A flexible framework for regionalization of base flow for river habit maintenance and its thresholds

Information on base flow for river habit maintenance (BF_RH) and its thresholds is necessary for water resource utilization and protection. BF_RH and its thresholds have significant spatial differences; however, it is still unclear how to identify and assess these characteristics. In this study, a technical framework was proposed to clarify the specific procedures and methods for regionalization of BF_RH and its thresholds in large-scale areas. The framework includes four parts: construction of controlling factor system, sub-region division, identification of dominant factors, and determination of the thresholds in sub-regions. The framework was then applied to China to analyzed the regionalized characteristics of BF_RH and its thresholds from a national perspective. The results illustrate the following: (1) the country is divided into nine sub-regions, and the controlling factors and their action paths to BF_RH vary greatly. The elements of climate, vegetation, soil, topography and morphology are satisfactory in explaining the variance of BF_RH and its thresholds, as R² of the partial least squares structural equation modeling is between 0.503 and 0.848. (2) The value of BF_RH/MAF (i.e. the proportion of BF_RH to mean annual natural flow) differs greatly among sub-regions. The mean value is the largest in the Northwest Region, reaching 20 %, while it is only 1.7 % in the Northeast Cold Region. (3) The dynamic and static thresholds are obtained by using the precipitation and other indices as the explanatory variables in the sub-regions. In general, the more abundant the water resources, the higher may be the threshold. Moreover, attention should be paid to the positive and negative effects of vegetation restoration on this threshold. The case study proves that the framework can guide the determination of BF_RH, especially for ungagged rivers. Importantly, the framework is flexible and highly adaptable in different regions.