Performance Evaluation of Global Hydrological Models for climate change projections in Pan-Arctic river basins

The Pan-Arctic region has become highly susceptible to the increased risks of climate warming. Climate warming has substantial implications in many biophysical states and processes that are strongly influenced by the threshold and phase change of the freezing point. While it may be challenging to anticipate the changes brought on by a changing climate, the potential effects on hydrological processes are profound and manifold. However, Global Hydrological Models (GHMs) make it possible to simulate the terrestrial water cycle on a global scale, so they are used in this study to analyze model simulation results of land surface hydrologic dynamics processes of six Pan-Arctic river basins. Here, GHMs simulated cold region processes, including river discharge and snow water equivalent. Here simulations from nine GHMs participating in the second phase of the Inter-Sectoral Impact Model Inter-comparison Project (ISIMIP2a) were evaluated. The simulated discharges by each individual hydrological model were compared against the observed discharges for the period 1971-2000. In addition to a visual comparison, three efficiency criteria-NSE, PBIAS, and Bias in Standard Deviation were used to validate the model simulations for monthly hydrographs, seasonal dynamics, high flows, and low flows. Furthermore, the models were evaluated and rated on their performance using the threshold values assigned for each efficiency criteria, and then aggregated indices were estimated for each model and basin using rating scores of 1 (good performance), 0.5 (weak) and 0 (poor) for every criterion and gauging stations. The study revealed large uncertainty levels in simulated river discharges with uncalibrated models showing considerable bias when compared against observed discharge. Therefore, significant differences in model performance were identified. WaterGAP2, MATSIRO, and MPI-HM models performed better than the rest of the models. These models had a better ability to replicate observed discharge than the rest. Four to five models out of nine showed acceptable performance for high and low flows (aggregated index > 60%) in the Ob basin, in other basins their performance was much weaker. The models were able to reproduce seasonality of snow cover, but the bias was quite high in many cases (not estimated numerically). The large uncertainty and bias observed in the simulated values highlight the urgent need for model improvements of cold region hydrological processes in Global Hydrological Models.