Seasonal heavy precipitation sensitivity to moisture corrections in the western Mediterranean across resolutions

Abstract

The controlling role of atmospheric water vapour for heavy precipitation leading to extreme events has been widely demonstrated, along with the existing gap of adequate moisture observations and the frequent biases present in model simulations concerning this fundamental variable. In this study, we profit from a state-of-the-art dense network of GPS measurements over Europe retrieving a homogenized GPS-derived Zenith Total Delay (GPS-ZTD) data set up to 10 min of temporal resolution, to assess the seasonal sensitivity of convection-related processes and heavy precipitation modelling to atmospheric humidity corrections. For this purpose, we perform nudging experiments with the COSMO-CLM model at two spatial resolutions, 7 km (parameterized convection) and 2.8 km (explicitly resolved convection) covering the autumn period of 2012, when the Special Observation Period (SOP) 1 of the Hydrological Cycle in the Mediterranean Experiment (HyMeX) program took place in the Western Mediterranean, which is our area of interest.The benefits and disadvantages of GPS-ZTD nudging and resulting moisture corrections are disentangled. The impact on high-resolution parameterized versus convection-permitting simulations is compared. A process-understanding methodology and a local-to-regional approach are used. Our results show a beneficial impact on the seasonal scale at both model grid spacings improving the representation of the chain of processes leading to heavy precipitation, contrary to the non-systematic improvement at the event and sub-event scales. The correction of atmospheric moisture entails a reduction of about 10% in the total column water vapour and corrections on single locations up to 10 mm counteracting the model wet bias across scales. The location, structure, and amount of total precipitation are positively affected. Particularly, the combination of high-resolution atmospheric humidity observations and fine convection-permitting simulations shows great potential for correction of the precipitation daily cycle, key for accurate precipitation modelling. The difference in the density of local and upstream observational networks and the lack of information on the vertical stratification of moisture are identified weaknesses, which could be determinants in obtaining more accurate corrections on seasonal to sub-seasonal scales after assimilation strategies.