Facilitating sensitivity analysis of hydrological models through knowledge-driven configuration and distributed online model services

Abstract

Hydrological models (HMs) are essential for understanding the complexities of the water cycle and runoff dynamics. Sensitivity analysis (SA), an essential component of HMs, plays a key role in identifying the parameters that have the greatest impact on model outcomes. It helps to simplify the complexity of hydrological systems and provides a comprehensive understanding of the underlying physical processes. However, the complexity of HMs and the diversity of SA methods pose significant challenges for researchers, making the SA configuration process intricate and requiring substantial computational resources. To address these challenges, we propose a comprehensive strategy that integrates knowledge-driven configuration services with distributed online model services. First, we establish a rule-based knowledge repository and a case-based knowledge repository. These repositories provide general configuration guidance and similar SA case recommendations, respectively, to support decision-making in critical SA steps. This ensures that the configuration of SA is accurate and reliable. Secondly, we encapsulate HMs as web services and leverage distributed computing resources to optimize execution efficiency. Then, we integrate the HM services with the SA modules to achieve a complete SA experiment. Based on this strategy, we finally developed a prototype system that offers a user-friendly tool for conducting SA with enhanced computational performance and streamlined workflow. The watershed-scale HM, SWAT, was used to test the effectiveness of the prototype system. The results demonstrate that this strategy enables more comprehensive analysis and improves decision-making through configuration guidance, and holds promise for enhancing the reliability and efficiency of SA in hydrological modeling.