Pedotransfer Function (PTF) for Estimation Soil moisture using NDVI, land surface temperature (LST) and normalized moisture (NDMI) indices

Available soil moisture of the land surface is one the key variables in controlling of the cycle of heat and water exchange between the earth and the atmosphere that manage this process with evapotranspiration of the surface. The amount of moisture is important for hydrological, biological and biochemical cycles, too. With Soil moisture data at regular time intervals, we can determine the degree of drought improvement in regions with dry climates. Also, continuous monitoring of soil moisture in agricultural areas can help us to efficiently crops irrigate. Soil moisture is also used to identify forest fire areas. Therefore, monitoring of soil moisture is important in any regions and different time periods. According to factors such as lack of uniformity in the physical characteristics of the soil, topography, land cover, evapotranspiration and precipitation and etc. Soil moisture is known as a variable parameter in a spatial and temporal domain. Therefore, the use of conventional and traditional methods of soil moisture (such as gravimetric and neutron probe) is not appropriate to understanding the spatial and temporal behavior of this parameter in large scales. To resolve this problem in past two decades, remote sensing technology (especially in visible/infrared spectrum) widely used to estimate of soil moisture indirectly. The objective of this study was to estimate surface soil moisture using Normalized Difference Moisture Index (NDMI), Normalized Difference Vegetation Index (NDVI) and Land Surface Temperature (LST) indices.

For this purpose, Landsat 8 satellite imagery was downloaded at the same time as ground sampling. The samples were transferred to the laboratory and soil moisture was calculated by weighted method. Then, using the specialized software such as ArcGIS, the indices were estimated and the values of these indicators were transferred to SPSS software for statistical regression. Statistical analyzes the studied indices and soil moisture content was measured. In this study, a PTF were obtained to predict soil moisture condition using LST and NDVI and NDMI derived from Landsat 8 data. Multiple linear regression method was used to derive the PTF. Soil moisture and soil organic matter were measured in the study area. After derivation of the pedotransfer function, the accuracy of the derived PTF was evaluated. This research was carried out in the city of Dehzad from Izeh Country, Khuzestan province.

Comparison between measured and predicted soil moisture values indicated that the PTF had a good prediction (R2=0.78) and Coefficient of Residual Mass (CRM) also showed that the model gave good performance (CRM=0.001). Furthermoreو a soil moisture map was obtained for the study area. The result indicated which Normalized Difference Moisture Index (NDMI), Normalized Difference Vegetation Index (NDVI) and Land Surface Temperature (LST) indices can be used to predict soil surface moisture content successfully.

The result of this research is presented by a PTF and in the form of soil moisture map. The soil moisture map simulated by this model can predict 78% of soil moisture variation in the region.