A Model Setup for Mapping Snow Conditions in High-Mountain Himalaya
New Publication
article written by MRI
13.06.19 | 05:06

This paper describes and evaluates a snow mapping setup for the remote Langtang Valley in the Nepal Himalayas, which can deliver data for snow and water availability mapping all year round.

Seasonal snow cover is an important source of melt water for irrigation and hydropower production in many regions of the world, but can also be a cause of disasters, such as avalanches and floods. In the remote Himalayan environment there is a great demand for up-to-date information on the snow conditions for the purposes of planned hydropower development and disaster risk reduction initiatives. 

 Due to the importance of snow to society, many countries run an operational snow mapping service to provide updated information about snow conditions (Saloranta, 2016). The information derived from operational snow mapping is valuable for planning hydropower production and water resources management, for natural hazard forecasting (flood, avalanche), and for informing the public and tourists about trekking or skiing conditions in the mountains. In the Himalayas, such near real-time information about snow conditions is very limited at present, with efforts mostly focused on cloud-free satellite images showing the extent of the snow-covered area (SCA) (Immerzeel et al., 2009Gurung et al., 2017Huang et al., 2017). However, these satellite-based maps of SCA do not provide any direct information on snow depth and snow water equivalent (SWE), which is needed in hydropower applications and in flood or avalanche hazard forecasting. In light of the planned hydropower development initiatives (e.g., Alam et al., 2017) and of the recent snow-related disasters, there is a great demand for up-to-date information on the snow conditions in the remote Himalayan environments. Consequently, the main research question has been: how to enhance operational snow and water availability mapping in remote high-mountain areas, such as the Nepal Himalayas?

The study continues to explore and improve the mapping and simulation of the snow cover and snow melt rates in the Langtang catchment. The novel model features include: (i) estimation of snow melt rate parameters from dedicated snow observations using automatic SWE measurements (section Model Setup and Parameter Estimation), (ii) improved parameterization of melt water refreezing in the snowpack (section Model Description), (iii) inclusion of estimated sublimation/evaporation rates (section Model Setup and Parameter Estimation), (iv) estimation of more accurate snow precipitation rates for model forcing using a passive SWE sensor (Kirkham et al., submitted; section Liquid and Solid Precipitation), and (v) estimation of monthly precipitation distribution in Langtang Valley on the basis of open access global historic precipitation dataset (Beck et al., 2017a,b section Liquid and Solid Precipitation). The main focus in this paper is on seasonal snow but liquid precipitation estimates are also included. Moreover, in order to estimate how applicable the simulated estimates of snow conditions in the Langtang catchment are in comparison with the neighboring regions, we assess the spatial correlation of snow line elevation (SLE) in the part of Himalayas bordering to Nepal using SCA data derived from MODIS (Moderate Resolution Imaging and Spectroradiometer) satellite images in the period 2001–2017 (section MODIS Snow-Covered Area).

When searching for suitable models to be used in e.g., water management-related issues, there are many application-specific considerations to make (e.g., Saloranta et al., 2003). Basically, there is no standard universally “best” model, but whether a model is appropriate or not depends on the purpose it is used for. The setup for snow mapping application for remote high-mountain areas, described in this paper, features several elements which should promote and lower the threshold of model use in practical applications. These features include: near real-time data delivery, almost maintenance-free measurement station setup (minimum of two temperature sensors, tipping bucket and CS725), simplified modeling approach, precipitation distribution estimated from easily available climatology. Such “live” estimation of the seasonal snow cover, rainfall and snowmelt rates allows us to provide effectively up-to-date information for predicting hydropower production potential and possible flood risk, for identifying areas of avalanche risk and for forecasting seasonal meltwater supply patterns to people in high-mountain regions, where all of these things are generally poorly known.

The paper describes a setup for simplified operational monitoring and modeling of seasonal rainfall and snow distribution for remote high-mountain areas, which can deliver data for snow and water availability mapping all year round. The setup utilizes (1) robust and almost maintenance-free in-situ instrumentation with satellite transmission, (2) a freely available numerical snow model, and (3) estimation of model key parameters from local meteorological and snow observations as well as from freely available climatological data. These features should promote and lower the threshold of model use in practical applications.

The snow model, not specifically calibrated in our application, produces results which are in reasonable agreement with observed snow depth, SCA and SLE time-series in the Langtang catchment. The model results show slightly less snow than indicated by the satellite-based MODIS SCA-images (bias of −7 percentage points in SCA and 110 m in SLE). The RMSD variability measure between the simulated and observed (MODIS) snow cover is 16 percentage points for SCA and 670 m for SLE. As many of the high-mountain regions in central and eastern Nepal show high correlation (>0.8) with the estimated SLE in the Langtang catchment, the results may provide a first-order approximation of the snow conditions for these areas too.

The estimation of melt parameters and solid precipitation from passive gamma-radiation based SWE-sensor data, as well as the improved process-based parameterization and locally verified estimation of the significant processes of refreezing within, and sublimation/evaporation from the snow pack, are features which to our knowledge have not been previously applied in glacio-hydrological catchment models in the Himalayan region. The simulation results suggest that most of the snow melt water comes from rather recently settled snow, and only about 30% of the snow melt water originates from snow older than 30 days. The ratio of snow melt water refreezing to total snow melt is 36 and 48% during the all-year and the drier non-monsoon periods, respectively. A sub-daily model time step (3 h in our case) is essential to properly capture the diurnal melt-refreeze cycles. The estimated accumulated sublimation/evaporation loss from the snow cover is lower in the all-year than in the non-monsoon period (57 vs. 69 mm yr−1).

The simplified snow mapping approach should be able to provide useful and up-to-date information on snow cover, snow depth and water equivalent, as well as on weather conditions fit for the purposes and needs of e.g., hydropower companies, local authorities and other practical applications in remote high-mountain areas.


 Saloranta Tuomo, Thapa Amrit, Kirkham James D., Koch Inka, Melvold Kjetil, Stigter Emmy, Litt Maxime, Møen Knut. ‘A Model Setup for Mapping Snow Conditions in High-Mountain Himalaya’, Frontiers in Earth Science (2019)   https://www.frontiersin.org/article/10.3389/feart.2019.00129