基于野外高光谱的黄河三角洲滨海盐渍土盐分含量估测研究

Estimation of Coastal Saline Soil Salinity in the Yellow River Delta Based on Field Hyperspectra

为探索适合黄河三角洲滨海地区土壤盐分含量的快速准确估测方法,以黄河三角洲垦利县为研究区,采用野外原位土壤高光谱数据与室内实测土样盐分数据相结合的分析方法,进行土壤盐分估测研究。通过对土壤盐分高光谱反射率、一阶微分反射率与盐分含量进行相关性分析,筛选出土壤盐分敏感波段;通过波段组合构建并筛选出土壤盐分敏感光谱参量;通过对敏感波段进行主成分回归建模、多元逐步线性回归建模和运用光谱参量建模,筛选出最佳的估测模型。结果显示:一阶微分可放大样品间的光谱特征差异,提高了相关性;波段组合可消除部分背景因素影响,与盐分的相关性明显提高;利用敏感光谱参量构建的估测模型优于直接用敏感波段构建的模型,土壤盐分最佳估测模型为y=0.623-2003920.934*X507*X772+0.259*(X512+X1093)-469.717*X772/X512,R2为0.591,验证R2为0.556,RPD为1.624,RMSE为0.116,拟合度较好,稳定性较高。该研究为滨海区土壤盐分含量的野外高光谱估测提供了新的方法,同时为土壤盐分含量的高光谱定量研究提供理论和技术参考。

In order to explore a rapid and accurate method for estimating soil salt contents in the coastal area of the Yellow River Delta, this study tries to estimate soil salinity by a combined method of field hyper-spectral data and indoor measured soil salinity data collected from Kenli County. Firstly, the sensitive bands of soil salinity were screened out through correlation analysis between hyper-spectral reflectance and the first order differential reflectivity with the soil salinity. Then, the sensitive spectral parameters of soil salinity were identified through band combination and selection. Finally, the best estimation model was established based on the principal component regression modeling and multiple linear regression modeling by using sensitive bands and sensitive spectral parameters. The results show that the first order differential can enlarge the difference of the spectral characteristics of the samples and improve their correlation. Band combination was helpful for eliminating the background effects as well as improving the correlation with soil salinity. The estimation model based on sensitive spectral parameters was better than that based on the sensitive bands directly. This study provided a new method for the field hyper-spectral estimation of soil salinity contents in the coastal area, and it also provided a theoretical and technical reference for the hyper-spectral quantitative estimation of soil salt contents.