Following our recent results of assimilation of surface observations in a column model with and ensemble filter over the South Great Plains, we investigate the performance of similar methods in an area dominated by a sea-breeze circulation and assess probabilistic skill under different weather scenarios.

Surface observations of temperature, winds and humidity are assimilated with an ensemble filter into a column model of the planetary boundary layer (PBL). The column model is initialized and forced by profiles derived from either a climatology of mesoscale model forecasts or from recent mesoscale model forecasts. Advection is considered, and the advection speed is dynamically tuned with the surface observations to simulate the effect of assimilating data in three spatial dimensions.

Analysis and short term forecasts of profiles of temperature, humidity and winds in the PBL are verified against radiosondes. Similar verification of full 3D WRF model forecast profiles (at a horizontal resolution of a few kilometers) shows that the ensemble filter assimilation of the surface observations into the column model has a positive impact up to a height of 1000 m in certain regimes and weather conditions, independent of the 3D dynamics present in the 3D mesoscale model.

It is shown that using surface observations in an ensemble assimilation system with a single column model can lead to significant improvement over mesoscale model forecasts. PBL state estimates can be potentially valuable in nowcasting applications. Added benefit is realized by considering the entire analysis and very short-range forecast distributions probabilistically.