PERFORMANCE OF PRETREATMENTS AND MULTIVARIATE METHOD ON THE HYPERSPECTRAL ESTIMATION OF SOIL MOISTURE CONTENT

Abstract

Soil moisture controls the exchange of energy between the land surface and the atmosphere and is a significant factor affecting plant growth and productivity. Hyperspectral monitoring of soil water fraction could provide a theoretical basis for real-time estimation of spatial and temporal variations in soil moisture. To quantitatively evaluate the hyperspectral monitoring of soil moisture content (SMC): the SMC and its corresponding spectral reflectance were measured in the laboratory. In addition, the original spectral data was pre-processed by single and multiple transformations to study the effect of spectral preprocessing methods on the quantitative evaluation of soil moisture. The successive projections algorithm (SPA) was used to extract the corresponding wavelengths of soil moisture and the spectral monitoring model was established by using the partial least squares (PLS). The results show that (1) SMC and spectral reflectance show an obvious negative correlation, and the spectral reflectance gradually decreases with the increase of SMC. (2) Appropriate pretreatment methods can improve the correlation between SMC and spectral reflectance and improve the accuracy of the SMC monitoring model, of which T19 (R2 + SNV + FD) is the best spectral pretreatment method. (3) The optimal SMC monitoring model is T19-SPA-PLS (R2v = 0.986, RMSEv = 1.824, RPD = 8.239). This study provided a reference for spectral data processing and an effective method for the accurate estimation of SMC using hyperspectral remote sensing.