Interacting Effects of Precipitation and Potential Evapotranspiration Biases on Hydrological Modeling

The quality of precipitation (P) and potential evapotranspiration (PET) data greatly affects the hydrological modeling performance. Considerable attention has been paid to identifying the influence of biased P or PET inputs independently. However, few studies have explored the joint interaction of biases in P and PET inputs on hydrological simulations. Here, we investigate the mutual compensation of P and PET biases on the performance of two widely used conceptual hydrological models, the Xinanjiang model and the Probability Distributed Model. P and PET from HYREX (HYdrological Radar EXperiment) and CAMELS-GB (Catchment Attributes and Meteorology for Large-sample Studies in Great Britain) data sets are collected over five catchments with varying characteristics in Great Britain. Different biases are added to these original time series to generate 6560 biased input scenarios. The results suggest that there is a certain compensational relationship between the biases in P and PET inputs to reproduce desirable streamflow simulations. A new hydrological proxy named Compensational Interaction Angle (CIA) is identified and found to be stationary with various modeling periods, as well as stable with different hydrological models despite model equifinality. Further, the CIA highly relates to the long-term climate aridity ratio. The catchments with greater aridity have larger CIAs. This study offers a fresh perspective to analyze the input errors in hydrological modeling. The results can help to better understand P and PET interactions in hydrological modeling, and guide the selection/evaluation/bias-correction of P and PET data sets for hydrological applications.Plain Language Summary Hydrological modeling helps us understand, predict and manage water resources in the real world. Precipitation (P) and potential evapotranspiration (PET) are two key inputs. Hence their accuracy has a great impact on the modeling outputs. However, the measurements and estimations of P and PET are prone to different sources of errors. We explore the joint interaction of their accuracy on hydrological modeling by considering there are mutual compensational effects between them. It is found that there is a stable compensational relationship between their biases in producing desirable hydrological performance. Moreover, the compensation relationship is highly linked to the climate aridity condition of the region. This study provides a new perspective for hydrologists to explore the sensitivity of input errors in hydrological modeling. We strongly encourage the community to apply the proposed method in other regions to further explore and identify the compensational relationship between P and PET by using different hydrological models. This has the potential to enhance our understanding of the interactions between P and PET in hydrological modeling.