Our study focuses on the Naryn river catchment (57,833 km²) which originates in the Tien Shan mountains in Kyrgyzstan and flows west through the Ferghana Valley into Uzbekistan. There the Naryn river merges with the Kara Darya River to form the Syr Darya. Streamflow at gauging stations in the Naryn catchment have a natural signal, making this a suitable location to evaluate our model.

We implemented a degree day snow and glacier melt model to the Dynamic fluxEs and ConnectIvity for Predictions of HydRology (DECIPHeR) model. DECIPHeR uses hydrological response units to categorise the behaviour of water flow in different parts of the catchment. This means that DECIPHeR can be applied across very large catchments, yet is computationally efficient enough to include model uncertainty in streamflow predictions.  Including uncertainty in predictions is important because observations in mountainous regions can be very sparse.

We use the model to predict past changes in streamflow at six gauging stations in the catchment. Initial glacier thicknesses are inferred using the GlabTop2 method and Landsat observations of glacier outlines from the 1970s. The model can predict the increase in discharge observed during the spring associated with the onset of snow melting and peak discharge observed in the summer coinciding with glacier melting. Seasonal variations in the spatial distribution of snow extent agree well with MODIS snow cover observations.

The next step in the project is to apply the model to the future. To predict future changes in streamflow, we will use an ensemble of CORDEX regional climate model projections for Central Asia. We have bias corrected this high-resolution climate data (~25km gridded) using a distribution mapping technique with observations of precipitation and temperature. Any future changes in the timing and quantity of the snow and glacier melt contributions to streamflow will have implications for water supply in the Ferghana valley and downstream in Syr Darya.