The High lAtitude sNowfall DEtection aLgorithm for ATMS (HANDEL-ATMS): a new algorithm for snowfall retrieval at the high latitudes

Snowfall detection and quantification are challenging tasks in the Earth science field. Ground-based instruments provide only punctual measurements, and therefore the development of satellite-based snowfall retrieval methods is necessary for the implementation of a global monitoring system of snowfall. In particular, Passive Microwave (PMW) radiometric measurements are used for snowfall remote sensing: the retrieval is based on the scattering effect of the snowflakes visible in the high-frequency channels (> 80 GHz) on the upwelling radiation. However, the detection is made difficult by the weakness of this signature and by the contamination by the background surface emission/scattering signal. This phenomenon is particularly evident at high latitudes, where the prevalence of very light snowfall events and the extremely cold and dry environmental conditions make snowfall retrieval very difficult. The exploitation of operational microwave sounders on near-polar orbits such as the Advanced Technology Microwave Sounder (ATMS) allows for a very good coverage at high latitudes. Moreover, the wide range of channels (from 22 GHz to 183 GHz), allows for a radiometric characterization of the surface at the time of the overpass.

In this work the High lAtitude sNow Detection and rEtrieval aLgorithm for ATMS (HANDEL-ATMS), a new snowfall retrieval algorithm developed especially for high latitude environmental conditions and based on the ATMS observations, is described. The algorithm is based on the use of ATMS-CPR coincidence dataset, i. e. a dataset where each ATMS multichannel observation is associated with a vertical snow profile obtained by the CloudSat Cloud Profiling Radar (CPR) and therefore it is possible to analyze the relationship between the vertical precipitation structure and the PMW measurements in a direct way, without using simulated datasets.

The algorithm is composed of three main modules. The first module, based on the PMW Empirical clod Surface Classification Algorithm (PESCA), exploits ATMS low-frequency channel observations to obtain the surface classification and radiometric characterization at the time of the overpass. The second module estimates a set of clear-sky simulated brightness temperatures (TBs) by exploiting the previously derived surface radiometric properties. The clear-sky TBs set is compared with the observed TBs to highlight the snowfall signature. The third module is composed of four neural networks, which have been tuned against the CPR snowfall products. These networks, which exploit the deviation of the ATMS TBs from the clear-sky simulated TBs and the PESCA surface classification flag as inputs, return as outputs a snowfall detection flag and the surface snowfall rate estimate. HANDEL-ATMS shows very good detection capabilities - POD = 0.83, FAR = 0.18 and HSS=0.68. Estimation error statistics show an overestimation of very light snowfall events, but a good agreement for more intense events with respect to CPR snowfall products. The analysis of the results for an independent ATMS-CPR coincidence dataset and of selected snowfall events evidence the capability of HANDEL-ATMS to well detect and estimate snowfall also in presence of extreme environmental conditions typical of higher latitudes – dry and cold atmosphere and snow-covered background surface.