Towards developing a streamflow forecasting system for data-poor mountainous watershed: an approach using parameter transfer

Studying the hydrological responses of the Indian Himalayan Region (IHR) is crucial given the rise in the frequency of floods and other natural disasters. The hydrological processes in this area are more complicated due to the extreme weather pattern and varied topography. Streamflow forecasting is made more difficult by the extremely low number of stream gauge stations and the absence of accurate stream flow data. The problem of lack of observational data in ungauged watersheds can be resolved by transferring model parameters from similar gauged basins (Regionalisation). According to the traditional regionalization procedures using rainfall-runoff models, donor and recipient catchments must be similar in a variety of ways, including slope, size, drainage pattern, area, etc. It is extremely difficult to locate a catchment with all those similarities. In this study, we use a fully distributed hydrological model WATFLOOD for developing a streamflow forecast of the Alakananda River basin where the stream flow observation is very limited for the calibration of the hydrological model. WATFLOOD is working based on Grouped Response Unit (GRU). The requirement that has to be satisfied for regionalization using the WATFLOOD model is that land cover classes of the ungauged watershed should be represented in the gauged watershed irrespective of their spatial distribution. Also, there should be as many as possible gauged sub-watersheds that represent each land cover class. We identified a similar watershed that has similar land cover classes and sufficient stream flow gauges to represent each of the land cover classes. The three-step calibration process of the WATFLOOD model for both river basins is carried out to transfer the parameters. The results of ongoing work will be presented at the conference.