【作者】 吴曙雯；

【导师】 王人潮； 程家安；

【作者基本信息】 浙江大学 ， 遥感与信息技术， 2002， 博士

【摘要】 南方稻区是我国水稻主产区，近20年来水稻褐飞虱的发生、危害严重地威胁着该地区的水稻生产，因此及时、准确地预测褐飞虱的发生，不仅对于种植区水稻的正常发展，而且对区域农村经济可持续发展都具有重要意义。 传统的褐飞虱灾变预测方法主要是针对某一个县市进行的，缺乏空间分析因子。本研究以中国南方稻区为研究区域，采用地统计学和地理信息系统(GIS)方法，通过分析1981-1987年的虫情资料，模拟出褐飞虱越冬北界，分析其灾变发生程度的空间分布状况，并建立模拟模型。在此基础之上，设计开发出基于Web-GIS和ASP的中国南方稻区褐飞虱灾变预警系统，并初步应用到浙江地区。此系统是一个集成网络地理信息系统、模拟模型、网络数据库、人工智能和网络多媒体技术的网络系统。此外还初步研究了稻叶瘟对水稻冠层光谱特性的影响。 本研究主要内容如下： 1．中国南方稻区褐飞虱地理信息系统数据库的组建 采用Arc View 3.2和FoxPro 6.0软件组建1981-1987年中国南方稻区褐飞虱地理信息系统数据库，内容包括褐飞虱种群动态空间数据库和属性数据库的构建、发生程度、气象因子。此数据库具有可扩充性，为进一步分析大尺度南方稻区褐飞虱灾变发生的时空动态变化提供基础数据。 2．基于GIS的中国数字地形模型建立： 根据GIS数字地形模型(DTM)的构造原理和方法，用离散点移动拟合距离加权平均插值方法生成中国数字高程模型(DEM)。以DEM为基础，用离散点插值数值拟合方法构造了DTM的数学模型。用该模型从DEM中提取了中国不同地貌的地面单元平均高程、相对高程、高差、高程变异、坡度、坡向、地面粗糙度、坡面形态、谷脊特征线、沟谷密度、沟谷深度等各种地形结构线，由DEM产生了三维立体地形模型和各种地形剖面，以及地面辐照度的计算和地形阴阳坡的划分等，为全国一月均温的模拟和褐飞虱越冬北界的划定提供模型基础。 3．基于GIS的褐飞虱越冬北界的划定与模拟： 以地统计学的变差函数理论和 Kropng插值方法为工具，结合气候统计学，探讨地统计学在气象因于和褐飞虱越冬北界的划定与模拟上的应用途径和方法，与历史研究成果进行比较，在地理位置和总体趋势上具有一致性。 4．褐飞虱灾变发生程度的空间分布格局与动态： 以GIS和地统计学为工具，以1981－1987年褐飞虱发生程度为研究对象，分别从不同年份内和年份间的发生程度方面，在时空角度上分析褐飞虱空间分布及发生程度的规律性，发现大尺度中国南方稻区褐飞虱发生程度的主间分布类型呈规则分布，但发生程度均存在一定的空问相关性，主间相关强弱差异较大，且发生程度均存在不等的随机性变异。198卜1987年的褐飞虱发生程度均表现为由西南向东北方向发生程度增加的趋势，发生程度大的区域主要集中在福建、浙江、湖南和）－西。褐飞虱的发生程度、所在区域、方向趋势在同一年内及不同年份间均存在差别。褐飞虱发生程度在1981－1987年的基本分布格局表现为福建、浙江、湖南、广西东部发生偏重；四川、重庆、云南、贵州发生偏轻；厂 东、海南、江西、安徽、湖北、江苏、上海为中等发生。7年间的发生趋势，即198年至1983年发生程度趋于加重，但1984年却骤减，1985、1986年发生程度又加重、1987年出现严重的趋势。最终建立褐飞虱发生程度空间半方差变异函数模拟模型。 5．基于摸糊匹配的水稻病虫害诊断专家系统的设计与实现： 针对症状诊断知识的特点，提出了一种采用模糊匹配获取知识的方法未设计水稻病虫害诊断专家系统。在ASP网络服务器端脚本语言下实现，应用于基于Internet中国南方稻区褐飞虱灾变预警系统中，能通过模糊匹配产生实用的诊断规则，对缺少诊断知识的病虫害症状进行尝试性推理，达得从模糊到精确的诊断目的。 6．基于Internet的中国南方稻区褐飞虱灾变预警系统的组建： 以Internet为系统平台，以网络地理信息系统、网络数据库、人工智能、网络多媒体和动画制作为技术支撑的中国南方稻区褐飞虱灾变预警系统，具有时空动态分析、预警发布、预测预报、数据传输、病虫诊断、辅助决策和远程教育等功能，界面友好。 IX 7．稻叶瘟对水稻冠层光谱特异性影响的探讨： 水稻感染稻叶瘟后，绿光区、红光区和近红外区的水稻冠层光谱反射率随病情程度的加重分别呈现下降、上升和下降的趋势；绿光吸收边缘的特征波长值发生红移，红光吸收边缘和近红外吸收边缘的特征波长值发生蓝移。受害轻时近红外区反射率变化幅度大，受害重时绿光区和红光区反射率变化幅度大。研究为应用遥感技术早期探测重大病虫害的发生提供了实验依据。 本研究的技术发展和创新点有以下几个方面：，提出应用引 S技术模拟农作物病虫害越冬区域：利用 GIS技术和 气候统计学方法并结合农业气象学和病虫害及寄主的生物学特性能够预 测当年病虫害的越冬环境和地理分布区域，为中长期预测病虫害灾变发 生提供快更多还原

