Severe convection in the Mediterranean from microwave observations and a convection-permitting model

International audienceThis study investigates severe convection in the Mediterranean during the first Special Observation Period (SOP-1; 5 September to 6 November 2012) of the Hydrological Cycle in the Mediterranean Experiment (HyMeX) with the objectives of providing novel information about severe convection on its vertical structure, spatio-temporal variability as well as evaluating the ability of a convection-permitting model to reproduce this variability. Two criteria, namely deep convection (DC) and convective overshooting (COV), are computed using the water vapour channels of the Microwave Humidity Sounder (MHS). Special attention is paid to the COV as it is associated with particularly severe weather. For the first time, the COV criterion was assessed in the Mediterranean, using two case-studies conjointly observed by the airborne RASTA radar and MHS. COV is characterised by high ice water content (up to 2 g m−3) in the mid and upper troposphere (up to 12.5 km in the stratosphere). During the SOP-1, DC and COV occurred about 0.1 and 0.03% of the total observation time, respectively. The Atlantic weather regimes appear to affect the temporal distribution of these convective events. Most of the DC and COV occurrences were found along the western coasts of Italy and Greece, mainly during the 10–15 October and 25 October–3 November episodes. These two episodes, for which severe meteorological events (e.g. tornadoes) were reported, are significant when compared with the 2002–2013 climatology (above the 75th percentile). Both criteria are also employed to assess the current ability of the Meso-NH model to forecast severe convection using a model-to-satellite approach. The forecast DC and COV are found to be highly correlated in time with the observations, but are strongly underestimated. This suggests that the model missed a significant part of the most intense convective events and their associated hazards, and underlines the need for better characterisation of model uncertainties associated with severe convection