Abstract. Polarimetric radar systems are commonly used to study the microphysics of precipitation. While they offer continuous measurements with a large spatial coverage, retrieving information about the microphysical processes that govern the evolution of snowfall from the polarimetric signal is challenging. The present study develops a new method, called process identification based on vertical gradient signs (PIVSs), to spatially identify the occurrence of the main microphysical processes (aggregation and riming, crystal growth by vapor deposition and sublimation) in snowfall from dual-polarization Doppler radar scans. We first derive an analytical framework to assess in which meteorological conditions the local vertical gradients of radar variables reliably inform about microphysical processes. In such conditions, we then identify regions dominated by (i) vapor deposition, (ii) aggregation and riming and (iii) snowflake sublimation and possibly snowflake breakup, based on the sign of the local vertical gradients of the reflectivity ZH and the differential reflectivity ZDR. The method is then applied to data from two frontal snowfall events, namely one in coastal Adélie Land, Antarctica, and one in the Taebaek Mountains in South Korea. The validity of the method is assessed by comparing its outcome with snowflake observations, using a multi-angle snowflake camera, and with the output of a hydrometeor classification, based on polarimetric radar signal. The application of the method further makes it possible to better characterize and understand how snowfall forms, grows and decays in two different geographical and meteorological contexts. In particular, we are able to automatically derive and discuss the altitude and thickness of the layers where each process prevails for both case studies. We infer some microphysical characteristics in terms of radar variables from statistical analysis of the method output (e.g., ZH and ZDR distribution for each process). We, finally, highlight the potential for extensive application to cold precipitation events in different meteorological contexts.

Abstract. Polarimetric radar systems are commonly used to study the microphysics of precipitation. While they offer continuous measurements with a large spatial coverage, retrieving information about the microphysical processes that govern the evolution of snowfall from the polarimetric signal is challenging. The present study develops a new method, called Process Identification based on Vertical gradient Signs (PIVS), to spatially identify the occurrence of the main microphysical processes (aggregation and riming, crystal growth by vapor deposition, and sublimation) in snowfall from dual-polarization Doppler radar scans. We first derive an analytical framework to assess in which meteorological conditions the local vertical gradients of radar variables reliably inform about microphysical processes. In such conditions, we then identify regions dominated by (i) vapor deposition, (ii) aggregation and riming and (iii) snowflake sublimation and possibly snowflake breakup based on the sign of the local vertical gradients of the reflectivity ZH and the differential reflectivity ZDR. The method is then applied to data from two frontal snowfall events: one in coastal Adélie Land, Antarctica and one in the Taebaeck mountains in South Korea. The validity of the method is assessed by comparing its outcome with snowflake observations using a Multi-Angle Snowflake Camera and with the output of a hydrometeor classification based on polarimetric radar signal. The application of the method further makes it possible to better characterize and understand how snowfall forms, grows and decays in two different geographical and meteorological contexts. For the Antarctic case study, we show that crystal growth by vapor deposition dominates above 2500 m a.g.l., aggregation and riming prevail between 1500 and 2500 m a.g.l. and snowflake sublimation by low-level katabatic winds occurs below 1500 m a.g.l.. For the event in South Korea, aggregation and riming dominate between 4000 and 4800 m a.g.l., with local sublimation below and vapor deposition above. We infer some microphysical characteristics in terms of radar variables from statistical analysis of the method output (e.g. ZH and ZDR distribution for each process). We finally highlight the potential for extensive application to cold precipitation events in different meteorological contexts.

