Remote detection of bare soil moisture using a surface-temperature-based soil evaporation transfer coefficient

An approach for estimating soil moisture is presented and tested by using surface-temperature-based soil evaporation transfer coefficient (h(a)), a coefficient recently proposed through the equation h(a) (T-s - T-a)/(T-sd - T-a), where T-s, T-sd, and T-a are land surface temperature (LST), reference soil (dry soil without evaporation) surface temperature, and air temperature respectively. Our analysis and controllable experiment indicated that h(a) closely related to soil moisture, and therefore, a relationship between field soil moisture and h(a) could be developed for soil moisture estimation. Field experiments were carried out to test the relationship between h(a) and soil moisture. Time series Aqua-MODIS images were acquired between 11 Sep. 2006 and 1 Nov. 2007. Then, MODIS derived h(a) and simultaneous measured soil moisture for different soil depths were used to establish the relations between the two variables. Results showed that there was a logarithmic relationship between soil moisture and h(a) (P < 0.01). These logarithmic models were further validated by introducing another ground-truth data gathered from 46 meteorological stations in Hebei Province. Good agreement was observed between the measured and estimated soil moisture with RMSE of 0.0374 cm(3)/cm(3) and 0.0503 cm(3)/cm(3) for surface energy balance method at two soil depths (10 cm and 20 cm), with RMSE of 0.0467 cm(3)/cm(3) and 0.0581 cm(3)/cm(3) for maximum temperature method at two soil depths. For vegetated surfaces, the ratio of h(a) and NDVI suggested to be considered. The proposed approach has a great potential for soil moisture and drought evaluation by remote sensing. (C) 2010 Elsevier B.V. All rights reserved.