It is of considerable significance to estimate soil water content quickly and accurately by hyperspectral technique for the development of precision agriculture. In this paper, 76 soil samples collected from the Zhangqiu District of Jinan City, Shandong province, were taken as the research object. Firstly, nine transformation methods, such as differential, square, and square root, were used to transform the spectral data after denoising, and the estimation factors were selected according to the principle of maximum correlation. Then the grey-weighted distance correlation degree is used to recognize the pattern of the estimated samples, and the local linear regression estimation model of soil water content is established by using the known pattern samples closest to the samples to be identified. Finally, the determination coefficient and average relative error are adopted to evaluate the validity of the established model. The results showed that the maximum correlation coefficient among the five estimation factors was 0.94, and when the number of samples modeled by linear regression was 35, the estimation accuracy of the soil water content of 12 test samples was higher, among which the determination coefficient R2 was 0.993, and the average relative error was 3.50%. The results show that it is feasible and effective to estimate soil water content using the grey relational local linear regression model.

"> Soil Water content, Hyperspectral Remote Sensing, Grey Correlation Degree, Local Linear Regression