Open access publications in the SLU publication database

Abstract

Estimating dynamic storage as a metric can be used to make an overall assessment of catchment resilience to extreme weather events such as droughts and floods. Because of the complexity of direct empirical measurements, bucket-type hydrological models can be a suitable tool to simulate the catchment storage across a broad range of scales as they require minimal input data. However, these models consist of one or more conceptual structures based on several linear or nonlinear reservoirs and connections between these reservoirs. Therefore, choosing the most appropriate model structure to represent storage-discharge functioning in catchments is difficult. To bridge this gap, this study evaluated the performance of three different HBV model structures on 14 heterogeneous boreal catchments classified into four distinct catchment categories. The results showed that the three-bucket structure performed better in larger catchments with deeper sediment soils. In contrast, a single reservoir structure is sufficient to predict the storage-discharge behavior for a lake-influenced catchment with lower elevation above the stream network. Moreover, our results indicate that while the estimates of mean catchment storage varied between the different model structures, the ranking between the catchments largely agreed for the different structures. Hence, our results suggest that instead of a single model structure, using an ensemble averaging approach would not only better address the structural uncertainty but also facilitate further storage comparison between different catchments. Finally, based on Spearman rank correlation results, we found that catchment size and sediment soil were positively correlated with dynamic storage estimation.