遥感图像对江滩型血吸虫病疫区分类研究

Study on unsupervised classification in marshland of endemic area of Schistosomiasis Japonica using satellite TM image data

目的 利用卫星专题制图仪 (thematicmapper,TM)遥感图像对安徽贵池流坡村江滩型日本血吸虫病疫区洲滩植被进行分类。方法 通过主成分分析 ,确定用于分类的TM3、TM4和TM5等3个波段遥感数据 ,然后在遥感图像处理软件支持下对卫星TM图像进行假彩色合成及非监督分类 ,最后通过现场调查确定不同类别的地物属性 ,并将分类结果进一步合并、整理。结果 通过非监督分类和实地勘察 ,最终共划分出 5个地表类别 ,其中意杨树林、芦苇和草滩均为钉螺孳生地。根据 1998年春季的查螺结果 ,研究现场的活螺密度和感染螺密度分别为 0 5 10 /0 11m2 和 0 0 0 7/0 11m2 。结论非监督分类是计算机分析遥感图像的方法之一 ,有助于对与生态有关疾病的环境进行分类 ,为流行病学的研究提供了新的方法。

Objective To categorize the vegetation of marshland in the endemic areas of Schistosomiasis Japonica using satellite thematic mapper (TM）data in Liupo Village, Anhui Province. Methods Data of three satellite TM bands, namely TM3, TM4 and TM5, were selected to be used in classification with principal component analysis. Then, satellite TM images were applied to false color compo site and unsupervised classification with computer software to manage the images . Finally, the resulting clusters were identified and reclassified by site visit s to categorize the land coverage and vegetation. Results Five classifications o f land coverage were generated, including poplar forest, bulrush and marshland, all with snail habitants. Based on the survey in the spring of 1998, the overal l live snail density and infested snail density in the study areas were 0.510/ 0 11 m 2 and 0 007/0 11 m 2, respectively. Conclusion Unsupervised classification, which is a technique for the interpretation of remotely sensed imagery, can contribute to the classification of environment with snail habitants and ecological diseases and become a new method in epidemiological studies.