TU Berlin

Climatic and tectonic natural hazards in Central Asia


High and Central Asia is one of the most tectonically active regions on earth, influenced by two major climate systems: the monsoon and the westerly winds. Large landslides are one of the greatest hazards in Central Asia and pose a threat to settlements, human life, and infrastructure.

Climatic factors, such as extreme precipitation and rapid snowmelt, can cause landslides. Our principal goal is to understand the climatic triggers and the triggering mechanisms of landslide hazards in Central Asia. For this purpose, detailed atmospheric data with a high spatial and temporal resolution is a key requirement. Regional climate models (RCMs) can produce meteorological fields with adequate spatiotemporal resolution using dynamical downscaling methods. A new version of the High Asia Refined Analysis (HAR v2) was developed during the project.

Climatic factors such as extreme precipitation and rapid snowmelt can cause landslides. For our studies, which focus on Central Asia, detailed climate data with high spatial and temporal resolution are a key requirement. Regional climate models (RCMs) can produce meteorological fields with adequate spatiotemporal resolution using numerical downscaling methods.

A new version of the High Asia Refined Analysis ( HAR v2 ) is under development using the Weather Research and Forecast Model (WRF). Compared to the older version , HAR v2 covers a larger area of Central Asia, a longer time period, and includes new 2 km areas.

Fig. 1: WRF domain setup for HAR v2.


The HAR v2 dataset is generated by dynamical downscaling using the Weather Research and Forecasting model (WRF) version 4.1. ERA5 reanalysis data provided by ECMWF is used as forcing data. In addition, snow depth from the Japanese 55-year Reanalysis is applied to correct snow depth initialized from ERA5, since ERA5 largely overestimates snow depth over the Tibetan Plateau ( Orsolini et al., 2019 ).

The domain setup (Fig.1) consists of two-way nested domains with 30 km and 10 km grid spacing. Note that, HAR v2 only provides results from the 10 km domain. The forcing strategy is daily re-initialization adopted from the HAR ( Maussion et al., 2011 , 2014 ). Each run starts at 12:00 UTC and contains 36 h, with the first 12 h as spin-up time. This strategy avoids the model from deviating too far from the forcing data and provides computational flexibility since daily runs are totally independent of each other and can be computed in parallel and in any sequence.

Same as HAR, the output of the model is post-processed into product-files: one single file per variable and per year at various aggregation levels. The data is currently available from 2004 to 2018.

More information and download link of the HAR v2 can be found here (https://www.klima.tu-berlin.de/HARv2).


We combined the HAR v2 data with historical landslide inventories from the Global Landslide Catalog (GLC) (Kirschbaum et al., 2010) and the Global Fatal Landslide Database (GFLD) (Froude and Petley, 2018) to analyze the atmospheric conditions that initialized landslides in Kyrgyzstan and Tajikistan. Objective thresholds for rainfall, snowmelt, as well as the sum of rainfall and snowmelt (rainfall+snowmelt) were defined, such that these thresholds can best separate weather events that triggered landslides from those that did not result in any slope failure. Mean intensity (Imean), maximum intensity (Imax), and the accumulated amount (Q) of weather events were used as predictors. The results show that thresholds defined by rainfall+snowmelt have the best predictive performance (Imean=5.05 mm d-1; Imax=14.05 mm d-1; and Q=15.56 mm). Fig. 2 presents the mean annual exceedance maps derived from these thresholds. The mean annual exceedance is defined as the number of events per year that exceed the defined thresholds. These maps depict the climatic disposition of landslide hazards in Kyrgyzstan and Tajikistan and have added value in landslide susceptibility mapping, since they are derived from weather-scale triggering conditions, and therefore, contain information on extreme processes.

Fig. 2: Mean annual exceedance (number of events per year) of (a) Imean=5.05 mm d-1; (b) Imax=14.05 mm d-1; and (c) Q=15.56 mm for rainfall+snowmelt. Black circles: landslide events from Global Landslide Catalog (GLC) and Global Fatal Landslide Database (GFLD).


  • Wang, X., Tolksdorf, V., Otto, M., & Scherer, D. (2021). WRF‐based dynamical downscaling of ERA5 reanalysis data for High Mountain Asia: Towards a new version of the High Asia Refined analysis. International Journal of Climatology, 41(1), 743-762.
  • Wang, X., Otto, M., & Scherer, D. (2021). Atmospheric triggering conditions and climatic disposition of landslides in Kyrgyzstan and Tajikistan at the beginning of the 21st century. Natural Hazards and Earth System Sciences Discussions, 1-24.


  • Maussion, F., Scherer, D., Finkelnburg, R., Richters, J., Yang, W., & Yao, T. (2011). WRF simulation of a precipitation event over the Tibetan Plateau, China–an assessment using remote sensing and ground observations. Hydrology and Earth System Sciences, 15(6), 1795-1817.
  • Maussion, F., Scherer, D., Mölg, T., Collier, E., Curio, J., & Finkelnburg, R. (2014). Precipitation seasonality and variability over the Tibetan Plateau as resolved by the High Asia Reanalysis. Journal of Climate, 27(5), 1910-1927.
  • Orsolini, Y., Wegmann, M., Dutra, E., Liu, B., Balsamo, G., Yang, K., ... & Arduini, G. (2019). Evaluation of snow depth and snow cover over the Tibetan Plateau in global reanalyses using in situ and satellite remote sensing observations. The Cryosphere, 13(8), 2221-2239.
  • Kirschbaum, D. B., Adler, R., Hong, Y., Hill, S., & Lerner-Lam, A. (2010). A global landslide catalog for hazard applications: method, results, and limitations. Natural Hazards, 52(3), 561-575
  • Froude, M. J., & Petley, D. N. (2018). Global fatal landslide occurrence from 2004 to 2016. Natural Hazards and Earth System Sciences, 18(8), 2161-2181.