Comparison of soil moisture estimation based on LST-NDVI of Landsat 8 images with field soil sampling

Abstract

Estimation of soil moisture is very important to soil and agricultural scientists and requires many
attentions towards obtaining high and accurate results. So, satellite remote sensing data and
images have been extensively used to determining soil moisture especially with the help of
Landsat 8 OLI sensor images. In this research, 39 points were sampled for soil moisture at the
same time of Landsat 8 satellite overpass. LST and NDVI indices were used for modelling. In order
to estimate the accuracy of the results, the soil moisture obtained from the images was compared
with the results obtained from the soil science laboratory. The results showed that the estimated
moisture by the slope method of the Temperature Vegetation index (TVX) and the percentage of
soil moisture measured by use of LST-NDVI indices has a correlation of R2=0.64. The validation
results of soil moisture estimation model also showed that the model is able to predict surface soil
moisture with RMSE of 3.14. Also, for the study area, the equation %SM=-0.936LST0.515NDVI+36.7 was proposed to estimate the percentage of soil moisture