Do we still need reflectance? From radiance to snow properties in mountainous terrain: a case study with the EMIT imaging spectrometer

Abstract. Global patterns of snow darkening and melting, induced by grain metamorphism and the accumulation of small light-absorbing particles (LAPs), such as mineral dust, black carbon, volcanic ash, or algae cells, lead to an intensified radiative forcing and retreat of Earth's snow cover. Mapping and quantifying snow grain size and LAPs on both temporal and spatial scales are needed to improve the prediction of melt rates and their impacts on climate change. High-resolution visible-to-shortwave-infrared (VSWIR) imaging spectrometers herald a new era of passive spaceborne remote sensing, which will help to fulfill this objective. This technology provides measurements of reflected solar radiation in continuous spectral channels throughout the solar spectrum, allowing for the detection of narrow ice and LAP absorption bands. One of these instruments is NASA's Earth Surface Mineral Dust Source Investigation (EMIT) that was launched to the International Space Station (ISS) in July 2022. EMIT observations include snow cover in low- to mid-latitude mountainous regions, such as the western US, the Andes in South America, and high-mountain Asia. Accurate retrievals of snow surface properties, including grain size, liquid water content, and concentrations of mineral dust and algae, require precise, ideally joint accounting for atmospheric, topographic, and anisotropic effects in the reflected radiance. However, some methods still either neglect physical effects of the surface or utilize the surface reflectance as an intermediate non-physical quantity, in part without proper error propagation from atmospheric modeling and obtained from statistical modeling. Moreover, the term “surface reflectance” is often used with ambiguity in the literature, which instantly raises the question of whether we still need this quantity as a retrieval product. In this contribution, we present a novel forward model that couples the MODTRAN atmosphere radiative transfer code with a physics-based snow reflectance model that utilizes the multistream DISORT program. Our model allows us to estimate snow surface and atmosphere properties directly from measured radiance. We apply the approach to EMIT images from Patagonia, South America, and compare our results to the EMIT L2A products that retrieve surface reflectance as a free parameter. We find discrepancies in snow grain size of up to 200 µm and in dust mass mixing ratio of up to 75 µg g−1. Furthermore, we demonstrate differences in instantaneous LAP radiative forcing of up to 400 W m−2. We conclude that we still need reflectance but only if it is clearly defined and preferably modeled as a quantity within the forward model. These findings will be essential for the conception of retrieval algorithms for future orbital imaging spectroscopy missions, such as NASA's Surface Biology and Geology (SBG).