Improving Surface Reflectance Estimation for VIIRS AOT Retrieval over Land

Surface reflectance ratios of M1/M5, M2/M5, M3/M5, and M11/M5 are crucial to the VIIRS AOT retrieval over land. Having better estimates of these ratios can improve AOT retrievals. Current VIIRS aerosol retrieval algorithm uses constant ratios over the whole globe. We present an investigation of the surface reflectance ratios over the global AERONET sites and demonstrate that we can improve the AOT retrieval by using more accurate surface reflectance ratios. Surface reflectance in the M1, M2, M3, M11 and M5 channels were retrieved at global AERONET sites using AERONET-retrieved AOT. M1/M5, M2/M5, M3/M5, and M11/M5 surface reflectance ratios were then calculated for each site separately. These ratios vary strongly from site to site. This variability is the reason for the biases and standard deviations in the AOT retrieval when a single fixed value of the ratio is used. When instead the individual ratios are used in the AOT retrieval, their statistics (naturally) improve. In order to obtain these ratios without the help of the AERONET AOT, we tested a clear sky method, which assumes a low AOT and looks for the lowest atmospheric corrected M3/M5 ratio over a period of time. The clear sky ratios were obtained for each AERONET sites, which were then applied in land AOT retrieval. The resulting AOT retrievals also showed improvements over those from the original algorithm.