摘要
以南宁市为研究区,利用手持GPS实地调查取样,获得甘蔗及其它作物训练样本区。通过分析甘蔗等作物年度内光谱特征变化规律,依据不同作物物候历不同的原理,利用多时相HJ-1CCD卫星数据,采用监督分类、决策树分类和逐步剔除相结合的遥感方法,提取研究区甘蔗种植信息。野外验证和分类精度分析表明,无论从空间分布趋势还是面积数据来看,其信息精确度均较高,分类精度达到92.3%。由此可见,环境减灾卫星数据结合分类方法能基本满足制糖企业对甘蔗种植面积提取精度的要求,可以在制糖企业中推广应用。
The training area of sugarcane and other crops was surveyed by using handheld GPS in Nanning. The infor- mation of sugarcane planting was extracted by analyzing the variation regularity of spectral characteristics of sugarcane, by using remote sensing methods combining supervised classification, gradual elimination and decision tree classifica- tion, based on the multi-temporal HJ - 1 CCD satellite data in training area. The results showed that the classification precision reached 92.3% both from space distribution trend or area data, the environmental mitigation satellite data （ HJ - 1 ） and the classification method could meet the requirements of precision of sugarcane plant area for sugar com- panies, which could be applied to the sugar companies and other research.
出处
《中国农业气象》
CSCD
北大核心
2012年第2期265-270,共6页
Chinese Journal of Agrometeorology
基金
南宁市科技攻关项目(201002041B)
国家科技支撑计划课题(2008BADB8B01)
南宁糖业有限责任公司服务项目
关键词
HJ-1卫星
甘蔗
面积调查
HJ- 1 satellite
Sugarcane
Area investigation
