适用于多种卫星数据的太湖水体漫衰减系数估算算法

下行漫衰减系数(K_d)是描述水下光场的重要参数,决定水体真光层深度,影响着浮游藻类初级生产力及其分布特征.基于2008—2013年太湖4次大规模野外试验数据,分析太湖水体漫衰减系数特征及其影响因素,建立适用于多种卫星数据且较高精度的太湖水体490 nm处下行漫衰减系数估算模型.结果表明:无机悬浮物是太湖水体漫衰减系数的主要影响因素;红绿波段比值(674 nm/555 nm)最适合于太湖K_d(490)估算,模型反演精度较高(N=72,R~2=0.72,RMSE=0.89 m~(-1),MAPE=21.58%);利用实测光谱数据,模拟得到MODIS/EOS、OLCI/Sentinel-3、GOCI/COMS和MSI/Sentinel-2等主要传感器波段的信号,构建适用于多种卫星传感器K_d(490)估算的红绿波段模型,建模精度较高(N=72,R~20.7,RMSE0.9 m~(-1),MAPE22.0%),且进行了验证(N=37,R~20.7,RMSE0.9 m~(-1),MAPE22.0%).

Remote sensing estimation algorithm of diffuse attenuation coefficient applicable to different satellite data in Lake Taihu, China

The downwelling diffuse attenuation coefficient (K_d) plays a critical role in underwater light field. It determines the euphotic depth, and affects the distribution characteristics of phytoplankton and primary productivity. Based on in situ data collected from 4 cruise surveys in Lake Taihu during 2008-2013, both the variations and driving factors of K_d were analyzed, and an algorithm to estimate downwelling diffuse attenuation coefficient at 490 nm for a variety of satellites'' data was developed and validated. The results show that:(1) suspended particular inorganic matter is the decisive factor of K_d in Lake Taihu;(2) the algorithm of red-green band ratio is the most suitable for estimating K_d(490) in Lake Taihu (N=72, R²=0.72, RMSE=0.89 m^-1, MAPE=21.58%);(3) with the simulated remote sensing reflectance of the main sensor bands such as MODIS/EOS, OLCI/Sentinel-3, GOCI/COMS and MSI/Sentinel-2, the red-green ratio algorithm to estimate K_d(490) was established for a variety of remote sensing sensors(N=72,R²>0.7,RMSE<0.9 m^-1,MAPE<22.0%), and the validation was good as well(N=37,R²>0.7,RMSE<0.9 m^-1,MAPE<22.0%).