Impact of two-way coupling and sea-surface temperature on precipitation forecasts in regional atmosphere and ocean models

Six coupled atmosphere-ocean set-ups of ALADIN and POM limited-area models are used to simulate five heavy precipitation events over the Adriatic Sea. We evaluate several approaches to applying the one- or two-way atmosphere-ocean coupling (during the forecast of the warm-up assimilation cycle, during the main forecast, or both) and using static sea-surface temperature (SST) information in various resolutions: from POM, from MFS (a regional ocean model) or from OSTIA analysis (used also by the global model ECMWF/IFS). The set-ups are designed in a way that allows for independent evaluation of various SST sources in ALADIN and the strategy of two-way coupling. SST quality in these set-ups is verified against satellite observations. The impact on precipitation forecasts is quantified by verification over 900 stations. Results depend on the weather situation, the frequency of the update of SST and the application of two-way coupling before and during the events. When SST is used statically in ALADIN, operational-like forecasts using daily SST analysis from OSTIA are more accurate than those using MFS and especially POM; this illustrates the importance of using fresh information from observations. The two-way coupling outperformed the one-way coupling in the cases with heavy and localized convection as the dominating process. The skill of the two-way coupled experiments was similar in situations with large-scale synoptic forcing. It was found that including the two-way coupling in the assimilation cycle a few days before the studied weather events impacts the SST forecast but has neutral impact on precipitation scores. Results suggest that ocean data assimilation is necessary in the two-way coupled system in order to realistically update SST in the system with fresh observations.