基于MODIS数据提取广西甘蔗信息技术初步研究  

A Primary Study of MODIS-Based Information Recognition Technology for Sugarcane in Guangxi

作　　者：丁美花[1] 谭宗琨[1] 熊文兵[2] 匡昭敏[1] 何燕 杨鑫[1] 莫伟华[1] 欧钊荣[1] 

机构地区：[1]广西气象减灾研究所,南宁530022 [2]广西气象科技服务中心,南宁530022

出　　处：《西南大学学报（自然科学版）》2008年第9期94-100,共7页Journal of Southwest University(Natural Science Edition)

基　　金：广西科学基金资助项目(桂科基0778006-11);国家科技支撑计划课题资助项目(2008BADB8B01);“十一五”国家科技支撑计划重点资助项目(2006BAD04B03);中国气象局气象新技术推广资助项目(CMATG2006M42);广西自然科学基金项目(桂科自0640053)

摘　　要：甘蔗种植空间分布及面积估算是重要的农情信息,能够为政府决策、糖业企业经营规模提供科学、客观的技术支持.利用MODIS-NDVI年度内的变化规律,依据不同作物物候历不同的原理,并根据训练样本区的甘蔗光谱特性,在全区或分区范围内采用监督分类和逐步剔除相结合的方法,将广西各地甘蔗种植信息提取出来,同时进行了面积估算.通过对连片区域和一些特异点的野外验证,发现结果是非常准确的,误差在15%以内.因此本文的结果具有科学性,初步解决了"同物异谱"和"异物同谱"所造成的信息提取不准确的问题.利用多时相EOS/MODIS遥感数据提取的甘蔗种植空间分布信息与甘蔗实际种植空间分布是基本相符的.Using the annual variation patterns of MODIS-NDVI （moderate-resolution imaging spectroradiometer-Normalized Difference Vegetation Index）, according to the principle that different crops have different phenology calendars and to the spectral characteristics of sugarcane in the training samples, the authors of the present paper adopted the method of supervised classification and gradual elimination to extract information about sugarcane cultivation in Guangxi and estimated the size of its area. Field investigation of some differential spots and regions showed that the above-mentioned MODIS-based information recognition was accurate, with an error of ≤15%. The spatial distribution information of sugarcane given out by the remote sensing recognition method using the time series of EOS/MODIS data is basically consistent with actual distribution.

关 键 词：MODIS NDVI 监督分类 甘蔗 

分 类 号：V557.3[航空宇航科学与技术—人机与环境工程]