Soil moisture monitoring using hyper-spectral remote sensing technology

Precision agriculture needs surface soil moisture information accurately and quickly. Hyper-spectral remote sensing can be used to detect slight differences in soil moisture because of high-resolution and multi-band in spectral. Taking black soil in Jilin Province of China as the research object, this paper analyzed soil hyper-spectral characteristics and extracted parameters by using methods of spectral differentiation, feature-band extraction and multiple stepwise regression analysis. The relationship between soil moisture and soil spectra was analyzed combining with soil moisture laboratory measurement. Five models were used to quantitative inversion in order to seek soil moisture monitoring by applying hyper-spectral remote sensing. The conclusion was as follows: (1)in the below field water holding capacity condition, the sensitive bands of black soil spectral reflectance and its reciprocal and logarithmic transformation mainly focused in 400-470nm, 1950-2050nm and 2100-2200nm. The highest correlation coefficient between laboratory spectral data and soil moisture reached to 0.89 at 2156nm; (2)The best prediction model of black soil moisture content by using hyper-spectral remote sensing was Y = 22.16 + 26278.2x₁₃₂₈-47785.1x₁₄₃₉- 201.42x₁₇₄₂ + 49306.34x₂₁₅₆, x= (lgR)´, Y representing soil moisture content (%), coefficient of determination was 0.931.