In November 2018 the new CH2018 climate scenarios for Switzerland were released as part of the national climate change adaptation strategy. The CH2018 dataset is based on the EURO-CORDEX regional climate model (RCM) ensemble and will serve as reference input for a broad range of climate impact studies in Switzerland. A variety of products serving different purposes are offered and are distributed to users. These products include simple graphics and quantitative ranges of climate change signals over different regions of Switzerland, but also transient daily data that can be directly used to drive a variety of impact models. The transient scenario data have been generated by bias-correcting and downscaling the RCM output using empirical uni-variate quantile mapping.

User feedback so far indicates a high value of the scenarios, but intrinsic data limitations and methodological concerns prevent their direct use in specific individual applications. In our workshop contribution we will summarize the possibilities of the new CH2018 scenarios, but also their basic limitations. Furthermore, we will give an overview on additional user needs that have been collected so far and on our strategy to accommodate these additional needs and thereby further develop the CH2018 scenario dataset.

The project Trajectories aims to improve our understanding of the interactions between mountain societies and their environment and the way they adapt in the face of climate disturbance. The study period covers the mid-19th century (1850) to the mid-21st century (2050).

Three French alpine sites are identified as pilot sites: the Arve valley, the Maurienne valley, and the Meije area. These three regions are chosen for the various challenges they face when dealing with the effects of climate change. The Arve Valley, at the foot of Mont Blanc, is subject to changes in hydrological, glacial and flood hazards linked to global warming. The Maurienne Valley has an economy marked by the triptych tourism - agriculture - industry strongly linked to climate change. Finally, the Meije area benefits from alpine biodiversity threatened by climate change and land use planning.

In order to investigate the co-evolutions between climatic stresses and the different "objects" studied within these valleys (snow cover; agriculture; mountain and snow tourism; local economy; biodiversity; pastoral practices) through multidisciplinary modelling approaches, we first need to identify the climatic variables of interest by conducting a series of interviews with a representative panel of scientists of the project (historian; economist; ecologist; statistician of extremes; social geographer specialized in alpine mammals). This will allow us to identify the nature, spatial and temporal scales and resolutions of the climatic proxies that are considered as potential “critical” variables in the studied systems.

Therefore, we will evaluate how such proxy variables could be retrieved from regional climate simulations realized within the past (1900 – 2010) and the future periods (2010 – 2100) with the MAR model (Gallée et al., 1994; Valla et al., in preparation).

Climate change affects the world's mountain regions and may jeopardize the services provided by mountains. This is especially important for mountain regions located within the boundaries of several countries, for example, Altai Mountains that stretch over Mongolia, Russia, China and Kazakhstan. The Altai mountains is a big region located in the Northern Asia, that is extended for more than 1200 km from the north to the south and can be divided into two areas: Russian Altai at north and Mongolian Altai at south. In this work we report the results of comparison of climate (both temperature and precipitation) trends in Russian and Mongolian Altai in 1979-2017. It was found that the mostly intensive warming is seen in Mongolian Altai (+0,25°С per decade), thus at south of Altai mountains; whereas in Russian Altai the warming was +0,19°С per decade. The trends of the last thirty years (1988-2017), show less significant warming, namely +0,15°С per decade in Mongolian and +0,11°С per decade in Russian Altai. Wherein during the cold seasons at north and south of Altai the negative temperature trends are observed, so in average the winters become colder; whereas during the warm seasons the trends are positive, so the summers become warmer. The precipitations in Mongolian Altai decrease on -2,8 mm yr^-1 per decade, whereas in Russian Altai increase slightly on +0,3 mm yr^-1 per decade. The analysis of the last thirty years shows that the similar trends are seen (-2,9 mm yr^-1 per decade) only at south of Altai mountains. The situation in Russian Altai is changed, thus the precipitation trend over 1989-2017 becomes negative (-2,6 mm yr^-1 per decade) and close to the values observed at north of Altai mountains. These abrupt climate changes related to the changes in atmospheric circulation and increase incidence of hazard in Altai Mountains.

