Calibration of the SPAW model for estimating plant available water in the Kurdistan region of Iraq

The SPAW computer model can predict soil moisture under various tensions based on particle size distribution, gravel content, compaction, salinity, and organic matter. This ability may be decreased in soils with high levels of calcium carbonate due to its effect on soil properties related to moisture content. Therefore, this research evaluated and calibrated this computer model for predicting field capacity (FC) and wilting point (WP) in calcareous soil. A dataset of ninety-two soil samples used to cover a wide span of calcium carbonate and clay content. Seventy-six soil samples were used as input data (training set) to determine FC and WP through SPAW computer model application. The calibrated SPAW models (SPAWcal) were formulated via relationship between predicted data from the SPAW computer model and training set. Sixteen soil samples were used as a testing dataset for validating predicted FC and WP values by (SPAW & SPAWcal). The results showed that the accuracy of the SPAWcal model outperformed the SPAW original model, as predicted data by SPAWcal model were closer to actual data. The RMSE decreased from 3.316 to 2.642 for WP and from 3.104 to 2.247 for FC. The Bland-Altman analysis presented a shift in the mean difference from overestimation to underestimation (from -1.648 to 0.455 for FC and from -1.320 to 0.251 for WP). These results confirm that in view of calcium carbonate content in the estimation of both FC and WP significantly improves prediction accuracy. Thus, the calibrated SPAWcal model bids a more accurate and reliable tool for soil moisture prediction in calcareous soils.