Influence of Data Length on the Determination of Data Adjustment Parameters in Conceptual Hydrological Modeling: A Case Study Using the Xinanjiang Model

Abstract

Minimum data length is vital to guarantee accuracy in hydrological analysis. In practice, it is sometimes determined by the experiences of hydrologists, leading the selection of the acceptable minimum data length to an arguable issue among hydrologists. Therefore, this study aims to investigate the impact of data length on parameter estimation and hydrological model performance, especially for data-scarce regions. Using four primary datasets from river basins in Japan and USA, subsets were generated from a 28-year dataset and used to estimate data adjustment parameters based on the aridity index approach to improve the parameter estimation. The influence of their length on hydrological analysis is evaluated using the Xinanjiang (XAJ) model; also, the effectiveness of outlier removal on the parameter estimation is checked using regression analysis. Here, we present the estimation of the most acceptable minimum data length in parameter estimation for assessing the XAJ model and the effectiveness of parameter adjustment by removing the outliers in observed datasets. The results show that between 10-year to 13-year datasets are generally sufficient for the robust estimate of the most acceptable minimum data length in the XAJ model. Moreover, removing outliers can improve parameter estimation in all study basins.