【作者】 张治英；

【导师】 徐德忠；

【作者基本信息】 中国人民解放军第四军医大学 ， 流行病与卫生统计学， 2003， 博士

【摘要】 日本血吸虫病是我国长江以南地区重要的公共卫生问题，对人群健康造成极大的威胁，日本血吸虫病的分布与其中间宿主钉螺的分布范围高度一致，呈明显的地方性，而钉螺与其孳生微环境的自然因素密切相关，因此研究环境因素与钉螺分布的关系已成为钉螺控制及血吸虫病预防研究的重要内容之一。遥感资料由于能够快速、准确、周期性地提供地面环境监测资料，为环境因素与血吸虫病发生及钉螺分布关系的研究提供了新的工具，本研究在2000年调查的江宁县钉螺分布资料的基础上，利用GIS及其空间分析技术研究了遥感资料在江宁县江滩钉螺分布预测及其孳生地监测中的应用。 根据2000年调查显示，江宁县钉螺孳生地有江滩与山区型两种，其中江滩钉螺孳生地面积为1079.28万m~2，占江宁县钉螺孳生地总面积的98.27％，且江滩地区的平均活螺密度明显高于山区(江滩为1.69只/框，山区为0.827只/框)，因而江滩是江宁县钉螺控制及血吸虫病预防的重点地区。 我们以江宁县1：5万数字化要素图为结构数据库，收集的2000年江宁县的螺情资料为属性数据库，在ArcView8.1软件的支持下建立了江宁县钉螺孳生地分布的地理信息系统，结果显示江宁县钉螺孳生地在空间中的分布是不均一的，呈区域性分布；进一步空间分析发现在江滩有2个、山区有4个活螺的高聚集区，其平均活螺密度明显高于周围地区(p＜0.0001)，说明这些地区有适于钉螺孳生繁殖的某些因素存在；同时我们利用变异函数对江宁县江滩钉螺的空间分布特征进行了描述，显示江宁县江滩钉螺在空间中的分布存 第四皿吕大琢傅士学位防不一讪耍一在自相关性，结果表明当空间距离小于0刀301时，活螺密度的分布变化与距离有关，且其变异可以由变异函数八）来描述，建立的变异函数方程为： 厂 0（h＝0） l，L’L3 I。＿－－．3h lh“ y（）叫 14石61X（三一一二一了一三一二一了（0＜h＜0．0301）” I“”——””2 n。，。12 2 n。。nl3” L14石61（》＞0刀301） 为了分析江宁县江滩环境因素与活螺分布的关系，我们从 Landsat ETM＋图像中提取江宁县江滩钉螺孽生地环境因素的遥感替代指标，并用逐步多元回归分析研究了其与江滩钉螺分布的关系结果发现，江宁县江滩活螺密度与其孽生环境的修正土壤调整植被指数（MSAVI）跟m及缨帽转换湿度指数（WetllllSS）有关，且存在Y—2．481＋3二19MSAVI－9．143WetnCSS－0．26iT，其中h为活螺密度的平方根；分析显示该回归方程的决定系数d－0．2820＜0刀001＼决定系数偏小，认为由以上3个环境因素指标估计江宁县江滩钉螺的分布存在较大的残差，可能有重要因素没有考虑入回归方程中因此我们在回归分析的基础上，结合回归残差的空间分布特征，利用地统计学普通克立格法 （Ordin聊 Kropng）建立了回归残差的空间分布预测图（Y＊，并将其与回归方程（Y；）一起用于预测江宁县江滩钉螺的空间分布，其模型为Y’－Y；＋Y。（其中Y’为活螺密度的平方根＼ 分析发现该模型的决定系数R乙0．851 （P。0刀00），较回归方程有明显提高；进一步用该模型预测江宁县江滩地区的活螺密度并与实际调查结果进行比较，计算其符合率，结果显示其Kappa值为63％，可见二者基本符合，说明以该模型为基础，利用遥感资料能有效预测江宁县江滩钉螺的分布状况。 同时，根据既往研究结果及钉螺孽生分布的特点，即有钉螺的地方必有植被，没有植被的地方没有钉螺分布，且一定的植被对钉螺分布有指示作用，因此我们进一步探讨了遥感图像在江宁县江滩植被监测中的应用洲门对以Landsat ETM＋图像组成的 4种特征图像进行非监督分类，经过现场勘察及分 2 第四罩否大鼻俗士学位沂文一纫三一离度评价，结果显示以 ETM的 CHZ 3 4波段组成的 ETM234伪彩色复合图像滤掉无植被覆盖像元后的分类识别效果最好，能有效地将地表主要的植被景观分托因此我们将ETM234伪彩色复合图像去掉无植被覆盖像元后非监督分类的结果与江宁县江滩钉螺孽生地分布矢量图重叠，提取各孽生地地表的主要植被特征，分析发现江宁县江滩钉螺革生地的主要地表植被为芦苇梭叶草杂草和树林，其中以芦苇及梭叶草为优势植被的钉螺孽生地占江宁县江滩钉螺孽生地的65％，且其活螺密度明显高于以杂草及树林为优势植被的钉螺孽生地（P＜0刀5），因此认为通过遥感图像分类识别地表植被类型可用于预测江宁县江滩钉螺擎生地的分布状况。 然而由于应用遥感资料预测江宁县江滩钉螺分布是一个复杂的过程，基层血防工作者不是遥感领域的专家，且不是专业的软件操作员，因此本研究在ERDASS．5软件的支持下，以建立的预测模型为基础利用决策树的原理和方法建立了以遥感资料定性监测江宁县江滩钉螺荤生地分布及定量预测荤生地钉螺?更多还原

