The Parallel Method And Implementation Of Ecological Remote Sensing Parameter Inversion Based On MapReduce Model

With the continuous development of remote sensing technology, remote sensing data acquisition represents three-multi (multi-platform, multi-sensor, multi-angle) and three-high (high spatial resolution, high spectral resolution and high temporal resolution) trend. The rapid increase of remote sensing data and the limited data processing capabilities of traditional remote sensing image processing software make the image processing speed become the bottleneck of remote sensing technology application and development in different areas. In order to improve the remote sensing image processing speed and meet the real-time or near real-time processing demand of remote sensing application in ecological environment monitoring and disaster forecast, this paper investiges the potential application of cloud computing and parallelization computing techniques to remote sensing data processing, uses long time series of MODIS data to carry out Three Rivers District ecological environment parameter parallel inversion based on MapReduce model in open source cloud computing platform Hadoop. The main contents in this paper are as the followings:1) Cloud computing platform and MapReduce parallel model execution mechanism. This paper in-depth studies cloud computing platform and MapReduce parallel programming model, builds cluster system based on cloud computing platform, establishes the parallel framework based on MapReduce model, analyzes the basic idea of cloud computing and the cloud computing platform architecture in detail, in-depth parses the implementation mechanism and running mode of parallel frame based on MapReduce model.2) Data partitioning method for the parallel processing of remote sensing data. This paper summarizes the existing remote sensing data parallel processing sub-block methods, in-depth studies different remote sensing image processing methods need for remote sensing data sub-block in parallel computing. According to the selected research area, data source, the proposed inversion parameter and the characteristics and actual demand of the selected image inversion method, in view of different process and stage of MapReduce parallel frame, the paper designs the compatible data sub-block method, to ensure the selected ecological environment parameters of the study area be parallel inversed efficiently and reliably.3) Parallel framework combination application based on MapReduce model. On the basis of serial remote sensing data processing, this paper in-depth analyzes remote sensing image application requirement in different process or stage, studies and designs the parallel framework combination of different ecological remote sensing parameter used in parallel inversion; studies the frame combination method and forms three frame combination application modes, which are one frame combination application, independent frame combination application and relational combination application, to complete parallel inversion of variety ecological environment factors.4) Remote sensing image parallel processing method based on MapReduce model. Combing MapReduce model and ecological remote sensing parameters inversion model and algorithm, this paper makes ecological remote sensing parameter inversion method be integrated into MapReduce model parallel frame, achieves to combing ecological remote sensing parameter inversion algorithm and MapReduce parallel frame organically, and studies the implementation method of applying MapReduce parallel model to different ecological remote sensing parameter inversion.