We present a case study that uses climate information at various spatial and temporal resolutions for modelling water balance at the landscape level.

Landscape-scale water balance can be used to predict water availability to support forest vegetation and thus natural open azonal vegetation patches caused by insufficient water availability may be identified as distinct from open secondary vegetation that has resulted from post grazing abandonment in mountain areas. Additionally, the impacts of projected climate change may be for future habitat suitability for forest growth. One important factor for such work is the use of adequate spatio-temporal resolution of the climate data used.

We used climate information at various levels resolutions, ranging from modelled 5-km horizontal resolution of current and projected climate, WorldClim temperature and precipitation models at 2.5 min and 30’’ resolution and monthly temporal availability. We contrasted these with data derived from 10 meteorological stations within an area of ca. 100 km², - adjacent to the basin of interest - and adjusted terrain impacts, using a DEM at 30-m resolution to account for the differences resulting for water economy from clear-sky radiation. We applied these different data for calculating water balance to identify drought-sensitive areas in a hydrological basin which coincides with the limits of a conservation area in south-eastern Brazil.

We compare results for using the different sources of information on current and simulated future climate (temperature, precipitation), on the spatial pattern of precipitation and their impact on modelling potential water limitation to forest growth. We highlight the importance of using adequate climate data and auxiliary information for meaningful projections of climate change impacts on soil water availability and vegetation distribution.

We are developers and users of an integrated hydrological model (GEOtop), which has been applied to a variety of scientific problems, ranging from runoff estimation in small – medium catchments (< 1000 m²), studies related to the water-soil-vegetation interactions, snow cover dynamics in mountain areas, climate change impact assessment. We would like to discuss the needs in term of input data accuracy, when this kind of models are used for climate impact assessment studies in mountain regions. In particular, the need to have a temporally and spatially coherent representation of the daily cycle of all main surface meteorological variables (i.e. temperature, humidity, wind, solar radiation, precipitation). The difficulties in capturing local phenomena as clouds, shadows, thermal inversion.

We hope the workshop could be an opportunity to foster research in providing open, high quality climatic input dataset for impact studies in mountain regions.

GEOtop is an integrated hydrological model that simulates the heat and water budgets at and below the soil surface (Rigon, et al 2006; Endrizzi et al 2014). It describes the three-dimensional water flow in the soil and the energy exchange with the atmosphere, considering the radiative and turbulent fluxes over complex topography. Furthermore, it reproduces soil freezing and thawing processes, and it simulates the temporal evolution of snow cover, soil temperature and moisture. The model has been applied to a variety of scientific problems, ranging from estimation of runoff and water budget in small – medium catchments (< 1000 m²), studies related to the water-soil-vegetation interactions, snow cover in mountain areas, climate change impact assessment. The model has also been coupled with powerful meteorological preprocessing tools as MeteoIO (Bavay et al, 2014) or downscaling tools, optimized for complex topography (Fiddes et al 2012).

In this poster we present the approaches used by the model to have a temporally and spatially coherent representation of the daily cycle of all main surface input meteorological variables (i.e. temperature, humidity, wind, solar radiation precipitation). Moreover, we will discuss the challenges when the model used for climate impact assessment studies in small mountain catchments.

Most glacier models implement simple but efficient melt models which relate air temperature with melt: the so-called "temperature index models". To date, it has not been proven that physically more realistic models are more accurate, especially in view of the uncertainties in the forcing data (i.e. GCMs).

Although simple, temperature index models are non-linear and very sensitive to temperature variations above a pre-defined threshold. Therefore, not only bias correction but also variability has to be addressed by dynamical or statistical methods targetting the (long-term) forcing of glacier models.

The current generation of global glacier models all use very similar methods to bias-correct GCM data, and I see much room for improvement: both on the statistical and dynamical downscaling side.

