Mapping Soil Salinity Risk Using Multivariate Geostatistics.
Annamaria Castrignan˛1, Nicola Lopez1, Stefania Fanni2, and Rita Puddu2. (1) CRA - Agronomic Research Institute, Via Celso Ulpiani, 5, Bari, Italy, (2) CRAS-ERA Sardegna, Viale Trieste, 111, Cagliari, Italy
Soil salinization is related to both natural factors and changes in land use including agricultural intensification. The costal plain of Muravera-Villaputzu (southeastern Sardinia, Italy), renowed for the citrus cultivations, has been recognized in the last decades as experiencing severe and progressive salinization of the groundwater and then of the irrigated soils. Monitoring selected soil indicators provides data on salinization degree, also allowing unexpected relationships may be elucidated. Maps can be produced of actual soil salinity or potential risk of salinity and of soil attributes (e.g. Electric Conductivity (ECe), Exchange Sodium Percentage (ESP), Sodium Absorption Rate (SAR)) or combination of attributes. Assessment of soil salinity risk requires identifying critical indicators of salination and defining an operative methodology of integration, which combines and weights each factor appropriately in a global index. We have developed an approach to integrate an unlimited number of soil indicators. This method utilises an indicator tranform, which transforms measured data values into a binary value according to specific criteria. The criteria, developed independently for each indicator, are critical values or ranges of values which could negatively impact soil productivity. Each soil indicator is expressed on a landscape basis as the probability of areas taking the risk of salinization. To determine such a probability in unsampled points, we have utilised an approach based on non parametric geostatistics, called indicator cokriging, and GIS techniques. We have then applied factor kriging to define two regionalised factors summarizing the effects of different variables on soil salinization at different spatial scales. To illustrate how this approach can be used to evaluate soil salinity risk, we present an example of application to an area of southeastern Sardinia, which is considered representative of many Mediterranean zones at high soil salinity risk. We selected two variables (ECe and ESP) deemed most relevant for soil salinity assessment. We produced maps of both individual soil indicators and regionalised factors indicating the areas on a landscape basis that have a high probability to be saline and then cause a significant reduction in citrus production. In addition, this procedure allows to identify which indicator parameters are mostly responsible for zones at high salinization risk and to record the evolution of process, thus allowing specific management plans or land use policies to be developed.