作物种植面积和产量统计数据降尺度空间表达及时空变化分析

High-precision and downscaled spatializaton information of agricultural statistics such as crop area, yield and production are very significant to the researches on climate change, food security, etc. In this project, based on agricultural statistics at county level, methods of downscaled spatialization of agricultural statistics such as crop area, crop yield and production would be studied depending on the technologies of remote sensing, GIS, crop simulation model and optimization algorithm. The study region was in Huanghuaihai Plain which is the most important grain producing region and the research crop was winter wheat.Firstly, based on the model of extracting crop distribution using the time series remote sensing data of NDVI and the model of spatial crop yield simulation through assimilating remote sensing data such as LAI into crop growth model, regarded the statistics of crop acreage and yield in each county as the control parameter respectively, the results of downscaled spatialization of crop area statistic and crop yield statistic were gotten with high precision at regional level and grid level in each county supported by global optimization algorithm SCE-UA and cost function.Then, in order to improve the new method of downscaled spatializaton information of agricultural statistics and develop an optimized technical solution with high precision and low uncertainty, the uncertainty analysis of the impact on downscaled spatializaton of agricultural statistics and evaluation of some critical process helping to improve the accuracy of the final downscaled spatializaton results will be done such as the impact of the fluctuations and error of agricultural statistical data, the smoothing method of remote sensing time series data, spatio-temporal scales of assimilation variables and spatial grid scale of downscaled spatializaton. Thirdly, based on the spatialization results of crop yield statistic and crop area statistic, spatialization result of crop production statistic would be gotten through zonal statistics using standard grid net. At last, based on the above downscaled spatitialization results of crop area, crop yield and production of different period in the last 13 years from 2001 to 2013 in Huanghuaihai Plain, spatio-temporal variations of crop area and crop yield would be analyzed using the indices of Moran's I and aggregation. The model for the center of gravity of grain-output was built and the change of grain-output pattern was analyzed from 2001 to 2013 with the model. In short, this research would provide a new method and novel idea for the downscaled spatialization of agricultural statistical data which would promote development of interdisciplinary fields such as global climate change, land use/land cover, etc.

高精度作物面积和产量统计数据降尺度空间化信息对气候变化、粮食安全等研究具有重要意义。项目以我国黄淮海为研究区，以冬小麦为研究对象，运用遥感、GIS、作物模拟和优化算法等开展作物面积、产量统计数据降尺度空间化表达方法研究。在利用时序NDVI构建作物面积空间分布模型、遥感同化生长模型构建作物单产空间模拟模型基础上，以作物面积和单产统计数据为控制参数，在全局优化算法和代价函数支持下，实现区域精度和网格精度双优的作物面积、单产统计数据高精度空间化表达；然后对空间化结果具有不确定性影响和利于提高精度的过程进行评价分析，形成精度较高、不确定性降低的统计数据空间化优化方案；最后获取黄淮海过去13年（2001-2013）不同时期统计数据空间化信息，科学阐明区域作物面积/单产时空变化特征及总产重心时空格局演进过程。本项目为农业统计数据降尺度空间化提供了新方法和新思路，将促进全球变化、土地利用等交叉学科发展。

高精度作物面积和产量统计数据降尺度空间化信息获取对气候变化、粮食安全等具有重要意义。项目以黄淮海地区为研究区，以河北衡水为典型试验区，以冬小麦为研究作物，运用遥感、GIS、作物模型、优化算法等开展作物面积、产量统计数据降尺度空间表达技术方法、不确定性分析、区域应用和时空特征分析研究。取得主要成果包括：（1）完成基础数据库构建，开展了典型试验区作物遥感同步试验，获取了大量研究数据；（2）完成作物面积统计数据降尺度空间表达技术方法研究和精度验证，创新提出基于时序NDVI和SCE-UA全局优化阈值模型的作物面积统计数据降尺度空间化方法，实现区域作物面积统计数据降尺度高精度表达；（3）完成作物单产统计数据降尺度空间化方法研究和精度验证，提出基于EPIC生长模型同化的作物单产统计数据降尺度模拟方法，构建了作物单产统计数据空间分布模拟模型和系统，实现区域单产统计数据降尺度高精度模拟；（4）在降尺度核心算法研究基础上，系统开展了作物面积、单产降尺度主要影响因素对精度影响和不确定性分析研究，为统计数据降尺度方案优化奠定坚实基础；（5）高精度获取了黄淮海过去13年（2001-2013）不同时期作物分布、作物产量空间化结果，实现作物面积、产量统计数据降尺度表达技术方法大范围应用；（6）系统分析黄淮海作物面积和产量时空特征，明确了作物面积、产量动态变化趋势和特征、经纬向分布规律、重心演进过程等；（7）项目发表论文14篇（含5篇SCI，6篇EI），其中第一资助在Remote Sensing of Environment、IEEE Geoscience and Remote Sensing Letters、遥感学报等发表8篇（含7篇SCI/EI），申请发明专利5项，软著1项，项目组出站博士后1名、培养硕士2名。项目成果为作物面积/产量基础农业科学数据获取提供了新方法和新思路，将促进农业遥感、全球变化和土地利用等领域发展。