Evaluation on the atmospheric correction methods for water color remote sensing by using MERIS image: A case study on chlorophyll-a concentration of Lake Poyang

Abstract

MERIS is the sensor of ENVISAT-1 satellite in orbit nearly 10 years, which was launched in 2002. The band setting and radiation sensitivity of MERIS could well be applied to chlorophyll-a concentration inference, but the atmospheric correction in Case II water is still a key problem. Taking the Lake Poyang, the biggest freshwater lake in China as a case, the results of ENVISAT-1 satellite MERIS images using FLAASH, 6S, BEAM and QUAC atmospheric correction algorithms in 2005 and 2011 with simultaneous in-situ spectral data was evaluated. The band combination factor of 12 chlorophyll-a retrieval models were also assessed. Results show that: (1) the order of atmospheric correction accuracy from high to low are FLAASH, 6S, BEAM, QUAC, with the mean relative error of 31.13%, 31.88%, 69.48% and 42.64%, respectively; and the determination coefficients (R2) of 0.60, 0.57, 0.38 and 0.24, respectively. (2) Among the band-combined chlorophyll-a retrieval models, FLAASH is the best one, followed by 6S, BEAM and QUAC. As for FLAASH algorithm, the three bands ((Rrs(510)/[Rrs(443)/Rrs(560)]) model is the most accurate one, the MRE is 25.12%, the determination coefficient is 0.74. This band combination of FLAASH atmospheric correction results is recommended for the chlorophyll-a inference in the Lake Poyang.