Optimizing the range of atmospheric condition parameters to avoid over and under-estimation of uncertainty

Abstract

This paper aims to solve the critical problem of uncertainty optimization of two atmospheric condition parameters: water vapor concentration (wv) and visibility. Both are used in atmospheric correction (AC) processes when retrieving surface reflectance from at-sensor radiance. Parameter uncertainty is often over or under estimated. A framework is proposed that determines the relation (indirect) of the parameters with estimated surface reflectance and uses this to reduce the uncertainty. Results show that optimizing the uncertainty of the parameters reduces the range of visibility from 18–42 to 31–39 km and the range of wv from 1.6–2.9 to 2.05–2.30 g cm−2 for the class agriculture. Also, the total number of simulations that is needed for further studies such as an uncertainty or a sensitivity analysis in AC is reduced. The study concludes that there is an inverse effect of wv and visibility on the surface reflectance.