Assimilation of surface pressure observations from personal weather stations in AROME-France

Abstract

Spatially dense surface pressure observations from personal weather stations (PWSs) are able to describe pressure patterns at the surface, such as those associated with convective events, in more detail than with standard weather stations (SWSs) only. In this study, the benefit of assimilating PWS observations with the 3DVar and the 3DEnVar data assimilation schemes of the AROME-France model is evaluated over a 1-month period and during a heavy precipitation event in the South of France. Observations of surface pressure from PWSs are bias-corrected, quality-controlled, and thinned with a spacing equal to the horizontal dimension of an AROME-France grid cell. Over France, almost half of the 55 187 available PWS observations are assimilated, which is 129 times more than the number of assimilated SWS observations. Despite the limited advantages found from their assimilation with the 3DVar assimilation scheme, the 3DEnVar assimilation scheme shows systematic improvement and reduces by 10.3 % the root-mean-square deviation in surface pressure between 1 h model forecasts and SWS observations over France. Significant improvement is observed over the first 9 h of the forecasts in mean sea level pressure. Finally, when PWS observations are assimilated with the 3DEnVar assimilation scheme, a surface pressure anomaly generated by a mesoscale convective system – observed by PWSs and not visible without them – is successfully assimilated. In that case, the forecasts of location and temporal evolution of the mesoscale convective system as well as rainfall are closer to the observations when PWS observations are assimilated.