Auto-calibration of a land surface and routing model over the Narmada and Yamuna River Basins

Floods are the most devastating natural occurring disaster in consideration of loss of lives as well as loss of economy. India often faces floods every year in some areas. The forecast model plays a critical role in flood forecasting to mitigate the loss of lives and economic losses. Large-scale hydrological models can be used to forecast floods. However, large-scale hydrological models perform better on a global scale. While coming to the local scale, their performance is limited compared to locally calibrated models. The model parameters derived from global datasets do not accurately represent the regional characteristics of a catchment and land surface conditions, and it isn't easy to measure the model parameters for the catchment. Using these models require that model parameters be calibrated (adjusted) so that model predictions closely replicate the observations. Here, we calibrate the LIS framework with the Noah-MP model coupled with the HYMAP2 over the Narmada and Yamuna River basin. We are using a Dynamically dimensioned search (DDS) algorithm for auto-calibration. DDS is an efficient tool to calibrate complex and large-scale hydrological models as it automatically reduces the number of evaluations of models to find the optimal solution. Multiple meteorological forcings are used to run the model in the LIS framework, and streamflow from the central water commission (CWC) is used as observed data.