基于MapReduce模型的生物量遥感并行反演方法研究

Remote sensing retrieval method for biomass based on MapReduce parallel model

MapReduce模型是一种基于云计算平台下新型的并行编程模型。文中将MapReduce并行编程模型应用到遥感影像并行化处理中,以2005-2009年5a生长季期(5-10月)MODIS13Q1数据产品为数据源,对青海省三江源地区的生物量(草地总生物量和可食草量)进行并行化反演,研究基于该模型的生物量遥感并行反演方法。实验分析结果表明:基于该模型的并行生物量遥感反演结果与经过精度验证的串行反演结果一致,并行化反演结果准确、可信;并行化反演效率较串行化反演效率有大幅提高,并随着计算节点的增加,并行效率不断提高。

MapReduce is a new parallel programming model based on cloud computing platform. The MapRe duce parallel programming model was applied to remote sensing image parallel processing, and Three River Source Region biomass （total biomass of grass and grazing capacity） in Qinghai Province was taken as an exam ple to study the remote sensing retrieval method for biomass in a paralleling way by using growing season period MODIS13Q1 data products in 2005 2009 as the data source. The experimental analysis shows that： parallel in version results based on the MapReduce model are consistent with serial inversion results which accuracy is vali dated and parallel inversion results are accurate and credible. The parallel retrieval efficiency has been greatly improved than the serial inversion efficiency; with the computing nodes increase, the parallel efficiency continues to increase.