Using Probability‐Based Geostatistics to Evaluate the Spatial Distribution of Groundwater Salinity for Irrigation

Groundwater is a key source of water for agriculture, although it may become salinized by several natural and anthropogenic sources. Saline groundwater for irrigation results in soil salinization and reduces crop productivity. This study used indicator kriging (IK) to evaluate the spatial suitability of groundwater electrical conductivity (EC) levels for irrigation in the Pingtung Plain, Taiwan, because of highly skewed data on EC. IK was used to establish a conditional cumulative distribution function (CCDF) for groundwater EC. For robust estimation for skewed data, two approaches of probability‐based estimation, estimated probability values and p‐quantile estimates based on the CCDF, were investigated at an EC level of 750 μS/cm for the Taiwanese standard for irrigation water quality. These estimates were compared against the observed EC values to determine the optimal classifications of suitable irrigation. An interquartile range was used to examine the local uncertainty of the estimates. The results revealed groundwater with EC levels exceeding 750 μS/cm, primarily in southern coastal regions. Estimated probability values of 0.7 and 0.7‐quantile estimates based on the CCDF at an EC level of 750 μS/cm were used to identify regions with suitable groundwater quality for irrigation. Moreover, regions with more saline groundwater had greater local uncertainty.