The current assessment of the Antarctic surface mass balance mostly relies on reanalysis products or climate model simulations. However, little is known about the ability of models to reliably represent the microphysical processes governing the precipitation. This study makes use of recent ground-based precipitation measurements at Dumont d'Urville station in Adelie Land to evaluate the representation of the precipitation microphysics in the Polar version of the Weather Research Forecast (Polar WRF) atmospheric model. During two summertime snowfall events, high-resolution simulations are compared to measurements from an X-band polarimetric radar and from a Multi-Angle Snowflake Camera (MASC). A radar simulator and a "MASC" simulator in Polar WRF make it possible to compare similar observed and simulated variables. Radiosoundings and surface-meteorological observations were used to assess the representation of the regional dynamics in the model. Five different microphysical parameterizations are tested. The simulated temperature, wind, and humidity fields are in good agreement with the observations. However, the amount of simulated surface precipitation shows large discrepancies with respect to observations, and it strongly differs between the simulations themselves, evidencing the critical role of the microphysics. The inspection of vertical profiles of reflectivity and mixing ratios revealed that the representation of the sublimation process by the low-level dry katabatic winds strongly influences the actual amount of precipitation at the ground surface. By comparing the simulated radar signal as well as MASC and model particle size distributions, it is also possible to identify the microphysical processes involved and to pinpoint shortcomings within the tested parameterizations.