Presenting a New Method for Soil-moisture Estimation Using Optical Remotely-sensed Imagery

Optical remote sensing of earth surface processes commonly relies on the Red, Green, Blue (RGB), Near Infrared (NIR) and Shortwave Infrared (SWIR) electromagnetic bands. In soil-moisture estimation method using optical remotely-sensed imagery, by assuming a linear relationship between the Red-NIR reflectance, the line of bare soil (soil line) is established as the base and then moisture isoclines are assumed perpendicular to the soil line. This study is intended to show that this assumption is not consistent with the actual Red-NIR space geometry, which in many cases introduces soil moisture estimation errors. Therefore, a new mathematical transformation method was proposed to the original Red-NIR space followed by newly-defined soil moisture isolines. This new Transformed Red-NIR (TRN) model was compared with the conventional Red-NIR (CRN) model using data from Salman Farsi sugarcane field located in Khozestan province in southwestern of Iran. Twelve Landsat-8 satellite images were used during the sugarcane growing season. For validation of the remotely sensed data, ground reference soil moisture was measured at 22 locations at five different depths. Results of the proposed new method significantly improved accuracy of the Red-NIR approach to remote sensing of soil moisture.