【作者】 邝继双；

【导师】 汪懋华；

【作者基本信息】 中国农业大学 ， 农业电气化与自动化， 2003， 博士

【摘要】 本研究是在“现代精细农业系统集成研究教育部重点实验室(中国农业大学)精细农业研究中心”和“中国农业大学信息与电气工程学院农业电气化与自动化国家级重点学科”所提供的研究环境与物质保障下完成的。国家发展计划委员会与北京市政府批准立项的“高技术产业化示范工程项目北京精准农业示范工程(编号A00300100584)”为课题研究提供了经费支持和田间试验条件。 在精细农业的研究、试验与实施过程中涉及到大量关于农田环境与作物生长情况的数据。由于这些数据一般都具有空间分布上的特性，要分析与处理这些数据，普通的关系型数据库已不能满足需要，因此需要开发一种田间信息系统，或者称之为农田地理信息系统，来对田间数据进行分析与处理，并对农田信息进行管理。本课题根据我国发展精细农业的需要，在对空间信息处理技术研究的基础上，进行了农田地理信息系统的设计与开发等方面的研究工作，其中主要包括： 1．按照现代软件工程的管理方法，分析和研究了农田地理信息系统的开发模型与开发方式，提出了采用螺旋式模型，以通用高级编程语言结合地理信息系统组件的系统开发方案。 2．比较分析了农田电子地图的主要测绘技术，设计了GPS测绘中单点重复定位——求平均，以及连续定位——移动平均等定位数据的处理方法。 3．选用了7参数布尔莎模型进行WGS-1984空间直角坐标系与北京1954空间直角坐标系之间的转换；使用了高斯-克里格投影进行大地坐标系与平面直角坐标系之间的转换。 4．研究了应用遥感图像作为参考底图进行电子地图更新的方法。遥感图像的几何校正采用一个二次多项式模型实现，来消除平移、缩放、旋转、仿射与弯曲等图像变形；遥感图像的坐标配准采用参数文件的方法进行。 5．在对田间土壤采样方法研究与分析的基础上，设计了一种参考已有的农田土壤信息，来分析土壤参数的分布变化规律，并在农田中的关键位置设置采样点的指导采样方法。此方法通过空间搜索与判断，来了解土壤参数在农田中变化的剧烈程度，并以此作为确定指导采样点位置的依据。 6．在田间信息分布图生成与插值方法的研究方面，设计了用于生成产量分布图具有数据过滤功能的局部平均法；并选用了一种带有平滑参数IDW插值方法；该方法可以有效的防止插值结果中出现数值突然变化的奇异点。 7．研究了模糊综合评价方法与分层聚类分析在农田管理区划分中的应用。两种方法的分类结果虽然有些局部差别，但是从整体来看其结果基本一致。 8．提出了一种确定目标产量的方法。该方法首先分析作物产量的限制因素；然后确定在这一因素的制约下，田间不同小区的最大可能产量，并以此作为目标产量，绘制出目标产量分布图。 9．在冬小麦变量施肥处方的确定过程中，根据作物的生长特点与需肥规律，选用了一种平衡施肥模型。该模型考虑了目标产量、土壤养分含量、肥料的有效含量以及肥料的利用率，并且还能够根据耕地的不同产量水平与具体情况对施肥量进行调整。 在北京精准农业示范区，对设计开发的农田地理信息、系统进行了GPS测绘、坐标系转换、电 中国农业大学博士学位论文 摘要一子地图更新、土壤采样规划、田间信息分布图与作物产量分布图生成、农田管理区划分、目标产量图生成、变量施肥处方图生成以及基本信息管理等方面的测试与实际工作。结果表明系统运行稳定，能够作为田间数据处理与农田信息管理的一种工具，为精细农业的研究与实践提供支持。更多还原

【Abstract】 This research was carried out at the Key Laboratory for Modern Precision Agriculture System Integration Research of Ministry of Education, the Research Centre for Precision Agriculture of China Agricultural University, and supported by the National Development Project Committee and Beijing Government Project - Hi-tech Industrialization Demonstration, Beijing Precision Agriculture Demonstration Project (Project No.A00300100584).For site-specific crop management of precision agriculture, it is needed to manipulate mass and various data gathered from different data sources. Since common management information systems are not suitable to deal with these spatial distributed data, it is needed to develop a field level geographic information system (FIS), in order to meet the needs of field data processing and field information management. Driven by the needs for carrying out precision agriculture research, a field-level geographic information system was designed and developed, based on the study of spatial information processing techniques. The main results are listed as the following:1. According to the modern software engineering technology, the system develop models and fashions were analyzed and compared. It was designed using a high-level programming language (VB6) and a geographic information system component (MO2) to develop the system, in accordance with spiral model.2. In order to obtain the digital base map of the field, different mapping techniques were compared. Single point repeat positioning - average and continuously positioning - moving average methods were designed for GPS mapping.3. For geographic coordinates and Cartesian coordinates transformation a model that include 7 parameters and the Guess-Kruger projection were used. The difference of calculation between the above method and Trimble TGO software is less than 1cm.4. In order to keep the digital base map up to date, methods of using remote sensing images as the reference for map refreshment were studied. A 2-D multinomial geometric correction quadratic model was used to correct the geometric distortion of images, and a file contents 6 parameters was used for images coordinates register.5. For lower the soil sample cost, a directed soil sample method was designed. This method could set sample points only at key locations in the field, according to the analysis result of the referenced field information.6. For yield map generating, a local-averaging interpolation method with a random error yield data filter was designed; And for field information distribution map generating, an IDW interpolation method with a smoothing parameter was used.7. Fuzzy integrative evaluation and cluster analysis was used for field management zoning. The results of these two methods were similar.8. A method for calculating the yield goal was developed. The method takes the maximum possible yield of each location that limited by the yield-restricting factor as the yield goal, and generates the yield goal map accordingly.9. For calculating the variable fertilizing dose, a mathematics model was selected. The model takes account of the yield goal, soil fertility, fertilizer available ingredientcontent, and the fertilizer using efficient.In the Beijing Precision Agriculture Demo Farm, the system was used for base map mapping, field boundary map update, soil nutrition map generating, yield map generating, and field basic information management. The result shows that the system is steady and effective to do these jobs, and could be a tool for field data processing and field information management.更多还原