【Abstract】 Schistosomiasis due to schistosoma japonicum (S. Japonicum) is a severe health problem along and down to the basin of the Yangtze River in China. Its distribution corresponds with that of the intermediate host Oncomcelania snails, which distribute in endemic. For the survival of snails relate closely to the environmental factors of habitats, the study on the relationships between the environmental factors and the distribution of snails are important for the prevention of schistosomiasis and control of snails. Imagery from satellite remote sensing is a digital database of environmental factors with excellent temporal and spatial references, so it becomes a useful tool for the prediction of alive-snails and surveillance of snail habitats. Based on the epidemiological investigation in 2000, this study was to explore the application of remote sensing (RS) to the prediction of the snails in marshland in Jiangning County using the geographic information system (GIS) and spatial analysis.The investigation showed that snails breed both in marshland and mountainous regions in Jiangning County in 2000. The areas of snail habitats in marshlands are about 10.7928 million square meters, which accounted about 98.27% of that of the total snail habitats in Jiangning county. And the average density of alive-snails in habitats of marshlands is about 1.68 per pixels, which are higher than that of mountainous regions. So we can draw our conclusion that theemphasis for the snail control and the prevention of schistosomiasis in Jiangning County should be put on the marshlands.The GIS established from the digitized topo-sheet of 1:50 000 and the co-ordinates of the geographical centroid for each snail-breeding site in 2000 were subjected to analyze the distribution characters of snail habitats in spatial in Jiangning County using ARC VIEW 8.1. It demonstrated that the snail habitats do not distribute randomly and there are some regions that had more snail habitats than others. The spatial scan statistics in further detected 2 spatial aggregations for alive-snails in marshlands and 4 in mountainous regions (P<0.001) in which the density of alive-snail were higher significantly than that outside these areas with P<0.001. This indicated that there are some factors in these aggregation areas favorable for the snail survival. Modeling the semi-variogram of the alive-snails in marshlands depicted the spatial autocorrelation of the alive-snails with rang about 0.0301, which means that the samples separated by distances closer than the range are spatially auto-correlated and the variogram of the alive-snails bydistance could be estimated by the formula ( h ) described asWhereas the samples separated by a distance greater than the range are spatially uncorrelated.The multi-variation regression analysis demonstrated that the density of alive-snails in marshland in Jiangning County related to the environmental factors of snail habitats derived from the imagery of Landsat ETM+. The regression formula showed as Y1=2.481+3.219MSAVI-9.143Wetness-0.261T, where Y1represented the square-root of alive-snails in habitats of marshland and MASVI, Wetness and T represented the environmental proxies derived from Landsat ETM+ imagery for vegetation index, land surface moisture and land surface temperature respectively. And the determination coefficient of the regression function is about 0.282 with P <0.0001. This indicated that there were great residuals using the regression function to estimate the density of alive-snails in marshland and some important determinants had not been included in the model. So we analyzed the spatial characters of the regression residuals in semi-variogram and established prediction model (Y2) to estimate the residuals regression function using the ordinary kriging. Then we used both the regression model for the density of alive-snails and the prediction model for the regression residuals to estimate the distribution of alive-snails. The last model for theprediction of alive-snails in marshlan更多还原