This is a report on the spatial variation of atmospheric carbon dioxide (CO<sub>2</sub>) concentration with urban congestion estimated from Airborne Visible/Infrared Imaging Spectrometer Next Generation (AVIRIS-NG) hyperspectral images procured for the first time in the tropical urban atmosphere of India. It combines hyperspectral field measurements and airborne image analysis and puts forward a modified version of differential optical absorption spectroscopic technique, termed as a-DOAS that defines the absorption depths through weighted average of radiance at absorbing and non-absorbing wavebands. Field spectroradiometry was carried out for part of two adjacent cities, namely Howrah and Kolkata, over sites of varied population density and vehicle abundance at both narrow and wide angle field-of-view. The local concentrations of atmospheric CO<sub>2</sub> at the ground surface and the solar illumination at the measuring spot were noted simultaneously. As an alternative to the general validation method of comparing the image-derived absorption depth with that simulated with radiative transfer model, the present work is based on the image-derived and the ground-based data. The relative differences of the surface reflectance for the pixels of different features were reduced by normalizing the pixel values with suitable constants. While calculating the CO<sub>2</sub> absorption depths, a correction for the adjacent water vapor absorption band was made with MODTRAN simulation. Since the pixel-to-pixel variation of radiance was too fine, the CO<sub>2</sub> map was generated by computing average through convolution and filtering with kernel of 11×11 pixels. The overall observation was lower concentration for vegetated regions in comparison with concreted urban areas.