Research On Application Of Remote Sensing Tupu

In the process of crop growth, it is of great significance to access crops growth information scientifically and rapidly for production field management and crop pest/disease monitoring and early warning. Nowadays, multi-source and multi-scale high resolution remote sensing images have gradually been widely used on agricultural condition monitoring, but there are still several disadvantages on practical application because of characteristics of remote sensing such as large volume of data, wide range of items, different organizational rules and complicated preprocessing procedure. Focusing on these problems, this manuscript considers actual requirements of crop natural disasters remote sensing monitoring and discrete distribution characteristics of monitoring objects in farm-scale, on that basis study storage environment, data organizational mechanism, efficient data computing model; implement massive remote sensing data rapid and steady storage, management, maintenance, archiving; and thereby focus our research on remote sensing data reorganization, cleaning method and practical data automatic preprocess method. At last, we design and implement farmland scale natural disaster remote sensing monitoring prototype system, using GF-1 WFV remote sensing data as experimental data, executing drought monitoring in local region of LiaoNing province to test the computational efficiency, computational accuracy, stability and practicability of our method. And the experiment shows that the data application method presented can fulfill the near real-time requirement of natural disasters remote sensing monitoring, and express advantages of well accuracy and stability, and have certain significance to researches on remote sensing based crop natural disasters monitoring.The main research work of this paper involves the following aspects:(1) Analyze applicable conditions and shortcomings of different data storage hardware architectures and distributed file system architectures, on that basis deploy distributed metadata model based data storage environment and compare our storage environment to traditional NAS Disk Array by different types of spatial data throughput tests. We also study the proper data format and compress mode by test the influence of data formats and compress modes to data storage, transmission and calculation efficiency, and thereby formulate unified data directory system and organization rule. Furthermore, we make the coordinate mechanism from the perspective of data storage, data calculation and data expression, choose proper computing mode by analyzing concepts, calculation process and suitable conditions of cluster parallel computing and task concurrency computing. At last, we deploy infrastructure environment for the following study base on the above-mentioned research.(2) Analyze advantages and disadvantages of different grid organization rules, reference the spatial reference, grid partition and coding method, data retrieve strategy, application characteristics of three typical grid system:SMG, GeoSOT, China National Standard Geographic Grid, research Raster Dataset Clean & Reconstitution Multil-grid (RDCRMG) according to the actual requirement of remote sensing based natural disaster monitoring, including grid design goals and principles, grid spatial reference, grid partition and coding scheme, data retrieve strategy, detection of grid structure. At last, the rule and process that reorganize the original data to unified grid system have been presented, and a rapid numerical transformation model-a linear rule approximation model (LRA-model) have been proposed for coordinate transformation between the WGS 84 Geographic Coordinate System and the WGS 84 UTM Plane Coordinate System. Furthermore, the information variation caused by data reorganization process has been tested by using percentage histogram and comentropy.(3) Design data automatic processing flow according to characteristics of GF-1 LIA data, and implement GF-1 LIA data automatic processing framework with well fault-tolerant ability, robustness and execution efficiency. In the aspect of radiation correction, we develop automatic radiometric calibration and atmospheric correction application. In the aspect of orthorectification, we combine digital elevation model (DEM) extraction with RPC model, design and implement automatic orthorectification process, and research the grid based DEM data extraction algorithm and data partitioning strategy. In the aspect of geometric rectification, we compare the correction accuracy, computational efficiency and stability of different automatic geometric correction algorithm with different terrian condition and data phase, and thereby explore proper geometric rectification strategy.(4) On the basis of infrastructure architecture, using GF-1 WFV remote sensing data as representative instance, design and implement farmland scale natural disaster remote sensing monitoring prototype system, applying RDCRMG based Natural disaster remote sensing monitoring model on MPDI/PDI model based drought monitoring to verify its computational efficiency and feasibility.