precipitation field in mountainous areas, and assume that future precipitation and temperature distributions are consistent with the historical distribution. This work attempts to improve the physical processes represented in the downscaling methodology by dynamically downscaling 16 Coupled Model Intercomparison (CMIP) GCMs across the western United States using the Intermediate Complexity Atmospheric Research (ICAR) model. GCMs were selected for their ability to reproduce the historical climate and to span the range of future precipitation and temperature projections in the western United States. ICAR was also evaluated for its ability to represent the historical precipitation and temperature. ICAR successfully captured the inter and intra annual variability of precipitation and temperature and showed similar patterns of precipitation to both a reanalysis based Weather Research and Forecasting (WRF) simulation and gridded observations. Afterwards, the dynamically downscaled ICAR precipitation and temperature projections were compared to statistically downscaled methods, such as the Localized Constructed Analogs (LOCA) method and the Bias Corrected and Spatial Disaggregation (BCSD) methods, for projected annual changes as well as more local scale changes. In the northwestern United States, ICAR’s ensemble mean simulated an increase in cool season precipitation on the windward side of the Cascade Mountain Range compared to statistically downscaled techniques. In contrast, on the leeward side of the mountain range, ICAR showed a decrease in precipitation relative to statistical methods which is based on the coarse grid scale of the GCM and the transition from snow to rain and the different fall speeds of rain and snow. This signal was consistent between different GCM’s. Meanwhile, in the interior mountain regions of the Upper Colorado River Basin, which generates streamflow for 40 million people and millions of acres of agricultural land across seven US states, ICAR’s ensemble mean produced no substantial change in precipitation while statistical models show increased precipitation. Therefore, in the Upper Colorado, climate projections that use ICAR suggest a further decrease in projected annual streamflow than projections that used statistical models due to warming and evapotranspiration increases within the basin.

In regions with complex topography, incoming short- and longwave radiation at the surface are strongly influenced by local and surrounding terrain. These influences affect (near-)surface variables like (air) temperature and snow cover, which in turn can impact atmospheric processes like the development of valley wind systems. Topography has an influence on the following surface radiation components: direct incoming shortwave radiation is typically most affected, whereas the radiation flux depends both on local slope as well as on neighbouring terrain, which can induce topographic shading. Incoming diffuse shortwave radiation can be enhanced by terrain reflection – this process is particularly relevant for snow-covered areas featuring a relatively high surface reflectivity. Finally, incoming longwave radiation can also be modulated by radiative exchange between facing slopes. In many atmospheric and climate models, topographic effects on surface radiation are not considered at all. Radiation exchange is exclusively modelled in the vertical direction using the column approximation, which does not allow to consider the above-mentioned effects explicitly. However, different parameterisations, either based on geometrical considerations or derived from Monte Carlo ray tracing simulations, have been developed in recent years. These parameterisations are applied on a grid or sub-grid scale and for a wide range of grid spacings (~100 m to 100 km). However, it is still relatively uncertain how well these parameterisations perform on different spatial scales. To assess this uncertainty on a kilometre-scale, we perform regional simulations with the COSMO (Consortium for Small-scale Modeling) climate model for the European Alps with a geometrical-based parametrisation scheme. Besides quantifying the uncertainty of specific parts of the scheme, we also evaluate the overall impact on simulated climate. Some above mentioned effects of topography on surface radiation, like terrain reflected shortwave radiation and the exchange of longwave radiation between sloping terrain, reveal a pronounced dependency on the spatial scale. We therefore focus on a sub-grid parameterisation of these effects.

Systematic wintertime cold biases in two meter air temperatures (T2m) have been observed in a variety of kilometric scale regional climate and weather forecasting models for montane and alpine regions across across the mid-latitudes, including the Alps, Sierra Nevada, Rockies, and South American Andes. Nevertheless, T2m is a vital climate variable essential for forecasting applications ranging from snow hydrology, rain-snow partitioning, avalanche hazards, ecological productivity, and energy demand. Moreover, such biases obscure the interpretation of climate change and elevation-dependent warming signals from models, leading to potentially large uncertainties in how these vital systems respond to warming perturbations. In calm conditions, T2m is the manifestation of the often snow-covered land surface energy balance, the turbulence characteristics of the often stably stratified surface layer, the thermodynamic structure of the PBL, and terrain modulated flow features (cold pools, katabatic winds) all of which are under-observed and even under-theorized in complex mountain terrain, complicating the diagnosis of the source(s) of model bias. In this study, we first present a systematic review of the literature that shows the commonality of T2m cold-bias across a wide range of simulation domains and models (WRF, VR-CESM, COSMO-CLM). Second, we leverage new data from the DOE SAIL and NOAA SPLASH Field campaigns located at a high-elevation site in Colorado’s Rocky Mountains. Measurements include doppler lidar velocities of boundary layer winds, surface air and skin temperatures, temperature and humidity profiles, and eddy covariance measurements among others. As such we are able to characterize the diel cycles of the PBL and surface energy balance in significant detail and isolate potential failure modes of models for accurately diagnosing T2m. Observation-to-model comparisons are made against operational products and regional climate simulations, and surface-layer parameterizations responsible for diagnosing T2m are scrutinized. The potential causes of model T2m bias are articulated and the utility of intensive field observing campaigns for model development is demonstrated.

