Salinity risk assessment using fuzzy multiple criteria evaluation

Abstract

AbstractFor agricultural productivity, primary salinity constitutes a hazard. Understanding the risk of secondary salinization is also important and it is necessary to map all these hazards and the overall risk. One approach is digital soil mapping. We have developed hazard maps of root zone (0–2 m) and subsurface (6–12 m) salinity, as well as groundwater height, using empirical best linear unbiased prediction using proximal and remotely sensed ancillary data. To convert the hazards into salinity risk maps and account for some uncertainty, fuzzy membership functions (FMF) were used in conjunction with semantic import models that consider the vulnerability of agricultural crops. Therefore, we developed memberships (μ) to three salinity risk classes and created maps of each, including root zone (μRSR) and subsurface (μSSR) salinity, and depth to groundwater (μGTR). To estimate groundwater (μGSR) and overall salinity risk (μOSR), fuzzy multiple criteria evaluation was undertaken. To account for μGSR, the μSSR and μGTR were combined using a t‐norm operator, and μOSR was accounted for by combining μRSR and μGSR using a t‐conorm operator. The approach was applied to an area growing irrigated cotton. Cross‐validation shows predicted μOSR was good, with an agreement rate of 70.6%. The predicted μOSR was consistent with that of many areas affected by point‐source salinity. The method provides an effective way of mapping salinity hazard and risk. Incorporation of FMF provides a meaningful continuum of membership and allows the incorporation of uncertainty. Salinity risk can be calculated relative to any vegetation community and locations where management actions to moderate soil salinization can be identified.