Potential of Passive Microwave around 183 GHz for Snowfall Detection in the Arctic

Abstract

This study evaluates the potential use of the Microwave Humidity Sounder (MHS) for snowfall detection in the Arctic. Using two years of colocated MHS and CloudSat observations, we develop an algorithm that is able to detect up to 90% of the most intense snowfall events (snow water path ≥400 g m−2 and 50% of the weak snowfall rate events (snow water path ≤50 g m−2. The brightness temperatures at 190.3 GHz and 183.3 ± 3 GHz, the integrated water vapor, and the temperature at 2 m are identified as the most important variables for snowfall detection. The algorithm tends to underestimate the snowfall occurrence over Greenland and mountainous areas (by as much as −30%), likely due to the dryness of these areas, and to overestimate the snowfall occurrence over the northern part of the Atlantic (by up to 30%), likely due to the occurrence of mixed phase precipitation. An interpretation of the selection of the variables and their importance provides a better understanding of the snowfall detection algorithm. This work lays the foundation for the development of a snowfall rate quantification algorithm.

Satellite Remote Sensing of Precipitation and the Terrestrial Water Cycle in a Changing Climate

“…Deep neural networks (DNN) are also being adopted as a viable alternative to merge together data from passive and active sensors at high latitudes [182]. This is accompanied by significant progress being made towards using PMW bands around 183 GHz for snowfall detection in Arctic conditions [183]. A better quantification of snowfall at high latitudes [184] and on a global scale [176] Deep neural networks (DNN) are also being adopted as a viable alternative to merge together data from passive and active sensors at high latitudes [182].…”

“…A better quantification of snowfall at high latitudes [184] and on a global scale [176] Deep neural networks (DNN) are also being adopted as a viable alternative to merge together data from passive and active sensors at high latitudes [182]. This is accompanied by significant progress being made towards using PMW bands around 183 GHz for snowfall detection in Arctic conditions [183]. A better quantification of snowfall at high latitudes [184] and on a global scale [176] will also contribute to unravelling the controversy around the impact of Arctic warming on increased rainfall/snowfall [185] and, more generally, to explaining the differences in model projections between the southern and northern hemispheres and wet and dry regions [186].…”

Comparative Assessment of Snowfall Retrieval From Microwave Humidity Sounders Using Machine Learning Methods

“…10 and 166 GHz) frequency channels have shown great potentials for retrieving snowfall over snow and ice surfaces (Ebtehaj & Kummerow, 2017). Furthermore, Edel et al (2019) demonstrated that among all MHS channels, near 183 GHz channels are the most important ones to detect snowfall. It should be noted that besides sensitivity towards scattering signatures, high frequency channels are useful to detect snowfall because they are less affected by surface features (Skofronick-Jackson et al, 2004).…”

“…Their report shows that snowfall rate estimated by combining GMI channels, environmental variables, and infrared bands can be improved over snowfall estimates only from MW channels. A recent study by Edel et al (2019) have demonstrated the ability of MHS channels to detect snow over the Arctic region.…”

“…Although snow retrieval over land is found more complex than over ocean because of the variations in emissivity from heterogeneous nature of surface (Skofronick‐Jackson & Johnson, 2011), combinations of lower‐ and higher‐frequency (e.g., 10 and 166 GHz) channels have shown great potentials for retrieving snowfall over snow and ice surfaces (Ebtehaj & Kummerow, 2017). Furthermore, Edel et al (2019) demonstrated that among all Microwave Humidity Sounder (MHS) channels, near‐183 GHz channels are the most important ones to detect snowfall. It should be noted that besides sensitivity toward scattering signatures, high‐frequency channels are useful to detect snowfall because they are less affected by surface features (Skofronick‐Jackson et al, 2004).…”