<p>Lagrangian studies are a widely-used and powerful way to analyse and interpret phenomenons in oceanography and atmospheric sciences. Such studies can be based on dataset either consisting of real trajectories (e.g. oceanic drifters or floats) or of virtual trajectories computed from velocity outputs from model or observation-derived velocities. Such data can help investigate pathways of water masses, pollutants or storms, or identify important convection areas to name a few. As many of these analyses are based on large volumes of data that can be challenging to examine, machine learning can provide an efficient and automated way to classify information or detect patterns.</p><p>Here, we present an application of unsupervised clustering to the identification of the main pathways of the shelf-break branch of the Labrador Current, a critical component of the North Atlantic circulation. The current flows southward along the Labrador Shelf and splits in the region of the Grand Banks, either retroflecting north-eastward and feeding the subpolar basin of the North Atlantic Ocean (SPNA) or continuing westward along the shelf-break, feeding the Slope Sea and the east coast of North America. The proportion feeding each area impacts their salinity and convection, as well as their biogeochemistry, with consequences on marine life.</p><p>Our dataset is composed of millions of virtual particle trajectories computed from the water velocities of the GLORYS12 ocean reanalysis. We implement an unsupervised Machine Learning clustering algorithm on the shape of the trajectories. The algorithm is a kernalized k-means++ algorithm with a minimal number of hyperparameters, coupled to a kernalized Principal Component Analysis (PCA) features reduction. We will present the pre-processing of the data, as well as canonical and physics-based methods for choosing the hyperparameters. </p><p>The algorithm identifies six main pathways of the Labrador Current. Applying the resulting classification method to 25 years of ocean reanalysis, we quantify the relative importance of the six pathways in time and construct a retroflection index that is used to study the drivers of the retroflection variability. This study highlights the potential of such a simple clustering method for Lagrangian trajectory analysis in oceanography or in other climate applications.</p>

<p align="justify"><span>The current assessment of the Antarctic surface mass balance mostly relies on reanalysis products or climate model simulations. The ability of models to reproduce the precipitation field at the regional and continental scales not only depends on the simulation of the atmospheric dynamics over the Southern Ocean and of the advection of moisture towards the ice sheet, but also on the representation of the microphysical processes that govern the formation and growth of ice crystals and snowflakes. This presentation reviews recent studies to stress the importance and challenges of evaluating the precipitation microphysics over Antarctica in climate models. It also discusses how recent observational campaigns including ground-based remote-sensing instruments can help pinpoint key processes that should be represented in models. We then present tangible examples of evaluation and improvement of microphysical schemes in the Polar WRF model thanks to radar and lidar observations acquired near Dumont d’Urville and Mawson stations on the Antarctic coast. Particular attention is devoted to three processes : i) the sublimation of snowfall within the katabatic layer that considerably reduces the amount of precipitation that actually reaches the surface ; ii) the snowflake aggregation responsible for rapid depletion of crystals within clouds ; iii) the generation of supercooled liquid water in frontal clouds that leads to crystal/snowflake riming. Such studies, albeit preliminary, could pave the way for further evaluations of clouds and precipitation in climate models in different Antarctic contexts, especially in the cold and pristine atmosphere of the Plateau.</span></p>

<p>Microphysical processes in cold precipitating clouds are not fully understood and their parametrization in atmospheric models remains challenging . In particular the lack of evaluation and validation of the microphysical parameterizations in polar regions questions the reliability of the ice sheet surface mass balance assessments. Recently, strong discrepancies have been found in the precipitation structure between simulations with different microphysical parameterizations over the Antarctic coast.</p><p>The dissimilarities between simulations seem to be due to different treatments of the riming, aggregation and sublimation processes.</p><p> </p><p>Evaluating the representation of a particular microphysical process in a model is delicate, especially because it is difficult to obtain in situ estimations, even qualitative, of a given microphysical process. In this study, we developed a method to identify the regions in radar scans where either aggregation and riming, vapor deposition or sublimation are the dominant microphysical processes.</p><p>This method is based on the vertical (downward) gradients of reflectivity and differential reflectivity computed over columns extracted from range height indicator scans. Because of the expected increase in size and decrease in oblateness of the particles, aggregation and riming are identified as regions with positive gradients of reflectivity and negative gradients of differential reflectivity. Because of the expected increase in size and oblateness, vapor deposition is identified as regions with positive gradients of reflectivity and positive gradients of differential reflectivity. Because of the expected decrease in size and in concentration, snowflake sublimation, and possibly snowflake breakup, are defined as regions with negative gradients of reflectivity.</p><p> </p><p>The method was employed on two frontal precipitation events, which took place during the austral summer APRES3 campaign (2015-2016) in Dumont d’Urville (DDU) station, Antarctic coast. Significant differences appear in the mean altitudinal distribution where each process takes place. Given that the radar signal extends up to 4500 m a.g.l., we could show that crystal growth dominates around 2800 m while aggregation and riming prevail around 1500 m. Sublimation mostly occurs below 900 m, concurring with previous studies stating that snowflakes preferentially sublimate in the relatively dry katabatic boundary layer.</p><p>Moreover the statistical distributions of different radar variables provides quantitative information to further characterize the microphysical processes of interest.</p><p>This method could be further used to assess the ability of atmospheric models to reproduce the correct microphysical processes at the correct locations.</p>