Much of my research has been on the cryospheric and hydrological impacts of climate change in the Himalaya. The scale mismatch between climate models and impact models here is of course very large. One of my interests is in climate model evaluation, particularly looking at the implications of biases in regional circulation, near-surface climate and land-atmosphere interactions. I would be very interested in discussing how to make best use of climate model outputs in light of their biases and limitations. This could include questions on: (1) methods for model selection/rejection in impact studies; (2) the potential for further use of observational and emergent constraints in mountain regions to constrain uncertainty in climate projections; (3) robustness of change signals in different climate variables and at different scales. In addition, I would be interested to discuss methods for bridging the gap in scale between climate and impact models. Relevant areas that I am working on currently include statistical downscaling/disaggregation (including weather generators and scaling-based approaches) and application of an intermediate complexity atmospheric model (ICAR). Some possible topics to discuss could include: (1) mapping different applications to appropriate climate scenario/downscaling approaches; (2) different approaches to dealing with and expressing uncertainty, including how to establish useful narratives for different stakeholders; (3) the potential for increased use of hybrid downscaling methods, such as how high resolution dynamical downscaling could be used sparingly in combination with less computationally demanding methods.

Climate records, with twenty years of data or more, are virtually absent in high mountain
regions. While global atmospheric model-based products (e.g., reanalysis data) provide an
important complement to available observations by offering long-term data sets for the entire
globe, their spatial resolution is generally too coarse to correcly represent complex orography
and thus mountain weather.
This study presents the statistical downscaling method sDoG that combines the completeness
of reanalysis data with the accuracy of in situ observational records, to obtain longer-term yet
high-resolution mountain weather and climate data for the past. The ability of sDoG to extend
short-term daily air temperature records to a baseline climate period is evaluated at the example
of the Vernagtbachstation in the European Alps (2640 MSL). Namely, sDoG is trained using
observations from 2002 to 2012 and evaluated over the remaining 23 years from 2001 back to
1979. Uncertainty estimates provided by cross validation within the training period are then
compared to the true uncertainties found for the evaluation period.
The results show that daily air temperature as modeled by sDoG agrees very closely to the
observations throughout the validation period. The cross-validation based uncertainty estimates
hold for the later period (post 1990) and for the winter half year throughout the entire validation
period. sDoG outperforms available reference models (bias corrected ERA interim, regional
climate model output, bias corrected nearby observation time series), at all time scales
(climatological cycle, day-to-day, year-to-year). However, uncertainties of sDoG are identified
for summers in the earlier period (prior 1990). These uncertainties are presumably linked to
microclimatic changes (here, nearby glacier tongue replaced by rocky terrain). We discuss how
such microclimatic changes can represent important limitations of the stationarity assumption
underlying statistical downscaling methods.

Climate change information required for impact modeling is of much finer spatial and temporal
scale than the climate models can provide. The highest spatial resolution provided by climate
models for Europe is 12.5 km while impact models require 100 m or less. Downscaling is a method
used to fill this gap. Among various approches available for downscaling, stochastic weather generators
have been widely used for impact anaysis in various fields such as agriculture, hydrology,
economics etc. Besides their widespread use, their potential to simulate extreme climate events
is largely unexplored. Extreme events like heat waves, droughts and wildfires often occur from
processes involving more than one weather variables. These kinds of events are called multivariate
exremes. The aim of this study is to evaluate the performance of a Richardson type 6 variate (precipitation,
minimum temperature, maximum temperature, solar radiation, relative humidity and
windspeed) single-site weather generator (SiSi) to simulate multivariate extremes. We evaluate the
weather generator at various sites in a mountainous catchment in the Austrian Alps. In addition
to that, we also include sites from different parts of Europe having varying climates, topography
and proximity. Results show that SiSi is able to simulate multivariate extremes generally well at
all sites. Among all extreme events, the weather generator has a tendency to underestimate the
extremes related to minimum temperature. The performance of SiSi doesn’t depend on the climate
type of a region or the elevation of a location. We conclude that the weather generator needs to be
trained at individual location which may require making different adjustments for each variable.

