An operational model for filling the black strips of the MODIS 1640 band and application to atmospheric correction

In this study, a LIMM (linear interpolation model for rebuilding the black strips of MODIS 1640 nm) model is proposed to rebuild the black strips of MODIS (moderate resolution imaging spectroradiometer) 1640 images, and for improving the performance of TSWNR (traditional 1240 nm band-based SWIR-NIR atmospheric correction model) model in deriving remote sensing reflectance in turbid coastal waters. By comparison with the field measurements, both the L3MAC (LIMM model-derived MODIS 1640 band-based atmospheric correction model) and TSWNR models can be used to derive remote sensing reflectance in the green and red regions, but the former is superior to the latter. Especially in summer in the Bohai Sea, use of the L3MAC model in estimating remote sensing reflectance decreases the MRE (mean root mean square error) values of estimation by >14% from the TSWNR model. Due to the great amount of detector noise in the MODIS SWIR wavelengths, there is still a >19% residual uncertainty in the L3MAC model-derived remote sensing, particularly in the NIR (near-infrared) and shortest blue regions, and both the L3MAC and TSWNR models produce >60% uncertainty in remote sensing reflectance retrievals. The success of the application of the L3MAC model to satellite data depends heavily on the detector noise in the MODIS SWIR (shortwave infrared) wavelengths. Our study suggests that more attention should be paid to how to minimize the effects of detector noise on the atmospheric correction results in the future.