When anticyclonic conditions persist over mountainous regions in winter, inversion layers (i.e. cold-air pools) develop in valleys and persist from a few days to a few weeks. During these inversion episodes, the atmosphere inside the valley is stable and vertical mixing is prevented, promoting the accumulation of pollutants close to the valley bottom and worsening air quality. Mountainous areas are experiencing a warming rate twice stronger than the global atmospheric temperature, thus, they are particularly sensitive to climate change. However, with radiative cooling at the bottom and a warmer air at the top, the fate of persistent inversions is unclear: are potential temperature gradients in an inversion layer statistically reinforced, unchanged or weakened with climate change? This question is the purpose of this paper, which addresses climate change impact on persistent inversions in the alpine Grenoble valley. For this purpose, the long-term projections of the regional climate model MAR (Modèle Atmosphérique Régional) at resolution 7 km forced by the global climate model MPI (developed by the Max Planck Institute) are used to perform a statistical study of inversion episodes over the 21st century. The trends of the main characteristics of the inversions, namely their duration, frequency and intensity, are investigated for two different scenarios (SSP2-4.5 and SSP5-8.5). The results show that the intensity of the inversions displays a statistically significant trend in the 21st century for the worst-case scenario only, with a decay rate of 0.059+-0.007 K/km/decade. A similar result holds for the frequency of inversion episodes. The impact of climate change on the detailed structure of inversion episodes is next investigated by comparing two such episodes, in the past and around 2050 for scenario SSP5 8.5. For this purpose, the WRF (Weather Research and Forecasting) model, forced by MAR, is used at a high resolution (111 m). The episodes are carefully selected so that a meaningful comparison can be performed. We find that these episodes present similar atmospheric circulation and heat deficit across the valley depth but a different atmospheric stability and (therefore) inversion height. The future episode is characterized by a stronger atmospheric stability and a lower inversion height, and warmer air both close to the surface and in altitude (as expected from future climate projections). Overall, this study shows that, for scenario SSP5 8.5, the atmosphere in the Grenoble valley tends to be less stable in the future but that strong inversion episodes can still occur.

Severe convective events accompanied by hailstorms and thunderstorms can lead to significant damage. Despite the catastrophic nature of these events, it remains a challenge to understand the characteristics and mechanisms of such severe events due to difficulties in observing and modeling such events. The challenges come from the rarity and complicated processes involved. The combination of ground-based and radar-based products can provide valuable information for hail and lightning and support the evaluation of models. The Alpine region is among the most important thunderstorm peril regions in Europe due to its notable topography. In this study, we assess the hail and lightning-related severe convective events performed with the COSMO-crCLIM model (GPU version of the Consortium for Small-scale Modeling) at 2.2 km horizontal grid spacing over this region. We run 10 years of multi-seasonal (April-September) simulations driven by reanalysis data. The model is equipped with the one-dimensional hail growth model HAILCAST and lightning potential index (LPI) as hail and lightning diagnostics. First, we evaluate hail and lightning climatology produced by the model against available Croatian station records, Swiss radar-based estimates, and LINET lightning observations. The modeled and observed seasonal cycle, diurnal cycle, and frequency of hailstones sizes are discussed. Overall, they show similar patterns. The diurnal cycle of hail and lightning is more pronounced in mountainous regions than in coastal regions. A higher frequency of severe events in summer is found over higher elevated topography. These promising results reveal that both HAILCAST and LPI can provide valuable information on hail and lightning, and are good candidates in future climate-change studies.

The interplay of the Indian Monsoon and the Himalayas is vital to many climatological aspects of the Himalayan foothill and foreland regions. A unique climate feature in the Himalayan foothill and foreland regions is a bi-modal diurnal cycle of precipitation with high rainfall amounts in the afternoon and around midnight. The reason for this nighttime precipitation maximum is not yet fully understood, and current climate models and also reanalyses do not represent the regions’ diurnal precipitation cycle. Nevertheless, estimating realistic spatiotemporal precipitation patterns is crucial for the climate community (e.g., for impact modelling). This study reviews discussions in the literature, available observational findings, and simulation results with the regional climate models (RCM) COSMO-CLM & ICON_CLM. Our simulations indicate that the models are not able to recover the nighttime’s precipitation behaviour with currently typical horizontal RCM grid-spacings (e.g., 10 or 50 km), but they can do so with convection-permitting grid-spacing (~3 km), which sufficiently resolves the relevant and diverse orographic thermal winds together with the moist monsoonal flow characteristics in the area.

The horizontal grid spacing of numerical weather prediction models keeps decreasing towards the hectometric range. However, although topography, land-use, and other static parameters are better resolved than at the kilometric range, truly complex, mountainous terrain still poses a challenge for the models. One of the reasons for possible challenges for NWP models are 1D boundary-layer parameterizations missing 3D contributions, which leads to unrealistic simulation of the mountain boundary layer already at the kilometric range. Furthermore, the turbulence grey zone, where simulated turbulence is both parameterized and resolved, is relevant at sub-kilometric resolutions. Finally, one decrease horizontal grid spacings towards large-eddy simulation (LES) ranges, where the largest eddies are already resolved, for process studies. However, scale interactions also pose a challenge at high horizontal resolutions over mountainous terrain. In this presentation, we show three modelling studies with different aims and grid spacings (with three NWP models): NWP at 1 km to asses turbulence parameterizations, LES to study ABL processes and scale interactions over an Alpine glacier, and, lastly, simulations from 1 km spanning the turbulence grey zone towards LES resolutions, to test the model's turbulence schemes and identify the horizontal resolution at which a 3D turbulence scheme starts to add value over a typical 1D scheme. All simulations are accompanied by high-quality turbulence observations which allow process-based model evaluations. With this data pool, a thorough evaluation of the turbulence representation in NWP models is possible for well-known case studies across scales. We assess whether increasing the horizontal resolution automatically improves the representation of the thermally-induced circulation, surface exchange, and boundary-layer processes over complex topography. This allows us to identify key challenges for turbulence parameterizations at the hectometric range over complex terrain and suggest possible improvements.

Steep slopes in narrow and deep terrain are challenging for commonly used numerical weather prediction models using terrain-following coordinates since they can generate numerical errors when evaluating horizontal gradients, and numerical instabilities. The model orography is therefore generally smoothed locally so as to remove steep slopes exceeding a defined threshold (typically in the range 30–40º). This process can lead to undesired changes to the shape of large-scale orographic features. A global smoothing algorithm designed to preserve properties of the orography (e.g., mean terrain height across the domain) is developed and applied to the narrow and deep Grenoble valley. The valley-floor width is between 2 and 5 km, and the valley depth ranges between 2000 and 3000 m. Results of numerical model simulations are then analysed to evaluate the effects of the slope threshold on valley-scale atmospheric dynamics for a cold-air-pool episode in the Grenoble valley. The dynamics is found to be almost insensitive to the slope threshold when comparing two simulations with thresholds of 28º and 42º, for which the near-surface vertical resolution differs slightly (25 m for 28º and 32 m for 42º). This indicates that it is the large-scale (unsmoothed) orographic features that control the flow structures and properties at the scale of this valley. However, significant differences in near-surface wind (up to 3 m/s) and temperature (up to 4 K) between the two simulations are found locally where the difference in terrain height is significant (larger than 150 m).

