Geostatistical Zoning of Soil Temperature in Maize Root-Zone under Different Irrigation Strategies

Soil temperature is one of the most effective parameters for plant growth. Although the spatiotemporal monitoring of soil temperature is important for providing an optimum condition to maximize the yield efficiency, indirect measurements for zoning  of soil temperature are essential due to difficulties in direct methods of measuring it and also the limited measurement points. Thus, the efficiency of some geostatistical methods including weighted moving average, kriging, and co-kriging with soil water content and electrical conductivity, as covariates, were investigated for estimating the soil temperature under three irrigation treatments: (FI) full irrigation, (PRD₇₅) partial root zone drying at 75%, and ( PRD₅₅) at 55% levels. TDR sensors were used to measure soil temperature, water content, and electrical conductivity. Comparing the observed and estimated soil temperature by geostatistical methods based on root mean square error (RMSE), mean absolute error (MAE), mean bias error (MBE), and index of agreement (d) indices confirmed the validity of the selected methods for zoning of soil temperature in the plant root zone under all treatments. Duncan variance analysis showed that the values of evaluation indices for kriging method were significantly lower than that for all other geostatistical methods under all treatments (P<0.05). Results confirmed that using soil water content as covariate in co-kriging method significantly decreased the range of influence in all treatments and, also, using the soil electrical conductivity as covariate in co-kriging method caused a discontinuity of the cross-variogram. Thus, it could be concluded that even under the especial soil water content condition in partial root-zone drying strategy, kriging method significantly decreases the estimation error of soil temperature compared with co-kriging method with soil water content as covariate.