摘要
通过对2010年5月2日太湖HJ-1A卫星超光谱影像的几何纠正和6S模型辐射校正,以及水体实测光谱数据和影像光谱数据分析,将太湖28个水体采样点光谱数据分别进行归一化处理和一阶微分处理后,选取和水质参数相关系数最大的波段或波段组合建立反演模型,获得太湖叶绿素a浓度以及悬浮物浓度的空间分布图.研究表明,超光谱影像B73波段(682.785nm)处的归一化光谱数据和叶绿素a浓度相关性最高,泥沙遥感参数(Sr)和悬浮物浓度相关性最高,与水体实测光谱数据的相关性分析结果相同,模型预测值和实测值的平均相对误差均在30%之内,水质空间分布图与实地调查结果一致,因此,HJ-1A卫星超光谱数据可以借鉴水体实测光谱数据不同水质参数敏感波段的分析结果,很好地应用于水质定量遥感.
Based on geometry correction using ERDAS software and radiation correction using 6S model for HJ-1A hyper-spectrum image(HSI) on May 2 in 2010 and the analysis of spectrum for water data and spectral data of hyper-spectrum image,this paper processes original spectrum data of 28 sample points using method of normalization and method of first-order derivation.Single-band and band combination are selected to establish inversion models of the concentration of chlorophyll-a and solid suspensions.Choosing the model with biggest correlation coefficient,the spatial distribution map of the concentration of chlorophyll-a and solid suspensions content in Taihu Lake is acquired.The research results show:Band-73 of hyper-spectrum image which has been normalized shows the biggest correlation coefficient of the concentration of chlorophyll-a,remote sensing sediment parameter shows the biggest correlation coefficient of the concentration of solid suspensions,the result is consistent with analysis of spectral data of hyper-spectrum image.Average relative errors of predicted and measured values are within 30 percent.Spatial distribution map of water quality is consistent with the result of field surveys.Therefore,based on reference of the analysis of sensitive band of spectrum for water data,HJ-1A hyper-spectrum image can give quantitative estimation of water quality parameters in Taihu Lake.
出处
《环境科学》
EI
CAS
CSCD
北大核心
2011年第11期3207-3214,共8页
Environmental Science
基金
国家水体污染控制与治理科技重大专项(2008ZX07528-005)
"十一五"国家科技支撑计划重点项目(2008BAC34B07-03
2008BAC34B01-2)
中央级公益性科研院所基本科研业务专项
关键词
HJ-1A卫星
超光谱数据
反演模型
叶绿素A浓度
悬浮物浓度
定量遥感
HJ-1A
hyper-spectrum image(HSI)
inversion models
chlorophyll-a concentration
solid suspensions concentration
quantitative estimate