The transport, diffusion and storage of passive tracers, such as greenhouse gases (GHGs), are controlled by Atmospheric Boundary Layer (ABL) processes. On a regional scale, local advection and turbulent transport and diffusion can modulate the vertical distribution of these trace gases. To test and optimise the representation of vertical mixing in numerical models, ICON hindcast simulations (model grid spacing of 1 km) have been performed over the Swiss Plateau for selected weather situations during summer and wintertime periods. These simulations are further evaluated against surface station and profiler observations and tower measurements at two the sites on Swiss Plateau namely Payerne and Beromüsnter, the latter a tall tower with a height of 212 m . The simulations have been done with different configurations of the operational turbulence scheme, based on a 1D Turbulent Kinetic Energy equation, and with a newly developed turbulence scheme, based on prognostic equations for two turbulence energies (2TE+APDF), and compared with meteorological observations of ABL profiles and surface fluxes at Payerne. Results for the stable wintertime situation show that with the two-energy scheme, the fog becomes more persistent in the model and agrees better with the observations. This improved simulation of the stable boundary layer will also be relevant for an improved simulation of periods with high air pollution and GHG concentrations. The results show that 2TE+APDF performs similar to or better than the combination of operational turbulence and shallow convection scheme in ICON for stable boundary layers.

We present a comprehensive analysis of a south foehn event in the Alpine Rhine Valley with ICON (Icosahedral Nonhydrostatic), including ICON in mesoscale mode with a horizontal grid spacing of 1 km (ICON-NWP) and in large eddy simulation mode with a horizontal grid spacing of 200 m and 40 m (ICON-LESs). The goal is (1) to study the ability and limitation of ICON-NWP over complex terrain in the context of foehn events and (2) to gain a better understanding of the foehn process in the Rhine Valley with ICON-LES. The three model setups are first evaluated with observational datasets collected from IOP2 of the PIANO (Penetration and Interruption of Alpine Foehn) project. As a second step, we study the evolution of a south foehn event in the Rhine Valley. The event took place from 20th to 24th in Nov 2016, driven by a significant deep low-pressure system. In the lower-Rhine Valley, the foehn flow experienced a temporary retreat and resumption. To understand the interaction between the foehn air and the cold air pool during this period as well as the foehn warming mechanisms in different valley locations, we deploy a Eulerian heat budget analysis over several control volumes throughout the Rhine Valley. Furthermore, the effect of unresolved small-scale side valleys in ICON-NWP is studied by comparing flow patterns and heat budget terms ICON-NWP and ICON-LES. This study reveals how the foehn air interacts with the underlying cold air pool and the local topography in the Rhine Valley during foehn and provides insights for the development of turbulence parameterization schemes over complex terrain.

Flow past an isolated steep mountain can lead to the formation of vortices, uplift, and clouds on its leeward side. Yet, the flow geometry on the leeward side may strongly depend on the properties of the oncoming flow and the size and shape of the orography. In our previous work we have carried through extensive Large Eddy Simulations, but so far we have mostly considered the time mean flow conditions. In the current work we widen our perspective and consider the transient aspects of the flow by performing a power spectral analysis of relevant time series. Our guiding question is whether the temporal structure contains periodic or quasi-periodic elements (such as vortex shedding) and under what circumstances they occur. Laboratory experiments and numerical simulations with idealized obstacles suggest quasi-periodicity, which can be quantified by the so-called Strouhal number. Indeed, for our numerical simulations with an idealized pyramid-shaped mountain we find a striking single peak in the power spectrum, and frequency corresponding to this peak decreases with increasing wind speed, as predicted by theory. We then go on to show to what extent this phenomenon depends on the upstream wind profile and on the exact shape of the orography.

Coherent plume structures characterize the convective boundary layer (CBL) over both flat and non-flat terrain. Over orography, they are found to be embedded in local thermal circulations developing over heated valley slopes. For this study, boundary-layer plumes over hilly terrain are investigated using large-eddy simulation. A conditional sampling method based on the concentration of a decaying passive tracer is implemented in order to identify the boundary-layer plumes objectively. Conditional sampling allows to quantify the contribution of plume structures to the vertical transport of heat and moisture. A first set of simulations analyses the flow over an idealized valley, where the terrain elevation only varies along one horizontal coordinate axis. In this case, vertical transport by coherent structures is the dominant contribution to the turbulent components of both heat and moisture flux. It is comparable in magnitude to the advective transport by the mean slope-wind circulation, although there are considerable differences between heat and moisture transport. A set of less idealized simulations considers the flow over three-dimensional terrain. In this case, conditional sampling is carried out by using a simple domain-decomposition approach. We demonstrate that thermal updrafts are more frequent on hilltops, but they are less persistent on the windward side when large-scale winds are present in the free atmosphere. The latter outcome is related to the finding that vertical moisture transport over complex terrain is reduced by upper-level winds.

Prandtl (1942) model for thermally-driven winds on ideal slopes provides a useful scheme to model the parallel flow promoted by a steady heating or cooling on a plane incline in an otherwise quiescent and stably stratified atmosphere. The model accounts for the effect of along-slope advection of the stably stratified atmosphere, whereas the assumption of along-slope invariance of this balanced flow implies neglecting density stratification. However this is inconsistent, as the density stratification is intrinsically associated with the thermal structure of the unperturbed atmosphere. Such an inconsistency is removed taking into account the full continuity equation and including a small, but nonnegligible, slope-normal velocity component. It turns out that the full solution can be expressed as a perturbation expansion, where at the leading order the solution reproduces the classical structure envisaged by Prandtl (1942) for the along-slope component, whereas at the second order includes a weakly nonlinear correction of the along-slope wind velocity component. This correction marks a difference between up-and downslope winds, consistently with results from field observations and numerical model simulations. The slope-normal velocity component accounts for entrainment/detrainment effects also observed in field measurements. Reference Prandtl, L.. 1942: Führer durch die Strömungslehre, Verlag Friedrich Vieweg & Sohn, Braunschweig.

