Obtaining soil moisture quickly and timely can grasp the needs of water of the crops, which is very important for the agricultural production. Soil spectral reflectance provides an alternative method to classical physical and chemical analysis of soil in laboratory for the estimation of a large range of key soil properties. Therefore, the soil moisture was quickly achieved by using hyperspectral technology and the application of ridge regression was explored in the optimization and quantitative analysis of hyperspectral bands. Totally 91 soil samples were collected from the soil depth of 0~5cm in Seder Boker area in the southern Israeli. These soil samples were analyzed in the process of physical and chemical properties in laboratory. After that, the raw hyperspectral reflectance of soil samples was measured by an ASD FieldSpec 3 instrument equipped with a high intensity contact probe under the darkroom conditions. Next, the raw spectral reflectance (REF) was transformed to three spectral indices, i.e. inverselog reflectance (LR), the first order differential reflectance (FDR) and continuum removal reflectance (CR). Regression models of soil moisture with different indices were established by three methods: partial least squares regression (PLSR), stepwise regression (SR) and ridge regression (RR). The inversion results of the model were validated and compared with each other. The results showed that the method of LR transform can eliminate the interference of external factors much better, and it appeared to be the optimal spectral index in stepwise regression model and ridge regression model (R2c were 0.981 and 0.975, and R2p were 0.971 and 0.979). For the three spectral indices about REF, FDR and CR, although the modeling effect of SR and RR was slightly lower than that of PLSR, the coefficient of modeling determination was above 0.9. Both SR and RR had simplified and optimized the model, but RR had better validation results and the number of bands used for modeling was only 0.3% of the full spectrum (400~2400Nm). After comparing the three regression models established with the four spectral indices, the LR-RR model not only had the characteristics of simple model and less calculation, but also improved the robustness of the model better by using biased estimation at the cost of losing the part accuracy. The result indicated that ridge regression method can not only achieve the efficient selection of hyperspectral bands, but also use the LR-RR hyperspectral inversion model for the reference of monitoring the aerospace hyperspectral remote sensing of regional soil moisture in the future.