Understanding the hydrological responses using multiple conceptual models in the Upper Indus basin

Código

I-EBHE0073

Autores

Malla Mani Kanta, Dhyan Singh Arya

Tema

WG 1.05: Comparative understanding of runoff generation processes

Resumo

The Upper Indus Basin (UIB), an important watershed in the Hindu Kush Himalayan region, ranks second worldwide in terms of its population exposed to flood risks. This basin's streamflow heavily relies on snow and glacier meltwater, especially in mountainous catchments with extensive glacier coverage. However, limited data on meteorological inputs, observed streamflow, land cover, and soil properties complicate hydrological modellingg in this basin. These challenges pose difficulties in the calibration and validation of models, introducing uncertainties that must be addressed to enhance the reliability of model predictions. Despite these challenges, conceptual models have seen exponential growth due to their cost-effectiveness in computational requirements, maintaining the accuracy of simulated streamflow with limited data needs. Conceptual models address hydrological processes in a catchment using a set of non-linear storage reservoirs, hence they are often referred to as grey-box models. In this study, four conceptual hydrological models?GR4J, HBV, HYMOD, and SIMHYD?were selected to simulate river discharge in the Upper Indus Basin over a study period from 1979 to 2019. These models were chosen for their extensive usage in various applications and their distinct characteristics in simulating rainfall-runoff processes. Due to the unavailability of observed data, global gridded precipitatio, and discharge data from ERA5 were used for model setup at the catchment outlet. Additionally, Degree-Day Factor-based snow and glacier modules were incorporated into the model structure to measure melt contributions. The calibrated models exhibited very good performance (NSE > 0.75) during both calibration and validation periods, with GR4J emerging as the top-performing model with an NSE >= 0.81. The study highlighted the dominance of glacier melt during the summer monsoon (JJAS), which drives peak discharge, followed by contributions from rain+baseflow and snowmelt. Sub-basin analysis revealed spatial variability in melt contributions, attributed to the employed model structures and soil moisture accounting routines. For example, during the summer monsoon, snowmelt predominated in the Astore, Shingo, Hunza, and Shyok sub-basins, while glacier melt was more significant in the Gilgit, Hunza, and Shyok sub-basins. Seasonal analysis emphasized the significant influence of model choice on predictions, as models exhibited disparities in both the timing and magnitude of melt contributions. The percentage contributions of glaciers and snowmelt varied among models. For instance, GR4J showed distinct patterns, such as peak glacier melt in August and peak snowmelt in June. Conversely, the HBV model exhibited a different pattern, with the highest glacier melt contribution in September and peak snowmelt in July. The superior performance of GR4J is evident from the computed diagnostic performance evaluation measures, showcasing its ability to effectively represent all flow segments and outperforming other models during calibration and validation periods. This enhanced performance is attributed to the parsimonious structure of GR4J, characterized by just four parameters, which facilitates improved calibration through reduced parameter dimensionality in a semi-distributed configuration. The findings of this study offer a critical understanding of the hydrological model structural uncertainty in generating the runoff.