The vertical temperature gradient dictates stability and is therefore one of the fundamental properties characterizing the boundary layer. In similarity theory, vertical fluxes are a function of the local gradient, but the limited vertical resolution of traditional point measurements require an assumption of the shape of the temperature profile. As a result, there are a variety of conceptual and theoretical frameworks describing the temperature profile of the boundary layer, e.g., assuming a log temperature profile between two levels. However, boundary layers influenced by submeso-scale processes (i.e., the stable boundary layer in complex terrain) can frequently invalidate these assumptions in ways that are hard to observe. Distributed Temperature Sensing (DTS) has the unique ability to resolve these processes with an effective resolution of ~0.5m in space and 10s in time. Vertically oriented DTS was deployed near the surface in two mountain valleys: 1) a mid-mountain valley (Fichtel Mountains, 12m vertical array, valley depth 50m-200m) and 2) a deep mountain valley (Inn Valley, 17m vertical array, valley depth 1500m). Using DTS we explicitly describe the “shape” of the nighttime boundary layer through clustering, yielding unique views of temperature profiles that do not adhere to textbook frameworks. With these clusters, we test the conceptual and empirical representations of temperature profiles and describe the impact of the temperature profile shapes on turbulent quantities. Temperature profiles vary at such fine scales that it is only possible to correctly estimate these profiles using vertically dense temperature observations, such as from DTS. These results highlight that estimating stability is a non-trivial exercise and suggest stability may be poorly constrained at most research sites using typical observation strategies (e.g., log-linear distribution of temperature sensors). These results provide observational and analytical strategies for improving estimates of the temperature profile and can be used to inform the design of future experimental studies.

Mountain ecosystems, which play an important role in facing the current and future climate crises, are supposed to be heavily affected by climatic changes and extremes. Therefore, efforts towards an accurate quantification of the carbon and water cycles in mountain forests and grasslands are crucial to determine their contribution to reaching the anthropogenic CO₂ emission reduction targets. However, among the challenges in quantifying land-atmosphere interactions in complex terrain, the unaccounted presence of advective CO₂ fluxes has the potential to bias the daily and longer-term CO₂ flux measurements towards unrealistic net uptake, a bias that still needs to be accounted for in order to reduce uncertainties related to role of ecosystems in the global carbon cycle. Here, we present results from three CO₂ advection experiments conducted at a European larch forest and a mountain grassland in Italy (2100 m asl). The setups consisted of: the main eddy covariance flux tower at each site, a sub-canopy eddy covariance flux system at the forest site, a home-assembled system for measuring CO₂ concentrations at three heights on the four sides of a 40 x 40 m control volume, a transect for measuring advection in a simplified 2D approach, and four automatic chambers measuring bare soil respiration (opaque chambers) and the net ecosystem CO₂ exchange from the vegetated forest floor (clear chambers). Results show that: i) advection is a not-negligible fraction of the total net ecosystem CO₂ exchange of these two ecosystems, ii) advection values are higher for the forest compared to the grassland site, iii) coupling measurements of above and below canopy eddy covariance in mountain sites could be essential for detecting/estimating the unaccounted CO₂ efflux, and iv) approaches for correcting for the missed fraction of the net ecosystem CO₂ exchange carried by advection could be applied to reconcile eddy covariance fluxes with estimates derived from vegetation models.

Turbulence is typically weak and oftentimes intermittent in the stable boundary layer, particularly under very stable conditions. Over complex, mountainous terrain, a number of locally induced processes occur regularly, which can impact near-surface turbulent exchange. Nocturnal temperature inversions, cold-air pools, thermally driven slope and valley winds, gravity waves, and meandering motions form frequently during synoptically undisturbed and clear-sky nights. A multi-year dataset is analyzed from five i-Box stations in the Inn Valley, Austria. The i-Box (Innsbruck Box) is a long-term measurement platform that was designed to study boundary-layer processes in complex, mountainous terrain. The five eddy-covariance stations are located within an approximately 6.5-km long section of the 2-3-km wide and 2000-m deep valley, with one station at the almost flat valley floor, one station at the foot of the south-facing sidewall, one station at an almost flat plateau along the south-facing sidewall, and two stations on steep slopes of the north-facing sidewall. In this presentation, the observed turbulence intensity and intermittency at the five stations will be related to the different flow processes. Depending on the location of the stations within the valley atmosphere, different processes dominate the local flow field. While the slope sites are characterized by downslope flows, which cause relatively large wind shear, stations close to the valley floor are frequently located within a strong valley inversion, with low wind speeds and oscillatory motions, each impacting the near-surface turbulent exchange in different ways.

The strong heterogeneity of the mountain snow-cover across different scales is mainly induced by complex snow -atmosphere interactions. We present findings from recent experimental studies on snow-atmosphere interactions which motivate new modelling strategies for the Swiss operational snow-hydrological model system. For the snow accumulation season, we can show that the local flow field at the ridge scale determines precipitation patterns by advecting hydrometeors downwind of mountain ridges and peaks. This process plays a decisive role in the final distribution of snow in mountain catchments. The HICAR (High-resolution Intermediate Complexity Atmospheric Research) model, a new model variant of the ICAR model, was developed to account for the interaction of the terrain-induced local flow field with the precipitation field. This is achieved through a novel combination of adjustments to a background wind field based on terrain descriptors, followed by a wind solver enforcing a mass-conservation constraint on the 3D wind field. This includes resolving blocking and channeling of winds, as well as speed up on the windward side of terrain and slowdown and recirculation on the leeward side. Importantly, these effects were incorporated without a large computational cost. We present comparisons of flow patterns and snow deposition maps of HICAR in comparison with WRF model outputs. With this new model, physically-based downscaling of precipitation and other atmospheric variables is made available for high-resolutions (100m) and large-spatial extents (10,000 km2) which are often demanded by operational land-surface models. For the melting season we present various experiments demonstrating the high spatio-temporal variability of heat and momentum fluxes over heterogenous land surfaces such as patchy snow covers or glaciers. Experiments using a high-resolution thermal infrared camera pointing at synthetic screens and multiple eddy-covariance sensors show the dynamics of turbulent temperature fluctuations over the patchy snow cover, horizontal heat advection, the formation of stable internal boundary layers (SIBL) and the penetration of gusts thorough the SIBL. Our experiments also highlight high temporal and spatial dynamics of local turbulence profiles of momentum and heat, indicating strong changes of the local thermodynamics at a glacier when larger-scale flows disturb the persistent katabatic flow. These finding motivate the coupling of a snow process model (FSM) to HICAR to solve the energy and mass exchange at the snow-atmosphere interface in the vertical but also in the horizontal direction. This allows us by the first time to fully account for complex snow-atmosphere interactions and feedback mechanisms.

