基于RS和GIS的西河流域土壤有机碳含量的空间反演

基于遥感(RS)和地理信息系统(GIS)方法掌握土壤有机碳含量是当前土壤学科的发展趋势。本文利用西河流域Landsat 8 OLI遥感影像及121个样点土壤表层(0~20 cm)有机碳含量和地面辅助数据建立预测模型并进行了空间反演。结果表明,仅用遥感光谱信息建立的土壤有机碳含量预测模型均达极显著水平(P=0.000),表明遥感光谱信息能用于表层土壤有机碳含量的预测建模。在引入成土母质、地貌类型和农地利用方式等地面辅助因子后,预测模型决定系数R2明显增大(P=0.000),由0.385提高到了0.579,这表明地面辅助因子能有效改善模型精度。空间反演显示,最优模型的插值图能较好地表现区域土壤有机碳含量分布的基本格局,但对于个别值区的反演效果仍有待进一步提升。

Spatial prediction of soil organic carbon based on RS and GIS in West River Valley

Understanding of soil organic carbon content ( SOC ) based on remote sensing ( RS ) and geographic information system ( GIS) techniques becomes the development trend of current discipline. In this research, remotely sensed spectral data from Landsat 8 OLI, soil organic carbon content (121 samples from 0~20 cm soil layer) and the related ground parameters in the West River Valley were integrated to construct models for spatial prediction in the area. Space inversion was employed to check the model reliability. Results indicated that evaluation of SOC content was achieved by the constructed models using re-mote sensing data (P=0. 000), implying that it was potentially reliable to predict SOC content. When the related ground pa-rameters of soil parent material, landforms and agricultural land use were considered, respectively, the R2 values of prediction models were significantly improved (P=0. 000). Concretely, the R2 was increased from 0. 385 to 0. 579, indicating the in-volvement of ground parameters was beneficial to prediction accuracy. In contrast to the determined content, optimal interpola-tion model better reflected the basic pattern of regional distribution for SOC content. However, the accuracy of spatial inversion should be improved in some special areas.