论文部分内容阅读
采用基于混合像元的结构分析方法和支持向量机(SVM)算法,建立了高分辨率遥感数据(TM)向低分辨率遥感数据(MODIS)的尺度转换模型,实现了由高分辨率遥感数据获得的NPP向低分辨率遥感数据获得的NPP的空间尺度转换。对低分辨率遥感数据(MODIS)估算的NPP结果进行了尺度效应校正。结果表明:SVM回归模型模拟出的尺度效应校正因子Rj_corrected与1-F中覆盖度草地之间的相关性较高,R2达到0.81。尺度效应校正前的NPPMODIS与NPPTM的相关性较低,R2仅为0.69,RMSE为3.47;尺度效应校正后的NPPMODIS_corrected与NPPTM的相关性较高,R2达到0.84,RMSE为1.87。因此,经过尺度效应校正后的NPP无论是在相关性还是在误差方面有了很大程度的提高。
A scale conversion model of high resolution remote sensing data (TM) to low resolution remote sensing data (MODIS) is established by using structural analysis of mixed pixels and support vector machine (SVM) algorithm. The high resolution remote sensing data Obtained NPP Spatial Scale Conversion of NPP to Low Resolution Remote Sensing Data. The scale effect correction of the NPP results estimated by MODIS was carried out. The results show that the correlation between the scale effect correction factor Rjcorrected by SVM regression model and the grassland with 1-F coverage is high, and R2 reaches 0.81. The correlation between NPPMODIS before scaling effect and NPPTM is low, with R2 of only 0.69 and RMSE of 3.47. Correlation of NPPMOD with scale effect has a high correlation with NPPTM, R2 of 0.84 and RMSE of 1.87. Therefore, the NPP corrected by scaling effect has been greatly improved both in the correlation and in the error.