The wind-precipitation-terrain interactions during snowfall, leading to preferential deposition, strongly alter the actual deposited snowfall in mountainous terrain. Although preferential deposition can be modelled with reasonable accuracy, fine scale 2D or 3D wind fields are required and existing wind modelling approaches need careful numeric model pre-adjustments and have high computational costs. With current climate projections indicating an increase of extreme precipitation events, accurately representing spatial snowfall patterns is becoming increasingly important. However, the required high-resolution wind fields are generally not available for large regions and over longer periods and broad model application is thus not feasible. Our goal was therefore to develop parameterizations to spatially downscale coarse snowfall information to the topography using as little fine-scale temporal input as possible and at low computational cost. We simulated preferred snowfall deposition patterns at fine scales using the snow transport module of a surface process model Alpine3D under controlled conditions forced with 3D wind fields computed using the non-hydrostatic, compressible atmospheric model ARPS (Advanced Regional Prediction System) on a large ensemble of synthetic mountains. The resulting database for fine-scale snowfall distributions, near-surface wind and topography characteristics allowed us to systematically investigate relationships between wind flow, topography, and deposited snow particles. We developed two statistical downscaling schemes: (1) a scheme using a combination of local surface vertical wind speed and local terrain slope; (2) a scheme based on spatial mean horizontal surface wind speed, coarse surface wind direction and local terrain slope and aspect. Using the first significant seasonal snowfall in a small alpine catchment above Davos, Switzerland provided an independent spatial snow depth data set for evaluations in a real mountain environment. Snow depth data was acquired using a photogrammetric drone survey. The event was modelled using the snow transport module of Alpine3D forced with ARPS wind fields. Downscaled and modelled preferred snowfall compared well. Downscaled snowfall from a previous statistical downscaling scheme resulted in fewer spatial deposition patterns and more scatter. Larger discrepancies with measured snow depth were observed on the steepest slopes, which were likely due to avalanches that released during the snowstorm or from previous snow deposits. Overall, our results show that the proposed downscaling schemes can be used to reliably reproduce spatial snow depth patterns with a computationally very efficient method and could be used for various model applications such as in hydrology, climate impact studies, or weather forecasts.

Data assimilation (DA) is the process of estimating the state of the atmosphere, or any other dynamical system, by combining observational information and a model of its evolution (the background). DA methods of different complexity exist, ranging from observation nudging to variational, ensemble, and hybrid approaches. Several variants of the ensemble Kalman filter have emerged as efficient techniques to assimilate conventional atmospheric observations in convection-permitting numerical weather prediction (NWP) models. They are thus routinely used by forecasting centres worldwide. Under restrictive assumptions, ensemble DA results in a statistically optimal estimate of the atmospheric state. These assumptions include linear evolution of the dynamical system, small departures between observations and background, absence of observation and model biases, Gaussian distributions of observations and background errors, and linearity of observation operators. Some of these constraints are easily violated in mountainous environments, particularly in the mountain boundary layer. The most fundamental problem is systematic model error, which is typically large due to inadequacies in boundary-layer parameterizations and is further amplified by deviations between the model orography and the truth. Model error, combined with observations that are too sparse to capture the high degree of spatial variability of the mountain boundary layer, typically causes suboptimal DA, and ultimately inaccurate analyses. In this contribution, some applications and future prospects of DA research in the field of mountain meteorology are presented. (i) Documenting model errors using DA products: A long-term analysis of first-guess departures from the convection-permitting analysis ensemble at Meteo Swiss, based on the COSMO model, reveals time-dependent biases in temperature analyses. (ii) Correcting model errors with DA techniques: The state augmentation method enables the minimization of errors both in state variables and in uncertain model parameters, based on observational evidence. (iii) DA in the turbulence grey zone: As operational NWP models approach sub-km grid scales, the computation of analyses that partially resolve the largest scales of atmospheric turbulence becomes feasible, at least in principle. Such high-resolution analyses are not necessarily useful as initial conditions for weather forecasting, but have several other potential applications. A few outstanding challenges that, in addition to model error, are relevant for the computation of grey-zone analyses are discussed.

The initial conditions of numerical weather predictions are provided by a data assimilation process, which combines the latest available forecasts (first guesses) and the latest available observations into a statistically optimal estimate of the atmospheric state. Most operational data assimilation algorithms are designed under the assumption that the frequency distribution of forecast errors is Gaussian and has zero mean, i.e., there is no systematic error. However, imperfect model formulation is an inevitable source of systematic error, which ultimately leads to inaccurate estimates of forecast uncertainty and suboptimal use of observations in data assimilation. Over mountainous terrain, oversimplified parameterization schemes (in particular for surface-layer and boundary-layer exchange), combined with large deviations between the modelled and true orography, lead to pronounced model errors. In this work, we use an ensemble post-processing method known as SAMOS (Standardized Anomaly Model Output Statistics) to correct systematic deviations between observations and first guesses prior to the data assimilation. A 5-year archive of observations and first guesses from the MeteoSwiss COSMO-1E analysis ensemble is considered, with emphasis on 2-m temperature forecasts. We demonstrate that the SAMOS correction reduces systematic errors both at mountain-top and valley sites, as well as their seasonal and diurnal variations. In addition, SAMOS processing enhances the standard deviation of ensemble first guesses. The procedure ultimately results in a better estimate of the forecast uncertainty and in a very nearly Gaussian distribution of forecast errors. Based on these results, we expect that the implementation of SAMOS processing of temperature innovations in the data assimilation cycle can lead to fewer observations being rejected and to an increased weight of the assimilated observations in the analyses.