We present overview and latest developments from the downscaling model TopoSCALE which can be used to derive point or grid model forcing timseries in mountain terrain based on 3D atmospheric fields from reanalysis, NWP or climate models. It performs 3D interpolations through coarse grid cells and pressure levels to obtain downscaled fields at each model timestep. Additionally it corrects radiative fluxes for high resolution surface geometry. We have recently added a data assimilation module to correct biases in forcing, largely originating from precipitation.

The strengths of this approach are tha it is fast (compared to dynamical downscaling) so can be applied at scale (large domain, long timeseries or ensembles) and does not rely on surface data, ecept global datasets (e.g. MODIS, SRTM). Thus it is a globally applicable method that is especially useful in data sparse regions.

TopoSCALE is now being used for the ESA CCI Permafrost. We present additional results from impact models run in High Mountain Asia.

In order to gain a better understanding of features, forcings and feedbacks related to temperature changes in mountainous regions, we analyze daily resolution alpine climate data from Switzerland covering an elevation range between 203 and 3580 m from 1981 to 2017 deriving highly resolved trends. The data set investigated includes daily resolution information on mean, maximum and minimum temperature, snow depth, cloud cover, global radiation, absolute air humidity, sunshine duration and weather type classifications. The two main research question we aim to address are: 1) Are temperatures in Switzerland changing at a different rate depending of elevation? and 2) What are underlying driving mechanisms of elevation dependent temperature changes? Our results indicate that temperature trends in Switzerland differ depending on the time of the year, day and elevation. Warming is strongest during spring and early summer with enhanced warming of daytime maximum temperatures. Elevation-dependent differences in temperature trends occur during autumn and winter with stronger warming at lower elevations. We attribute this elevation-dependend temperature signal mainly to elevation-based differences in trends of incoming solar radiation and elevation-sensitive responses to changes in frequencies of weather types. We assume solar radiation to change due to multiple mechanisms: circulation-driven changes in cloud coverage, direct aerosol effects enhanced at lower elevations due to higher initial aerosol concentrations and probably due to changes in fog and low stratus clouds. In general, effects of varying frequencies of weather types overlap with trends caused by transmission changes in short- and longwave radiation. Temperature signals arising from snow/ice albedo feedback mechanisms are probably small and hidden by other effects.

The Virtual Alpine Observatory (VAO) was initiated in 2012 by the German Environmental Research Station Schneefernerhaus (UFS) and has been operating as a network of nine European High Altitude Research Stations, based in the Alps and similar mountain ranges. VAO aims to address scientific questions related to alpine regions in great depth by means of its cross-border and interdisciplinary character.

One of the core elements of VAO is the Environmental Data Analysis Centre (AlpEnDAC) – a state-of-the-art IT-infrastructure which ensures that the measurements taken at the various sites areare easily and conveniently accessible.

Scientists using AlpEnDAC benefit from two main services:

- ‘Data.on-demand’: Uniform access to data irrespective of their actual location

- ‘Computing-on-demand’: AlpEnDAC also includes complex numeric computer models (e.g. trajectory and air quality models, meteorological models, etc.) hat can easily be used by scientists as required -- no special IT knowledge required! This service is in particular interesting for scientists not familiar with computational methods, who can run these numerical models through a simple web interface in the LRZ Compute Cloud by point and click.

Within the new project phase, that starts in 2019, a new automatic instrument control will be included that allows

- ‘Operating-on-Demand’: Advanced data-driven workflows, as a reaction on data events (e.g. concentration thresholds reached).

AlpEnDAC, as a collaborative effort of UAU, LRZ, DLR, UFS and bifa, develops services with an immediate value for research. It is funded by the Bavarian State Ministry of the Environment and Consumer Protection.