Evaluation of convection-resolving models using satellite data: The diurnal cycle of summer convection over the Alps

• Published in: Meteorologische Zeitschrift (Berlin, Germany : 1992) Vol. 25; no. 2; pp. 165 - 179
• Main Authors: Keller, Michael, Fuhrer, Oliver, Schmidli, Juerg, Stengel, Martin, Stöckli, Reto, Schär, Christoph
• Format: Journal Article
• Language: English
• Published: Borntraeger 01.01.2016
• Subjects: Alps; cloud validation; convection; convection-resolving; diurnal cycle; satellite observations
• ISSN: 0941-2948
• DOI: 10.1127/metz/2015/0715

Diurnal moist convection is an important element of summer precipitation over Central Europe and the Alps. It is poorly represented in models using parameterized convection. In this study, we investigate the diurnal cycle of convection during 11 days in June 2007 using the COSMO model. The numerical simulations are compared with satellite measurements of GERB and SEVIRI, AVHRR satellite-based cloud properties and ground-based precipitation and temperature measurements. The simulations use horizontal resolutions of 12 km (convection-parameterizing model, CPM) and 2 km (convection-resolving model, CRM) and either a one-moment microphysics scheme (1M) or a two-moment microphysics scheme (2M).They are conducted for a computational domain that covers an extended Alpine area from Northern Italy to Northern Germany. The CPM with 1M exhibits a significant overestimation of high cloud cover. This results in a compensation effect in the top of the atmosphere energy budget due to an underestimation of outgoing longwave radiation (OLR) and an overestimation of reflected solar radiation (RSR). The CRM reduces high cloud cover and improves the OLR bias from a domain mean of −20.1 to −2.6 W/m2. When using 2M with ice sedimentation in the CRM, high cloud cover is further reduced. The stronger diurnal cycle of high cloud cover and associated convection over the Alps, compared to less mountainous regions, is well represented by the CRM but underestimated by the CPM. Despite substantial differences in high cloud cover, the use of a 2M has no significant impact on the diurnal cycle of precipitation. Furthermore, a negative mid-level cloud bias is found for all simulations.