The SwissMetNet surface observation network operated by MeteoSwiss provides a dense coverage of near-surface measurements in Switzerland. In September 2021, 2m temperature and humidity measurements observed by these stations have been introduced in the operational data assimilation system KENDA. This system is based on an ensemble Kalman filter and runs the COSMO model at 1km grid spacing. In this contribution, we present the benefit and some challenges of the 2m temperature and humidity assimilation. The ensemble Kalman filter successfully assimilates the observations using realistic ensemble correlations taking into account the complex topography. Including the new observations significantly improves the temperature and humidity profiles of the near-surface model atmosphere and the representation of fog and low stratus in the COSMO model forecasts. Despite the improvements gained in the forecasts, some challenges remain. Especially in complex terrain like the Alps in our domain, near-surface measurements are not always representative for their surroundings. This is largely related to the height differences of the model and real-world topography, but also to imperfect parametrizations and other model deficiencies. To mitigate problems arising from height differences in the topography, we discard observations from stations with a height difference of more than 150m. Another issue is the vertical representativeness of near-surface measurements in strong inversion situations. Those situations are quite common on the Swiss Plateau, especially in winter. When the inversion layer is very close to or at the surface, the differences between model data and 2m measurements are only representative for a shallow vertical layer and not valid higher up in the atmosphere. This can cause a degradation of the analysis quality. We show examples from more than a year of practical assimilation experience.

GLORI (Global-to-Regional ICON) is a trilateral project between Germany, Italy and Switzerland, which aims at developing a global-to-regional Digital Twin based on the ICON modelling system. ICON is used for operational numerical weather prediction (NWP) at DWD and a growing number of partners within the COSMO Consortium. One of the high-resolution (~500m mesh size) regional model configurations targeted within GLORI will focus on the greater Alpine region (GLORI-Alps). The central project goal is to improve the prediction of extreme weather events and their impact on the community by providing on-demand high resolution forecasts for selected regions including atmospheric composition and hydrological modelling. From a technical point of view, a key requirement is the capability of the system to run on supercomputers accelerated using GPUs in order to attain the computational performance needed to provide forecasts at such high resolutions within a reasonable amount of wallclock time. Moreover, the enormous data amounts provided by such model configurations require revising the traditional workflow of storing everything on disks and in tape archives. To tackle this issue, a so-called Data Lake will be established to accomplish a timely data transfer to downstream models and end users. After a brief project overview, our presentation will focus on (ICON) model development aspects relevant to Alpine meteorology, and in particular to dedicated ultra-high resolution forecasts planned for the TEAMx Observational campaign in 2024/2025. Specific issues known at the time of writing this abstract include (i) a systematic dependence of boundary-layer mixing on the horizontal model resolution, getting too strong when moving from 2 km towards finer scales; (ii) the treatment of subgrid-scale orography at scales of O(1 km); (iii) apparent deficiencies in the wave-mean-flow interaction of resolved breaking gravity waves, which become most evident over the Himalayas. Regarding the last item, we find that ICON provides the best forecast scores over and in the lee of the Himalayas (Southeastern China) at the operational global mesh size of 13 km, where the effects of breaking gravity waves are largely parameterized. At higher model resolution, the forecast quality progressively deteriorates until reaching mesh sizes of O(1 km), where wave breaking is explicitly simulated. An additional important work package within the EXCLAIM project (which is closely related to the GLORI project) is coupling the LES-capable 3D Smagorinsky turbulence scheme to the surface transfer scheme used for operational NWP, which is a prerequisite for using this scheme for NWP purposes.

Winter precipitation poses a major challenge for numerical weather prediction and operational weather forecasting and frequently leads to air- and surface-transportation disruptions, vehicle accidents, avalanches, and other hazards. In the western continental United States (CONUS), complex terrain strongly modulates precipitation amount and type, necessitating kilometer- or sub-kilometer-scale prediction for applications from winter-road maintenance to avalanche mitigation activities. Over the past several years, the University of Utah has produced high-resolution snowfall forecasts over the western CONUS using operational modeling systems run by national and intergovernmental modeling groups including the global ensembles [e.g., the North American Ensemble Forecast System (NAEFS) comprised of the US and Canadian global ensembles] and the US Global Forecast System (GFS). The NAEFS consists of 52 members run at relatively low resolution and the GFS is a deterministic model run at ~13-km grid spacing. Quantitative precipitation forecasts (QPF) from these modeling systems are downscaled using climatological precipitation analyses to 800-m grid spacing. High-resolution snowfall forecasts above the estimated snow level are produced by machine learning techniques based on training to high quality manual snow-to-liquid ratio observations at several high mountain sites. Application to the 52 NSEFS members enables the derivation of probabilistic quantitative snowfall forecasts. The approach is relatively straightforward and does not require a long record of high-resolution precipitation analyses and forecasts. The latter is often difficult to obtain from operational forecast systems. Verification relative to observations collected by automated weather sensors and snow-safety professionals indicates that these forecasts improve upon model output and existing SLR techniques used by the US National Weather Service. Ongoing work to incorporate these techniques into the US High Resolution Rapid Refresh and next-generation Rapid-Refresh Forecast System will also be presented.

The re-routing or postponing of flights due to thunderstorms is one of the main causes for weather-related air traffic delays in Europe, leading not only to significant financial losses but also to increased CO2 emissions. Efficient and safe air traffic management therefore relies on highly accurate information about the location and timing of convective initiation and convective development at time-scales greater than the nowcasting range. However, this poses a great challenge even for convection-resolving numerical weather prediction (NWP) models, especially in regions with complex topography as in Switzerland. This study aims to provide improved probabilistic convection forecasts to support the pre-tactical flight planning phase which ranges from 2 to 16 hours into the future. We present the capability of COSMO-1E, which is the currently operational convection-resolving NWP model at MeteoSwiss, to predict convective activity over Switzerland and its bordering countries. COSMO-1E is an ensemble prediction system that is run every 3 hours at a spatial resolution of 1.1 x 1.1 km2. We compare traditional methods of convection prediction based on the direct model output from COSMO-1E against a machine learning (ML) approach. The ML system was trained using COSMO-1E forecasts from the convective season 2022 for the greater Swiss area and validated against lightning observations.

