Estimating GOCI daily PAR and validation

Abstract

Photosynthesis available radiation (PAR) is the most important source for primary production at the ocean. In these days, satellite remote-sensing has an advantage in terms of cost-effectiveness and spatio-temporal resolutions for observing global oceanic PAR. Geostationary Ocean Color Imager (GOCI) was employed to observe accurate daily PAR at the ocean surface around Korean peninsula from a PAR model based on the Plane-parallel theory. Six bands radiance data of GOCI L1B image that ranged from 400 to 700 nm and altitude angle of sun/sensor were used as input data of PAR model and other parameters such as water vapor were obtained from climatological data. Estimated GOCI daily PAR was compared with in-situ data observed during 2015 at two stations. GOCI daily PAR and in-situ daily PAR shows high correlation coefficient, 0.98, and root-mean-square error (RMSE) is 4.50 Ein/m2/day, representing seasonal bias during spring and winter season when GOCI daily PAR has been underestimated. To correct the underestimation the equation was modified using a linear regression between GOCI and in-situ daily PAR observed during clear sky conditions, which showed a decreased RMSE of 3.08 Ein/m2/day with the correction of the underestimation. Validation for the finally developed GOCI PAR algorithm was carried out using in-situ daily PAR observed during 2016, which showed a high correlation coefficient (0.98) and a low RMSE (2.69 Ein/ global oceanic PAR. Geostationary Ocean Color Imager (GOCI) was employed to observe accurate daily PAR at the ocean surface around Korean peninsula from a PAR model based on the Plane-parallel theory. Six bands radiance data of GOCI L1B image that ranged from 400 to 700 nm and altitude angle of sun/sensor were used as input data of PAR model and other parameters such as water vapor were obtained from climatological data. Estimated GOCI daily PAR was compared with in-situ data observed during 2015 at two stations. GOCI daily PAR and in-situ daily PAR shows high correlation coefficient, 0.98, and root-mean-square error (RMSE) is 4.50 Ein/m2/day, representing seasonal bias during spring and winter season when GOCI daily PAR has been underestimated. To correct the underestimation the equation was modified using a linear regression between GOCI and in-situ daily PAR observed during clear sky conditions, which showed a decreased RMSE of 3.08 Ein/m2/day with the correction of the underestimation. Validation for the finally developed GOCI PAR algorithm was carried out using in-situ daily PAR observed during 2016, which showed a high correlation coefficient (0.98) and a low RMSE (2.69 Ein/