Are visibility-derived AOT estimates suitable for parameterising satellite data atmospheric correction algorithms?

Aerosol Optical Thickness (AOT) is an important parameter in Radiative Transfer Models (RTMs) used for atmospheric correction of remotely-sensed data. It is often estimated from horizontal visibility measurements by use of the Koschmieder formula or other related methods built into RTMs. This paper assesses the accuracy of this estimation, in the context of atmospheric correction, by comparing AERONET AOT data with AOT estimates from UK Met Office visibility data at a site in Hampshire, UK. Root Mean Square Errors errors are calculated for a number of visibility categories (0–10km, 10–20km, 20–30km, 30–40km) and are found to be high for all visibilities (ranging from half to more than double the mean AOT for each category). For all visibilities < 30km, these errors are significantly higher than those from other AOT estimation methods. Simulations performed with the 6S RTM show that the effect of these errors on satellite-level radiances are large (up to 36W/m2/sr), and the change in vegetation indices (NDVI and ARVI) is smaller, but still significant. It is recommended that estimations of AOT based upon visibility measurements are only used if no alternatives are available, and that great caution is used when estimation is performed for visibilities < 10km.