Trapped lee waves and turbulent rotor activity are a hazard for aviation, so forecasters must take into account the likelihood of lee wave activity when advising aviation personnel. The UK Met Office’s operational high resolution Numerical Weather Prediction model, UKV, resolves lee wave activity but there is currently no automated operational way to recognise such regions directly from the high resolution UKV data, short of a forecaster looking at the data themselves. If lee waves can be detected from the UKV data, we can use this output to gain a better understanding of lee waves and their impact over the UK. For example, over which regions are lee waves likely to form? Which atmospheric conditions are the most conducive to lee wave propagation? A machine learning (ML) model, a U-Net, was trained to identify lee waves, and then retrained to derive characteristics about the waves (orientation, wavelength) from the model data instead. The wave amplitudes are also extracted using the wave mask from the segmentation model. These ML models were then applied to a 30 year dataset (1982 - 2012) of high resolution NWP output, driven by ERA-Interim data. From this, the location and characteristics of lee waves over Britain and Ireland from within this time period were extracted. In this talk, we explore the climatology of lee waves in the data over this period, including the relationship between waves and the orography; frequency of waves; seasonal effects; the effects of weather patterns on waves; and diurnal effects. In addition, the same ML models are applied to a high resolution regional climate model dataset (using the RCP8.5 emissions scenario) to analyse how waves might change in a warming climate. For example, are waves with stronger amplitudes produced? Are waves more likely?

Solar and thermal radiative heating and cooling is a key driver of valley flow. However, most of todays LES and NWP models employ 1D radiation schemes which neglect the horizontal transfer of energy. While hill-shading methods or tracing of elevation-maps allows to incorporate 3D radiative effects in clear sky scenes, i.e. as long as the obstacles are static, these methods fail in cases with dynamic heterogeneities such as clouds. We extended the TenStream radiative transfer solver to fully resolve 3D radiative effects in LES simulations. We present implementation details and results from simulations within WRF-LES (terrain-following coordinates) and PALM-LES (immersive boundary implementation).

How can tornadic supercells be influenced by complex orography? In this study the authors address this issue by studying a tornado outbreak that affected the Po Valley in northern Italy on 19 September 2021. During the event seven tornadoes (four of them ranked as F2 according to the Fujita scale) developed between Lombardia and Emilia-Romagna regions in a few hours. Although tornadoes are not rare in Italy, so many tornadoes in such a short time is an unusual event. The event was studied exploiting observations and numerical simulations obtained with the convection permitting MOLOCH model. Observations showed that during the event there were two low-level boundaries in the Po Valley: a cold front coming from the Alps and a dry line generated by the downslope winds from the Apennines. These two boundaries created a triple point, like those observed during tornado outbreaks in the US Midwest, but on a smaller scale. Numerical simulations with 500 m grid spacing showed that a warm and moist air tongue from the Adriatic Sea played a fundamental role in generating the supercells. Moreover, by means of numerical experiments, it has been proved that the structure and location of the moist air tongue was sensitive to the Froude number of the south-westerly flow from the Apennines: the greater the Froude number, the further north and narrower was the tongue of air, with impacts on the development of supercells. Finally, the dry line played a key role in the generation of tornadoes. In fact, kinematic and windshear parameters were comparable to typical values observed in the US-tornado events only along a narrow path ahead the dry line.

In preparation for the TEAMx field campaign, a group of scientists within the working group dedicated to orographic convection identified the week of July 23-29 2019 as an interesting inter-comparison period for state-of-the-art convection permitting models simulating real-case of orographic convection. This period has the advantages of being quite recent and representing a typical transition from convectively stable conditions to days with localized/stationary convection finally evolving into 1-2 days with organized convection. The focus area of the study is the Inn valley in Austria, but, for most of the days, convection occur and is sometimes also organized on larger scales. The AROME, ICON, MesoNH and WRF models were run for this inter-comparison, with a horizontal grid spacing close to 1km and explicit deep convection. In order to focus on the impact of model physics, models were run on a common domain, starting from ECMWF operational analysis and using hourly ECMWF forecasts as boundary conditions. Particular focus will be given to the location of the convection triggering in the models, and to the location/intensity/chronology of the precipitation events. Simulations will be compared to model-independent objective analyses products ingesting radar and/or raingauge observations (such as those provided by the INCA system in use at GeoSphere Austria, and by the VERA system in use at UBIMET). Results will illustrate the current strengths and weaknesses of NWP models in forecasting of typical Alpine convective events.

This contribution aims at presenting results from a model intercomparison study developed in the framework of the Working Group “Mountain Boundary Layer” of the TEAMx programme, with the main objective of evaluating the ability of different numerical weather prediction models to reproduce the development of thermally-driven winds and the associated thermodynamic fields in a mountain valley in real-case simulations. The use of real-case simulations allows the assessment of model performance against observations using a setup representative of operational forecasts, giving the opportunity to better discriminate the reasons behind the differences between models and to identify the physical processes mostly responsible for such discrepancies. Moreover, the simulation of a full diurnal cycle allows the evaluation of model skills in the different phases of thermally-driven circulations in a mountain valley (e.g., daytime vs. nighttime, transitions). Different meteorological models participate in this initiative, including both operational and research models, i.e., AROME, GRAMM-SCI, ICON, Meso-NH, the Unified Model and WRF. Simulations are performed with a common setup regarding horizontal and vertical resolution, domain extent, initial and boundary conditions, characteristics of land cover and orography datasets and duration of the simulation, in order to help the comparability and the interpretation of the results. In particular, one domain covering the entire Alps is used, with horizontal grid spacing of ~1 km and vertical grid spacing varying from ~20 m near the ground to ~400 m in the stratosphere. Initial and boundary conditions are provided by ECMWF Integrated Forecasting System (IFS) forecasts on model levels. Moreover, for some models, different simulations are performed, to evaluate the sensitivity to physics options or model settings. Model results are compared with measurements collected during the Intensive Observational Period 8 (IOP8, 13 September 2019) of the CROSSINN field campaign, performed in the Inn Valley (Austria), about 20 km east of the city of Innsbruck. IOP8 was characterized by weak synoptic forcing and clear-sky conditions, with the full development of thermally-driven winds. A wide range of instruments are available for model evaluation, including standard surface weather stations, flux towers, radiosondes, Doppler lidars and microwave radiometers, allowing for an extensive assessment of model performance and the identification of the physical processes mainly responsible for simulation errors. Model results are evaluated mainly considering the surface energy budget, the diurnal cycle of temperature, pressure and wind speed near the surface, and the vertical profiles of temperature and wind speed within the mountain boundary layer.

