In agricultural remote sensing detection, clear weather is the necessary condition to collect spectrum. The main purpose of this study is: (1) to explore the influence of different weather (sunny, cloudy, cloudy) on the polarization parameters (S₀, ε₀, f₀₀), linear degree of polarization (Dolp) and physical and chemical value modes of different wavelengths of the BRDF of Dongzao in South Xinjiang through the hyperspectral polarization detection technology; (2) to compare the experimental results under different weather conditions for the remote control of outdoor jujube quality sensing detection provides a certain reference for environmental adaptability and prospective application. In this study, the statistical relationship between the water content of Dongzao jujube in South Xinjiang and its spectral polarization characteristics were analyzed. The first derivative spectral form of S₀, ε₀, f₀₀, Dolp and the statistical equation of water content of Dongzao jujube in South Xinjiang were established under different weather conditions. Three indexes were selected: correlation coefficient r, the standard deviation of calibration set (RMSEC) and standard deviation of prediction set (RMSEP) The performance and prediction ability of the model are evaluated. The results of correlation analysis showed that there was a good correlation between water content and hyperspectral polarization data when the quality of winter jujube was detected in sunny, cloudy and cloudy weather conditions, but the correlation between polarization parameter, linear polarization and water content was better than that in sunny and cloudy weather conditions. The former model had the largest correlation coefficient, and its r-score was the highest They are 0.913 and 0.914, and the value of correlation coefficient r of dark box spectrum and moisture content model is closest to 0.926. The feasibility analysis results of the model show that the maximum standard deviation of the calibration set samples of polarization parameter, linear polarization degree and water content model under three weather conditions are 0.009 71 and 0.008 73, respectively. The smaller the RMSEC value is, the better the model is established. The results show that the maximum standard deviation of the three weather models is 0.012 3 and 0.011 7, respectively. The smaller the RMSEP value is, the better the prediction ability of the model is. The results of different weather experiments show that the polarization can weaken the strong light and strengthen the weak light. By comparing the experimental results of sunny, cloudy and cloudy days, the method has better environmental adaptability and has a wide application prospect in outdoor jujube quality remote sensing detection.