基于Sentinel卫星的浑浊水体叶绿素反演对比研究——以鄱阳湖为例

Comparison study on the retrieval of chlorophyll in turbid waters based on Sentinel satellites——a case study of Poyang Lake

Sentinel-2A卫星、Sentinel-3A卫星分别于2015年6月和2016年2月成功发射,其上搭载的MSI、OLCI传感器的空间分辨率、时间分辨率、波段设置等在内陆水体水环境遥感研究中具有较大的应用潜力.针对浑浊水体叶绿素的反演难题,以鄱阳湖为例,基于光学分区理论并结合同步实测数据,探讨了Sentinel系列卫星数据在湖泊叶绿素a遥感反演的可行性.研究表明:1)对于Sentinel-2A MSI数据,鄱阳湖北湖区以[1/Rrs(665)-1/Rrs(705)]*Rrs(740)作为反演因子构建的三波段模型拟合效果最好,决定系数R2是0.65,平均相对误差是38.53%;鄱阳湖南湖区差值模型Rrs(705)-Rrs(665)反演结果最好,R2是0.63,相对误差是39.87%.2)对于Sentinel-3A OLCI数据,鄱阳湖北湖区以[1/Rrs(665)-1/Rrs(673.75)]*Rrs(753.75)作为反演因子构建的三波段模型拟合效果最好,R2为0.65,平均相对误差为37.6%;鄱阳湖南湖区差值模型Rrs(708.75)-Rrs(665)反演结果最好,R2是0.62,平均相对误差为39.6%.3)Sentinel系列卫星的分区模型能在一定程度上解决鄱阳湖部分浑浊水体区域叶绿素反演不成功的问题,后续将研究更高精度的反演模型方法.

Sentinel-2A satellite and Sentinel-3A satellite were successfully launched in June 2015 and February 2016 respectively. The spatial resolution, temporal resolution and band settings of MSI and OLCI sensors have great potential in inland water environment research. In this study, the inversion problem of chlorophyll in turbid water was studied. Based on the theory of optical partitioning and the simultaneous in situ data, taking Poyang Lake as an example, the feasibility and potential of applying data from Sentinel system satellites in chlorophyll retrieval are discussed. The results show that： 1） as for Sentinel-2A MSI data, the three bands （[1/Rrs （665）-1/Rrs （705）]＊Rrs （740）） model is the most accurate one for northern lake with the determination coefficient of 0.65 and MRE of 38.53%. The difference （（Rrs （705）-Rrs （665）） model is the most accurate one for southern lake with the determination coefficient of 0.63 and MRE of 39.87%. 2） As for Sentinel-3A OLCI data, the three bands （[1/Rrs （665）-1/Rrs （673.75）]＊Rrs （753.75）） model is the most accurate one for northern lake, the determination coefficient of 0.65 and MRE of 37.6%. The difference （Rrs （708.75）-Rrs （665）） model is the most accurate one for southern lake with the determination coefficient of 0.62 and MRE of 39.6%. 3） The optical partition model of the Sentinel family satellites is able to solve the problems on chlorophyll inversion of some turbid water in Poyang Lake to a certain extent, and more precision inversion model method will be developed in the follow-up time.