| dc.description.abstract | Sufficient and Accurate runoff time series are vital information which are required in many watershed development projects. Unfortunately, most of the catchments in the world are ungauged, and this issue is more pronounced in small size catchments (Area <50 km2). This study compares between the standard Scaling method, which is widely used by hydropower developers in Norway, and three regionalization methods namely; Regression, Physical Similarity (Single Donor) and Physical Similarity (Pooling Group), in Predicting daily flows for Ungauged Basins (PUB). Climate, Runoff and Watershed Characteristics data of 101 small unregulated gauged catchments around Norway are collected. From which, 15 catchments are randomly selected to validate the PUB approaches. Two conceptual hydrological models are used; HBV (an overparameterized model) and DDD (a parsimonious model). In general, all regionalization methods scored significantly better results than the standard scaling, the regression method was better than the physical similarity methods, and the HBV model outperformed the DDD model. The best PUB method was Regression by HBV model which gave satisfactory results (KGE>0.5) in all 15 test catchments in contrast to the Scaling method with satisfactory results in only 7 catchments. In Physical Similarity, Pooling Group method is significantly better than Single Donor, and Pooling model outputs yielded better results than Pooling model parameters. Based on simulating runoff signatures, the pooling group by HBV model yielded the best performances in simulating operational flows (10%probability of exceedance) and the pooling group by DDD gave the most accurate estimation of low flows (70% | |
