摘要
基于空间变异理论,本文提出了一种基于半方差函数的STARFM改进模型。该模型利用半方差函数分析基准Landsat影像的空间统计特性,确定影像各波段的空间相关距离——变程,然后将变程的2倍作为STARFM模型滑动窗口的宽度,预测了指定MDOIS时刻的Landsat影像;改变滑动窗口大小后的预测影像和实际观测影像各波段的平均绝对差值和平均差值均大于利用半方差函数确定滑动窗口宽度的预测影像,表明STARFM改进模型具有优越性。
Based on spatial variation theory, this paper put forward an improved STARFM algorithm. The variograms computed on all bands of the based Landsat was conducted, then the move window width was set to twice of this rang, finally the Landsat image at the specified MODIS time was predicted. Results showed that the average absolute difference and average difference between the pre- dicted and actual images are increased while the width of window changed, proving the improvement of this algorithm.
出处
《测绘科学》
CSCD
北大核心
2013年第3期140-142,92,共4页
Science of Surveying and Mapping
基金
“基于全球碳循环模拟下的北半球温带草原生态系统功能评价研究”(Y2ZZ19101B)
863重大项目“星机地立体组网协同观测关键技术”(2012AA12A301)
