CAJ | 학술논문

鄱阳湖南矶湿地国家级自然保护区内时常有烧荒事件发生，这对保护区生物多样性和区域生态环境造成一定影响，快速、便捷、准确地获取烧荒地信息是该区域环境管理的紧迫需求。本文针对2012年1月初鄱阳湖南矶山湿地发生的烧荒事件，采用环境与灾害监测预报小卫星影像数据，以鄱阳湖南矶湿地国家级自然保护区所在的新建县为研究区。以面向对象的遥感影像分析方法，通过对影像的多尺度分割，根据影像的光谱信息、空间特征、纹理特征等，构建烧荒地提取的规则集，对影像中的烧荒地信息进行提取，并利用混淆矩阵对信息提取结果进行精度检验。精度检验结果表明，烧荒信息提取的总体精度为99.38%，Kappa系数为0.89，证明面向对象的遥感信息提取方法可以有效、快速地提取烧荒地信息，避免了传统方法中的椒盐状破碎图斑的现象。
파양호남기습지국가급자연보호구내시상유소황사건발생，저대보호구생물다양성화구역생태배경조성일정영향，쾌속、편첩、준학지획취소황지신식시해구역배경관리적긴박수구。본문침대2012년1월초파양호남기산습지발생적소황사건，채용배경여재해감측예보소위성영상수거，이파양호남기습지국가급자연보호구소재적신건현위연구구。이면향대상적요감영상분석방법，통과대영상적다척도분할，근거영상적광보신식、공간특정、문리특정등，구건소황지제취적규칙집，대영상중적소황지신식진행제취，병이용혼효구진대신식제취결과진행정도검험。정도검험결과표명，소황신식제취적총체정도위99.38%，Kappa계수위0.89，증명면향대상적요감신식제취방법가이유효、쾌속지제취소황지신식，피면료전통방법중적초염상파쇄도반적현상。
Moorburn was occurred frequently in Poyang Lake Nanji Wetland National Nature Reserve, which af-fected biodiversity and regional ecological environment to some extend. The traditional moorburn information extraction methods have their respective advantages, but it is not suitable for rapid and emergency moorburn monitoring. So it is an urgent demand to extract moorburn information for environmental management in Poyang Lake Nature Reserve in a quick, convenient and accurate way. This paper focused on the moorburn incident in Poyang Nanji Nature Reserve Area of Poyang Lake located. The environmental mitigation satellite data were ad-opted in Lake Nanji wetland in early January 2012, taking Xinjian County as the study area. The research was based on object-oriented remote sensing image analysis theory. Firstly, the image was multi-scale segmented and merged. The scale of the image segmentation and merging of 35 and 80 respectively. Then, the rule set of moor-burn extraction was built based on the spectral information, spatial feature, and texture feature of image. The spectral information rules were built based on Hue, NDVI, MTVI and the principal components after the princi-pal component analysis. The space characteristic rules were built based on area and roundness. The rules of tex-ture feature were built based on texture mean and texture entropy. Finally, the moorburn information was extract-ed, and the extraction result is tested with confusion matrix. Accuracy test shows that overall accuracy is 99.38%, and Kappa is 0.89. It demonstrates that the object-oriented method is a kind of effective, accurate and fast method for the extraction of moorburn information, and it can avoid salt and pepper broken figure spot in tra-ditional methods. The difference of moorburn information and other features is bigger in the first principal com-ponent after the PCA transformation in spectral characteristics and the texture entropy of texture features. But the difference is very small in space characteristics. The method can provide data support and technical reference for environmental protection, nature reserve management and moorburn management in the Poyang Lake area.