The measurement and control of soil moisture are the key technologies of precision agriculture. In order to real-time detect soil moisture content faster and more accurately, a portable soil moisture sensor based on NIR spectroscopy was developed. With the sixty soil samples collected from a winter jujube orchard, a linear regression model was established. The determination coefficients of the calibration ($R^2_c$) and validation ($R^2_v$) reached 0.88 and 0.92, respectively. The model passed F-test and t-test and showed robust. Subsequently, two spatial distribution maps of soil moisture were generated based on the data obtained by the portable soil moisture detector and the data obtained by oven drying method, respectively. Finally, the correlation between these two maps was investigated by using the software of Surfer 8.0. The zones of dry and wet soil could be distinguished easily in both maps. The results of the study showed that the developed detector was practical.