Validation of CloudSat-CPR Derived Precipitation Occurrence and Phase Estimates across Canada

Abstract

Snowfall affects the terrestrial climate system at high latitudes through its impacts on local meteorology, freshwater resources and energy balance. Precise snowfall monitoring is essential for cold countries such as Canada, and particularly in temperature-sensitive regions such as the Arctic; however, its size and remote location means the precipitation gauge network there is sparse. While satellite remote sensing of snowfall from instruments such as CloudSat-CPR offers a potential solution, satellite detection of precipitation phase has not been systematically evaluated across Canada. In this study, CloudSat-based precipitation occurrence and phase retrievals were validated at 26 stations across Canada maintained by Environment and Climate Change Canada (ECCC). Probability of Detection (POD), defined as the percentage agreement between coincident CloudSat and human-observed present weather information for precipitation (solid, liquid or no precipitation), and False Alarm Ratio (FAR) were used as the primary metrics for validation. The mean POD (FAR) for precipitation occurrence across Canada is 65.5% ± 4.3 (31.4% ± 5.1) and for no precipitation is 90.6% ± 1.4 (11% ± 2.5). The results show lower rates of detection under cloudier skies, in the presence of (freezing) drizzle and for lighter snowfall, which may be explained by a large number of false-positives due to CloudSat-CPR’s high instrumental sensitivity. When CloudSat correctly detects the occurrence of precipitation, it shows uniformly high POD (>80%) and low FAR (<10%) for classifying the phase of precipitation. Large databases of coincident ground and satellite measurements allow us to provide a new estimate of around 9% for the frequency of virga events, a factor of two smaller than a previous estimate for the Arctic. The results from this study show that CloudSat has useful accuracy in detecting precipitation occurrence and very high accuracy at classifying precipitation phase, over diverse climate zones across Canada. As such, there is significant potential for satellite monitoring of snowfall in remote, cold regions.

Marine Cold‐Air Outbreak Snowfall in the North Atlantic: A CloudSat Perspective

“…Snowfall rate is then derived in the 2C‐SNOW‐PROFILE product, an optimal estimation algorithm that uses CPR reflectivity alongside ancillary temperature and cloud mask data to identify a cloud layer producing snow. The CPR has a high (>85%) probability of detection of snowfall (Cao et al., 2014; Chen et al., 2016; Kodamana & Fletcher, 2021) and correctly assigns hydrometeors as frozen for 95% of snowfall events detected by surface observations (Kodamana & Fletcher, 2021). More specific details about the 2C‐SNOW‐PROFILE retrieval are available in the Algorithm Theoretical Basis Document (ATBD, Wood & L’Ecuyer, 2018).…”

A Novel Approach to Validate Satellite Snowfall Retrievals by Ground-Based Point Measurements

“…Therefore, the S/P ratio derived from the satellite data is still physically meaningful. There are a few studies in the literature on comparing CloudSat snowfall product with surface station measurements [52][53][54], ground-based radar measurements [55][56][57][58], and model reanalysis [59][60][61]. Most of these studies are conducted over high latitudes.…”

Using CloudSat‐Derived Snow Accumulation Estimates to Constrain Gridded Snow Water Equivalent Products

“…These reflectivity profiles provide information on cloud type, shape and the presence of hydrometeors within the cloud, along with corresponding internal precipitation rate estimates (Kulie et al., 2020). CloudSat also demonstrates high accuracy in determining precipitation phase to differentiate between rain, mixed‐phase and snowfall events (Kodamana & Fletcher, 2021). CloudSat has a 16 day repeat cycle with granule tracks that extend to an N/S orbital maxima which converge toward the poles.…”