Regional water cycle health evaluation and obstacle factor analysis: A case study of the Yellow River basin in Henan Province, China

• A DPSIR framework model was proposed, covering 21 multi-scale indicators. • RF was introduced as a water cycle health evaluation method. • The obstacle factors affecting the water cycle health were identified. • The main obstacle factors were categorized and analyzed. • Some policy implications were provided for the development of the water cycle. Ecological and socioeconomic systems are interconnected through water cycles. Due to intensive anthropogenic activities and global climate change, the water cycle has undergone profound changes over the last few decades. This study aimed to evaluate the health status of the regional water cycle, explore the etiology of water cycle disorders, and provide suggestions for improving the health of the water cycle. Based on the Driving force-Pressure-State-Impact-Response (DPSIR) model, an index system was constructed to evaluate regional water cycle health. The water cycle was evaluated using a random forest (RF) model, which can achieve desirable fitting and generalization properties. The Yellow River Basin in Henan Province (YRBHN), China, was used as an example to demonstrate the application of this system. An Extended Fourier Amplitude Sensitivity Test (EFAST)-Cloud model was used to validate the evaluation results of the RF model. The main influencing factors in the water cycle process were then identified using the obstacle degree model, and ensuing policy recommendations were presented. The results demonstrated that the health state of the water cycle in the YRBHN shifted from unhealthy to sub-healthy during 2011–2020, with the response layer showing the most evident positive change. The health states of Zhengzhou and Luoyang were superior to those of the other cities, with average health index numbers of 3.1006 and 3.0078, respectively. Although the evaluation index numbers of the EFAST-Cloud and RF models were slightly different, their evaluation results were similar, indicating that the RF model was applicable for water cycle health evaluation. The proportion of ecological water use and the response (R) layer provided the greatest obstacle degree in the index and criterion layers, respectively. Based on the current state of the water cycle in the YRBHN, some policy implications were proposed. This study proposed a new method for regional water cycle health evaluation and decision making that provides theoretical support for regional water resource management and is of considerable practical significance.