The sustainability production of dryland agriculture is threatened by salt accumulation in soil due to irrigation practices by saline waters. However, the dynamic processes of secondary soil salinization depend on some factors varying in time and space. The aim of this research was to introduce an approach for the prediction of soil salinity
in some irrigated pistachio (Pistacia vera L.) orchards facing secondary soil salinization. The study area was Ardakan (Yazd Province, Central Iran). In this approach, the Landsat ۸ satellite data bands and satellite–based driven data (indices) were used. The Partial Least Square Regression (PLSR) method was used to predict the variability of soil salinity
with minimum (zero) ground measurements. The predicted soil salinity
(electrical conductivity) of soil saturated paste extract (ECe) were compared by the measured ECe. The existing conventional methods (e.g. WatSuit
computation model) using ancillary measured data of irrigation water salinity
(ECiw) and corresponding leaching fractions (LF) were also used for evaluation. The Results of the satellite-based PLSR method showed an R۲ of about ۶۴% between predicted and measured soil salinity, while this indicator was about ۷۲% for the conventional model of WatSuit. The higher accuracy of the Watsuit model is owing to its dependence on ground measurements, while the introduced satellite-based PLSR approach was able to predict temporal changes of soil salinity
in patterns fitted to the irrigation intervals with zero dependence on the ground truth data.