实测高光谱和HSI影像的区域土壤盐渍化遥感监测研究

通过典型研究区不同盐渍化土壤光谱反射率数据的变换和分析，选择与土壤含盐量响应敏感波段，建立实测高光谱土壤含盐量反演模型，以校正HSI影像建立的土壤含盐量反演模型。结果表明：实测高光谱土壤含盐量反演模型与HSI影像土壤含盐量反演模型均有较好的精度，模型判定系数(R2)均高于0.57，且模型稳定性较好。校正后的HSI影像土壤含盐量反演模型，模型判定系数有了较大提高，R2从0.571提升至0.681，且通过了0.01的显著性水平，均方根误差(RMSE)值为0.277。模型能够较好地提高区域尺度条件下土壤盐渍化监测精度，运用此方法开展盐渍化土壤定量遥感监测是可行的。

Study on the Soil Salinization Monitoring Based on Measured Hyperspectral and HSI Data

The present paper selects the Kuqa Oasis as the study area, studies spectrum characteristics of soil salinity, and establishes soil spectrum library. Through transforming and analyzing varying degrees of soil salinization reflectance spectra data in the typical study area, and selecting the most sensitive spectral bands in response to salinization, we established the measured hyperspectral soil salinity monitoring model, and by correcting the soil salinity monitoring model established by HIS image through scale effect conversion improved the model accuracy under the conditions of a regional-scale monitoring of soil salinization. The results show that both measured hyperspectral soil salinity monitoring model and HSI image soil salinity inversion model have good accuracy, model determination coefficient (R2) is higher than 0.57 and the model stability is better. Compared with the corrected HSI image soil salinity inversion model and uncorrected HSI image soil salinity inversion model, the coefficient of determination has been greatly improved, which increased from 0.571 to 0.681, and through the 0.01 significance level, the root mean square error (RMSE) value is 0.277. The correction HIS image soil salinization monitoring model can better improve the model accuracy under the condition of regional scale soil salinization monitoring, and using this method to carry out the soil salinization quantitative remote sensing monitoring is feasible, and also can provide scientific reference for future research.