Wind and visibility conditions strongly influence the takeoff and landing capacity of airports. It is, therefore, crucial to predict wind and visibility to a degree that allows plannable aircraft operations within the next hours and days. In our case, the complex topography around the airport of Zurich makes a correct prediction challenging. While locally strong tailwinds may cause go-arounds, fog patches often slide onto the airfield from surrounding wetlands. To inform Air Traffic Control (ATC) about these weather events, MeteoSwiss currently issues a three-hourly ensemble wind prediction at a 10-minute and hourly resolution on a 1.1 x 1.1 km² grid, and a deterministic wind nowcast every 10 minutes for a single location at Zurich airport. For visibility, every three hours a probabilistic hourly forecast estimates the expected horizontal visibility conditions. We aim to provide an improved prediction of wind and visibility to ATC, at different locations distributed over the airport. We do so by developing Machine Learning (ML)-based probabilistic predictions for visibility and wind with a higher update rate and a homogeneous resolution of 10 minutes for the first few hours. The ML predictors include both numerical weather prediction inputs and additional measurement sources. This contribution presents the particularities of wind and visibility conditions at the airport of Zürich, their prediction with different ML approaches, and an evaluation scheme using other available predictions of visibility and wind.

Seasonal weather forecasting is challenging over complex topography areas due to an inaccurate representation of the local physical mechanisms at play. This study's goal is thus to evaluate the seasonal weather forecast biases and their variability in relation to elevation and seasonality in the Trentino-South Tyrol (north-eastern Italian Alps) region as a case study. To this end, the SEAS5 seasonal weather forecast data (i.e., total precipitation and 2 m temperature) from ECMWF has been used on a spatial scale of 0.125° x 0.125° with 25 ensemble members in the period 1981–2016, and it has been compared with a high-resolution observational gridded dataset (250m x 250m) of daily precipitation and mean temperature. Given the significant linear dependence identified, a simple monthly-dependent linear correction model has been tested in which regression parameters are obtained by splitting the data into calibration (70%) and validation (30%), implementing a random subsampling technique. Our research demonstrates that the linear dependence of monthly weather forecast (i.e., precipitation and temperature) biases on elevation is time-varying due to the different natural dynamic processes that occur at different times of the year and varies with ensemble members considered due to the different initial conditions. Thus, not only static environmental variables (i.e., elevation) but also respective seasonality and ensemble members should be considered for seasonal weather forecast corrections.

The Meteomatics Meteodrone is a small Unmanned Aircraft System (UAS) designed to collect high-resolution vertical profiles of atmospheric parameters such as temperature, humidity, wind speed and direction, and barometric pressure. Since its founding in 2012, Meteomatics has undertaken an iterative development of the Meteodrone technology, with regular releases of incremental enhancements. The newest model, the MM-670, features major improvements to measurement accuracy, flight capabilities and reliability, and safety and regulatory compliance, making it suitable for routine operational and research use. In 2023, Meteomatics is continuing to install a network of 15 Meteodrones around Switzerland. In this project, Meteodrones are launched remotely from semi-automated Meteobase drone-in-a-box systems and routinely flown to collect profiles to a maximum of approximately 6,000 metres AMSL. The data collected by the Meteodrones can fill gaps in the existing weather observation network, especially in the boundary layer regions where extreme weather events occur. This data is used for nowcasting purposes and to improve the accuracy of Meteomatics-developed 1 km-resolution Weather and Research Forecasting (WRF) models. In addition, Meteodrone systems can provide extremely useful information about the atmospheric state in mountainous areas, particularly where there is limited coverage from radiosondes or remote sensing. In this presentation, we will share learnings from our research that has enabled the transition to operational use of Meteodrones, including how the technology has evolved and specific issues have been investigated and addressed. In addition, future applications of this new technology, including improvements to nowcasts and forecasts in Alpine regions, will be discussed and evaluated.

The Northern Alpine forelands in Switzerland, Southern Germany and Austria suffered an unprecedented series of hailstorms between mid-June and late July 2021, which turned that year into the costliest on record for many hail insurance companies. These events characteristically unfolded in a deep southwesterly flow ahead of a longwave trough over Western Europe, the dominant weather pattern during these six weeks. This pattern infamously brings together the conditions for severe thunderstorms north of the Alps, where moisture-loaded upvalley and upslope circulations undercut an elevated mixed layer (EML) that the mid-level flow carries away from the high terrain. Deep convection usually initiates at the dryline near the Alpine rim, then intensifies and organizes when it moves against the upvalley wind systems over the forelands. In this regime, the capping by the EML temporarily suppresses vertical mixing of moisture and momentum, and thus bottles up both CAPE and vertical wind shear into a volatile “loaded gun” environment. A comparison of ERA5 reanalyses and reported hail events for the 2016-2022 period, obtained from social media platforms and from fire brigade data, establishes a training dataset that highlights the crucial role of steep vertical temperature gradients across the lower troposphere for the generation of large hail. Afterwards we focus on the 20-24 June 2021 hailstorm outbreak, which vividly illustrated the delicate interplay between the synoptic background and mesoscale, orography-related processes. We employ a mixture of high-resolution fields from the AROME numerical weather prediction model, observational data and webcam imagery to disentangle these processes. A special focus is on the origin of the EML, which can both be advected off elevated, dry landmasses like Iberia or northwestern Africa, or form in-situ over the Alps under strong insolation. Using analogues of remarkable hailstorms in other parts of the Alpine region, we find that the dominant reason for a maximized hail hazard along the Alpine rims is the presence of high and dry terrain nearby, though a prior trajectory over the Spanish Plateau strongly “primes” an airmass for a rapid build-up of steep vertical temperature gradients. The climatological footprint of hailstorms along with a solid understanding of relevant processes can give a forecaster on duty some valuable guidance. The most important lessons from this exceptional hailstorm outbreak can help with a quick decision-making in time-critical warning situations.