为探讨不同土壤水分含量对复垦土壤高光谱特征及土壤属性反演精度影响，以安徽省淮北市某煤矿复垦区为研究对象，采集土壤样本并测定土壤理化性质及风干前后的土壤光谱信息。分析复垦与非复垦土壤之间的差异，明确不同水分条件下的光谱吸收特征，采用配对样本t检验探究不同水分条件下光谱影响波段；利用辐射传输模型和直接标准化算法去除土壤水分影响，并结合吸光度变化比较2种算法剔除效果；以土壤全氮为例，采用不同的波段筛选方法皮尔森相关系数PCC、连续投影算法SPA及其组合算法PCC-SPA建立干土、湿土及校正后光谱的土壤属性预测模型。研究结果表明：风干后复垦土壤光谱反射率明显高于非复垦土壤，两种土壤光谱在可见光波段的差异明显高于近红外波段；土壤水分影响较大的中心波段为1 000、1 450、1 950和2 200 nm，当含水率过高或过低时还会引起可见光波段的差异；与直接标准化算法相比，辐射传输模型能更加有效地去除土壤水分影响，得到的光谱曲线平均吸光度与干土近乎重合；受土壤水分影响，复垦土壤全氮的预测模型降低45.38%，采用辐射传输模型去除土壤水分后建立的预测模型精度显著提高，验证集R2提高超30%，且与干土R2相差小于10%；3种波段筛选方法均能有效地筛选出重要波长变量，其中PCC-SPA筛选出的波段不超过7个，极大优化了模型结构，减少建模时间。该研究结果可作为该区域湿土土壤属性估测的重要方法，为进一步提高煤矿复垦土壤原位属性监测提供新的思路。

In order to explore the influence of different soil moisture content on reclaimed soil perspectra characteristics and inversion accuracy of soil properties, this paper takes the reclamation area of a coal mine in Huaibei, Anhui Province as the research object, soil physical and chemical properties and soil spectral information before and after air drying were measured. Analyze the differences between reclaimed and non-reclaimed soils, and clarify the spectral absorption characteristics under different moisture conditions. Paired sample ttest was used to explore the spectral influence bands under different moisture conditions. Radiation transfer model and direct standardization algorithm were used to remove soil moisture, and the removal effects of the two algorithms were compared with the change of absorbance . Taking soil total nitrogen as an example,different spectral screening methods (Successive Projections Algorithm, SPA、Pearson Correlation Coefficient, PCC、PCC-SPA) were used to establish soil properties prediction models for dry soil, wet soil and corrected spectral. The results show that:① the spectral reflectance of reclaimed soil was significantly higher than that of non-reclaimed soil after air drying, and the spectral difference between reclaimed soil and non-reclaimed soil in visible band was significantly higher than that in near-infrared band. The central bands of soil moisture were 1 000, 1 450,1 950 and 2 200 nm, and the difference of visible band was also caused when the water content was too high or too low. ② Compared with the direct standardized algorithm, the radiative transfer model can remove soil moisture more effectively, and the average absorbance of the obtained spectral curve almost coincides with that of dry soil.Under the influence of soil moisture, the prediction model of total nitrogen in reclaimed soil decreased by 45.38%. The accuracy of the prediction model established after removing soil moisture by radiative transfer model was significantly improved, and the validation set R2 was increased by more than 30%, and the difference with dry soil R2 was less than 10%. ③ Three band screening methods can effectively screen out important wavelength variables, among which no more than 7 bands are screened by PCC-SPA, which greatly optimizes the model structure and reduces the modeling time.The results of this study can be used as an important method to estimate the soil properties of wet soil in this region, and provide a new idea for further improving the in-situ monitoring of soil properties in coal mine reclamation.