Remote sensing and relief data to predict soil saturated hydraulic conductivity in a calcareous watershed, Iran

Soil saturated hydraulic conductivity (Ks) is a crucial property in hydrology, soil erosion and conservation, water resource and agriculture. In this study, we developed regression models for predicting Ks by employing easily-to-measure soil properties, remote sensing and relief data in a calcareous watershed, Iran. Surfer soil samples (0-30 cm) were collected in 106 points under different land uses (i.e. farmlands, rangelands and bare lands). Various easily-to-measure soil properties including sand, silt, clay, organic carbon, bulk density, calcium carbonate equivalent and soil moisture were determined for each sampling point. Moreover, the remote sensing data were derived from Landsat 8 satellite and relief data were acquired from DEM with 30 m resolution. The Ks was predicted by developed PTFs in literature and by using four new scenarios including the easily-to-measure soil properties (Scenario I), relief indices (Scenario II), remote sensing indices (Scenario III) and easily-to-measure soil variables plus relief indices plus remote sensing indices (Scenario IIII). The results indicated that for all data, the Crosby's model was the best model for predicting Ks (ME =-31.37 mm h(-1), RMSE = 40.45 mm h(-1) and R-2 = 0.585). Models used soil particles (sand and clay) and soil particles plus organic matter showed the highest performance for predicting Ks in farmlands, rangelands and bare lands, respectively. Additionally, applying the scenario IIII in all data resulted a high accuracy for predicting Ks data (ME =-4.186 mm h(-1), RMSE = 21.908 mm h(-1) and R-2 = 0.713). Moreover, scenario IIII in different land uses yielded better results in the prediction of Ks compared to other scenarios. It was concluded that the combining the soil data and environmental data could highly improve Ks prediction.