利用2004~2012年在南海获得的9个航次的实测Chl-a数据, 采用NASA标准业务化算法OC3和针对低Chl-a水体所发展的最新算法OCI反演获得了相应的MODIS-Aqua Chl-a产品。通过建立实测与遥感产品的时空匹配数据对, 开展了Chl-a产品的适用性评估, 并对比分析了上述两种算法的性能。在此基础上, 利用南海实测遥感反射率(Rrs(λ))和MODIS-Aqua Rrs(λ)产品以及相应实测Chl-a的匹配数据集, 分别对算法OC3和OCI进行了区域性修正。结果显示: 基于算法OC3和OCI反演所得的MODIS-Aqua Chl-a产品值均高估了实测值, 平均绝对误差(APD) 的精度分别为56.30%和42.58%, 且算法OCI可明显改善低Chl-a水体(<0.25 mg·m-3)的反演精度;采用南海MODIS-Aqua Rrs(λ)产品与实测Chl-a匹配数据集(N=82)修正后的区域性算法NOC3和NOCI的精度均有不同程度提高, APD精度分别为37.85%和36.74%;采用现场实测Rrs(λ)与Chl-a匹配数据集(N=123)进行区域性修正后的算法INOC3和INOCI的APD精度分别为36.61%和37.79%, 上述两种方案精度较为接近。因此, 对于南海海域而言, 算法的区域性修正对于改善MODIS-Aqua Chl-a产品精度非常重要。

The performance of MODIS-Aqua Chl-a derived from the standard OC3 band-ratio algorithm and a new band-difference OCI algorithm are compared in the South China Sea(SCS) from the MODIS instrument on the Aqua satellite with nine cruises of chlorophyll measurements collected from 2004 to 2012. Then, the algorithms of OC3 and OCI were corrected by the parameter adjustments based on the satellite, in situ remote sensing reflectance (Rrs(λ)) and corresponding in situ Chl-a pairs respectively. The results show that MODIS-Aqua Chl-a products derived from OC3 and OCI algorithms overestimate the in situ data, with the Average Relative Error(APD) of 59.41% and 44 17%, respectively. However, the algorithm OCI significantly improves the retrieval accuracy for low chlorophyll concentration waters(<0.25 mg·m-3). After regional correction by MODIS-Aqua Rrs(λ) and corresponding Chl-a pairs, the algorithms are likely to fare better, and the modified algorithm NOC3 and NOCI show their APD accuracies are 37.85% and 36.74%, respectively. Moreover, the algorithms NOC3 and NOCI modified by in situ Rrs(λ) and corresponding Chl-a pairs also give a good estimateion with the APD accuracy of 36.61% and 37.79%. Therefore, it suggests that to conduct the regional correction for the operational algorithms is important to improve the accuracy of Chl-a estimation in the SCS.