【Abstract】 China South is one of the main regions for rice production in China, where in recent 20 years Brown Planthopper (Nilaparvata lugens (Stal)) (BPH) damaged the rice production seriously. Therefore, the timely and accurate prediction of BPH occurrence is important not only to the rice normal development of the planting region, but also to the sustainable development of regional rural economy.The traditional method used for prediction of BPH disaster is only aimed at one county and lack of spatial factors. Geostatistics methods and GIS were used to simulated the northern boundary of BPH overwintering sites, analyze spatial distribution of BPH disaster levels and built its dynamic model in the whole region from 1981 to 1987 in this study. As a result, China South BPH Disaster Early-warning System based on Web-GIS and Active Sever Page, as well as its preliminary application in Zhejiang region, were designed and developed. The web was a internet software system, which integrated GIS, simulation model, expert system and multimedia technology. Besides, effects of rice leaf blast on spectrum reflectance of rice was studied simply.The main contents of this study are as follows:1. Establishment of GIS database for BPH in China SouthArc View 3.2 GIS and FoxPro 6.0 were used to develop database for BPH. The database included spatial and attribute database. The spatial database included China’s boundary map, river map, road map, elevation map, soil category map, etc. and the attributes database included BPH disaster levels, meteorologic phenomena. The GIS database provided primary data for BPH spatial analysis.2. Establishment of China digital terrain model (DTM) based on GISUsing the principles and approaches of building DTM based on GIS, China Digital Elevation Model (DEM) was generated with the distance weighted average interpolation of shifting imitation for dispersed elevation points. On the basis of the DEM, mathematical models for DTM were built with numerical value interpolating imitation for dispersed elevation points. The average elevation of ground135unit, relative elevation, elevation difference, elevation variation, slope, aspect, ground roughness, ground formation, gully feature lines and hill ridge, valley density, gully depth, and others were extracted. Moreover, 3D model of China topographic map with shade slope and sunny slope was structured. The result was applied to the simulation of January average temperature in China and the computation of northern boundary of BPH overwintering sites.3. Definement and simulation of northern boundary of BPH overwintering sites based on CIS: As a tool of Geostatistical theory and Kriging interpolation method, the simulation ofineteorologic factors and northern boundary of BPH overwintering sites were discussed with climatic statistics. The results are consistent with historical research on the whole.4. Analysis on BPH disaster spatial distribution and its dynamic models:Statistical theory and GIS method were used to analyze BPH disaster level spatial distribution and its dynamics with inner-years and inter-years in China South from 1981 to 1987. The result of traditional statistics showed that the distribution pattern of BPH disaster level is regular on large scale, and the result of Geostatistics indicated that BPH disaster level was correlated to the spatial factor., but the range changed greatly in 1981-1987. The trend of BPH disaster is serious with the direction from south-west to north-east. The BPH high risk regions located mainly Fujian, zhejiang, Hunan and Guangxi, and lower risk ones lied in Sichuan, Chongqing, yunnan, Guizhou, and others were middle risk. The disaster level, regional distribution, directional trend of BPH were different in inner-years and inter-years. The disaster trend was that the BPH risk increased from 1981 to 1983, decreased greatly in 1984, aggravated in 1985 and 1986, and became serious in 1987. The spatial semivariance models of BPH were built on the basis of these result更多还原