基于Sentinel-2 MSI影像的秦皇岛海域叶绿素a浓度遥感反演

基于2013-2018年秦皇岛海域实测遥感反射率和叶绿素a浓度数据，建立了该海域Sentinel-2MSI影像的叶绿素a浓度遥感反演模型。结果表明：443 nm、490 nm和560 nm处的等效遥感反射率比值与叶绿素a浓度相关系数普遍高于其他波段或组合，通过经典的OC3Mv6算法拟合分析，得到秦皇岛海域叶绿素a浓度遥感反演的最佳算法，R²=0.804,MAPE=40.2%,RMSE=4.73 mg/m³；利用2016年7月6日的实测叶绿素a浓度数据对Sentinel-2 MSI遥感反演结果进行了真实性检验，MAPE=35.9%，可以满足应用要求；采用2020年2月、5月、7月及10月Sentinel-2 MSI影像进行叶绿素a浓度反演，发现春、夏季秦皇岛海域叶绿素a浓度梯度变化显著，而秋、冬季叶绿素a浓度分布相对均匀，且春、夏季沿海海域叶绿素a浓度明显高于秋、冬季。

Retrieval of chlorophyll a concentration from Sentinel-2 MSI image in Qinhuangdao coastal area

Based on the in-situ remote sensing reflectance (R_rs) and chlorophyll a concentration (Chl a) data in the Qinhuangdao coastal area from 2013 to 2018, here a regional Chl a retrieval model was developed specifically for the Sentinel-2 MSI image. The results showed that the correlation coefficients between R_rs(443)/R_rs(560), R_rs(490)/R_rs(560) and Chl a were generally higher than that of other bands or combinations. The classical OC3Mv6 algorithm was used for fitting, the best Chl a retrieval model in the Qinhuangdao coastal area was obtained with R2=0.804, MAPE=40.2% and RMSE=4.73 mg/m3; using the in-situ Chl a data on July 6, 2016, the accuracy of the Sentinel-2 MSI remote sensing inversion result was tested, and the MAPE was 35.9%, which basically meet the application requirements; the regional Chl a retrieval model was carried out for the Sentinel-2 MSI image in February, May, July and October 2020, and it was found that the variations of Chl a were significant in spring and summer, while Chl a in autumn and winter were relatively uniform, and Chl a in coastal waters in spring and summer was significantly higher than that in autumn and winter.