PreVOCA aims to assess the performance of global and regional atmospheric models in simulating the southeast Pacific region including the location of the VOCALS REx field campaign. Model representation of the regional circulation and the cloudy marine boundary layer is critical to the study of chemical transport, aerosol and microphysical processes, air-sea coupling, and mesoscale cloud features (e.g. POCS) and is important for regional forecasting. We compare results from 14 models with each other and with observations. Participating models include global operational forecasting models, global GCMs and regional models.

Simulations cover the month of October 2006 and are analyzed over the region from the equator to 40S and from 70W to 110W, extending westward from the coast of South America. The models typically have 50km resolution within this area. Several coupled models participated, though most models used specified SST. Observations of cloud fraction, boundary-layer depth, liquid water path, water vapor path are available from various satellite sources including MODIS, TMI, AMSR, COSMIC, and CALIPSO. Radiative flux climatologies can be compared with ISCCP FD data. Also at 20S 85W the the NOAA ESRL cruise data are combined to provide boundary layer depth and sounding climatologies.

The models generally show good agreement in mean vertical velocity fields at 850 hPa over the study region with weak subsidence over much of the domain and much stronger subsidence (> 0.1 Pa/s) near the central Chilean coast. The climatological near-surface winds also agree well with QuikSCAT (typically within 1 m/s) over most of the domain.

However the models disagree strongly in October mean cloud fraction over the study region, with substantial variation both near and away from the coast. A few models have a cloud fraction climatology that is reasonably consistent with MODIS cloud-fraction climatology while most models have large discrepancies. The models also show a substantial disagreement in mean boundary layer depth. Along 20S where several observational comparisons are available, most models have boundary layer depth several hundred meters less than observations with especially strong negative height biases near the coast.

Day-to-day and diurnal variability in the simulations is also explored.