Enhanced Runoff Modeling by Incorporating Information from the GR4J Hydrological Model and Multiple Remotely Sensed Precipitation Datasets

Abstract

Reliable runoff modeling is essential for water resource allocation and management. However, a key uncertainty source is that the true precipitation field is difficult to measure, making reliable runoff modeling still challenging. To account for this uncertainty, this study developed a two-step approach combining ensemble average and cumulative distribution correction (i.e., EC) to incorporate information from the GR4J (modèle du Génie Rural à 4 paramètres Journalier) hydrological model and multiple remotely sensed precipitation datasets. In the EC approach, firstly, the ensemble average is applied to construct transitional fluxes using the reproduced runoff information, which is yielded by applying various remotely sensed precipitation datasets to drive the GR4J model. Subsequently, the cumulative distribution correction is applied to enhance the transitional fluxes to model runoff. In our experiments, the effectiveness of the EC approach was investigated by runoff modeling to incorporate information from the GR4J model and six precipitation datasets in the Pingtang Watershed (PW; Southwest China), and the single precipitation dataset-based approaches and the ensemble average were used as benchmarks. The results show that the EC method performed better than the benchmarks and had a satisfactory performance with Nash–Sutcliffe values of 0.68 during calibration and validation. Meanwhile, the EC method exhibited a more stable performance than the ensemble averaging method under different incorporation scenarios. However, the single precipitation dataset-based approaches tended to underestimate runoff (regression coefficients < 1), and there were similar errors between the calibration and validation stages. To further illustrate the effectiveness of the EC model, five watersheds (including the PW) of different hydrometeorological features were used to test the EC model and its benchmarks. The results show that both the EC model and the ensemble averaging had good transferability, but the EC model had better performance across all the test watersheds. Conversely, the single precipitation dataset-based approaches exhibited significant regional variations and, therefore, had low transferability. The current study concludes that the EC approach can be a robust alternative to model runoff and highlights the value of the incorporation of multiple precipitation datasets in runoff modeling.