Retrieval of columnar water vapour over land from backscattered solar radiation using the Medium Resolution Imaging Spectrometer

We describe a new algorithm to derive columnar water vapour under cloud-free conditions over land from backscattered solar radiation in the near-infrared. The algorithm will be used in ESAʹs Medium Resolution Imaging Spectrometer (MERIS) ground processor. It is based on radiative transfer simulations, where the radiance ratio between the MERIS channels 15 (900 nm) and 14 (885 nm) is used in an inversion procedure based on regressions. The theoretical accuracy of the algorithm is about 1.7 kg/m2. We discuss and quantify possible error sources using radiative transfer simulations. These error sources are variable aerosol optical thickness, type, and vertical distribution, spectral variations in land surface reflectance, deviations between the actual and nominal bandsetting of the MERIS, sensor noise, variations in surface pressure; and temperature variations. For validation, we re-calibrate the algorithm to the bandsettings of the Modular Optoelectronical Scanner (MOS), which has been flown on the Indian IRS platform since 1996. Comparisons of the retrieved water vapour path (WVP) with colocated radiosoundings for 239 cases in the period 1996–1999 show a RMSE of 2.49 kg/m2 with a BIAS component of 0.04 kg/m2.