Predicting hydrology in ungauged basins using the DDD model

Abstract

Streamflow data from gauged catchments plays an important role for water resources management applications such as water resource planning, flood risk management and assessment of the impact of environmental- and climate change. A hydrological model successful at predicting in ungauged basins is needed for hydrological estimation for the million basins around the globe that are ungauged and has a great potential for better predicting the hydrological consequences of climate change. This study aimed at evaluating the DDD model and its performance on predictions in ungauged basins, as well as comparing and evaluating different regionalization methods on catchments in Norway. Regionalization is defined as methods that allow for the transfer of hydrological information from gauged catchments to ungauged catchments. The comparison of methods was done at two levels, as a whole over Norway, as well as regionally for catchments in specific regions over Norway. This study shows that the new and improved DDD model is good at predicting hydrology in ungauged basins, with average Kling-Gupta efficiency values ranging from 0.7 up to 0.77 for the different regionalization methods. The different regionalization methods perform satisfactorily, with good KGE scores. The best regionalization method to use was the multiple regression method, in which the average KGE value were 0.77, compared to 0.72 and 0.7 for the output average and parameter average, respectively.