Atmospheric Column Water Vapor Retrieval using Atmospheric Precorrected Differential Absorption Technique from AVIRIS-NG Data

Abstract

In this paper, the atmospheric column water vapor (WV) has been retrieved using atmospheric precorrected differential absorption (APDA) technique from the hyperspectral data of Next-Generation Airborne Visible/Infrared Imaging Spectrometer (AVIRIS-NG). The APDA technique is extended method of Continuum interpolation band ratio (CIBR), and it is based on Linear regression ratio (LIRR) method. In APDA the additional atmospheric precorrected term reduces atmospheric path radiance, which provides good accuracy over lower surface albedo. Moreover owing to use of more than three absorption bands, this technique gives better performance over large surface reflectance variation. Optimum absorption bands selection has been performed by derived at-sensor radiance using MODTRAN (MODerate resolution atmospheric TRANsmission) 5.3 radiative transfer model (RTM), and six WV absorption bands are selected between 1120nm to 1150nm. At-sensor radiance simulations have been carried out for varying atmospheric and surface parameters, and then a band ratio RAPDA is derived from absorbing and non-absorbing band, corresponding to different WV concentration. The theoretical validation of the technique has been performed using actual radiosonde WV values, which shows good agreement with a root mean square error (RMSE) of 0.0145 gm/cm2. The technique has been first-time applied to AVIRIS-NG Indian campaign data containing heterogeneous surface of two sites of Ahmedabad and Anand. Validation of retrieved WV was carried out with the NASA's AVIRIS-NG level-2 WV products on pixel-to-pixel basis. The RMSE for Ahmedabad and Anand is 0.164 gm/cm2 and 0.112 gm/cm2, respectively, which demonstrates a good performance of the proposed technique for the WV retrieval.