基于高光谱的喀斯特地区典型农田土壤有机质含量反演

Inversion of Soil Organic Matter Content in Typical Farmland of Karst Region Based on Hyperspectral Data

【目的】利用高光谱数据定量反演喀斯特地区土壤有机质含量,为喀斯特地区快速、大范围、实时地监测土壤有机质含量提供更多的技术手段。【方法】利用机载高光谱成像系统和便携式地物光谱仪分别获取土壤光谱数据,基于原始光谱反射率和不同光谱变换数据,分析其与土壤有机质含量的相关性,以偏最小二乘法建立模型预测土壤有机质含量。【结果】2种数据源都可以用于土壤有机质含量预测,其中,基于ASD光谱一阶微分变换建立的模型预测精度较高,验证集决定系数(Rv^2)为0.910,相对分析误差(RPD)为2.68;基于GS光谱二阶微分变换建立的模型预测效果较好,验证集Rv^2为0.772,RPD为1.49。【结论】ASD光谱与GS光谱建模预测精度相差较大,ASD光谱客观条件影响较小、光谱波段更宽、光谱分辨率更高,具有更好的预测能力;低空无人机获取的GS光谱也具有一定的预测能力。

【Objective】The soil organic matter content in karst region is quantitatively inverted by the hyperspectral data to provide more technological means for rapid,extensiveness and actual time monitoring of soil organic matter content in karst region.【Method】The soil spectral data are acquired by airborne hyperspectral imaging system and portable ground spectrometer and the correlations between the acquired soil spectrum data and soil organic matter content are analyzed based on original spectrum reflectance and different spectrum transformation data.The prediction model of soil organic matter content is established by the partial least squares.【Result】Two data sources both can be used for prediction of soil organic matter content.The precision of the established model based on first order differential transformation of ASD spectrum is high and the verification set coefficient(Rv 2)and relative percent deviation(RPD)are 0.910 and 2.68 separately.The prediction effect of the established model based on second order differential transformation of GS spectrum is better and the verification set coefficient(Rv 2)and relative percent deviation(RPD)are 0.772 and 1.49 respectively.【Conclusion】There is a great difference in prediction accuracy of soil organic matter content between the established model based on ASD spectrum and the established model based on GS spectrum.ASD spectrum with the advantages of less objective condition influence,wider spectral band and higher spectral resolution has the better prediction capacity of soil organic matter content but GS spectrum acquired by the low-altitude UAV has a certain prediction capacity of soil organic matter content.