Non-Lambertian snow surface reflection models for simulated top-of-the-atmosphere radiances in the NIR and SWIR wavelengths

Accurately modeling snow reflectance is one way to improve satellite observations in the high -latitude regions. Snow surfaces are known to be challenging for atmospheric retrievals in the short-wave infrared (SWIR) wavelength regime due to their low reflectance. For example, current algorithms for satellite -based remote sensing of atmospheric carbon dioxide (CO2) do not take into account the unique reflective properties of snow surfaces. In this paper, we present a measurement -based snow surface reflectance model in the near -infrared (NIR; 755-775 nm) and SWIR (1590-1620 nm, 2040-2080 nm) bands used in remote sensing of atmospheric CO2. We study snow reflectance in detail using a novel atmospheric radiative transfer model (RTM) software and a measurement -based model for snow bi-directional reflectance distribution function (BRDF) to identify how the observations could be optimized in regards of observation geometry and wavelengths. The novel simulation software for NIR-SWIR atmospheric radiative transfer, Raysca, is presented and validated. Topof -the -atmosphere radiance simulations show that forward -viewing geometries over snow-covered surfaces yield higher radiances than the traditional nadir -viewing geometries, which could indicate the preferability of forward -viewing observation modes in the retrieval of atmospheric CO2. Similarly, atmospheric observations in the 1.6 mu m CO2 absorption band might be preferable to the 2.0 mu m CO2 absorption band due to higher radiances in the 1.6 mu m band.