Sensitivity of Simulated Sea Breezes to Initial Conditions in Complex Coastal Regions

Abstract

Abstract Mesoscale simulations of sea breezes are sensitive to the analysis product used to initialize the simulations, primarily due to the representation of the coastline and the coastal sea surface temperatures (SSTs) in the analyses. The use of spatially coarse initial conditions, relative to the horizontal resolution of the mesoscale model grid, can introduce errors in the representation of coastal SSTs, in part due to the incorrect designation of the land surface. As a result, portions of the coastal ocean are initialized with land surface temperature values and vice versa. The diurnal variation of the sea surface is typically smaller than over land on meso- and synoptic-scale time scales. Therefore, it is common practice to retain a temporally static SST in numerical simulations, causing initial SST errors to persist through the duration of the simulation. These SST errors influence horizontal coastal temperature and humidity gradients and thereby the development of the sea-breeze circulations. The authors developed a technique to modify the initial surface conditions created from a reanalysis product [North American Regional Reanalysis (NARR)] for simulations of two sea-breeze events over the New England coast to more accurately represent the finescale structure of the coastline and the spatial representation of the coastal land surface and SST. Using this technique, the coastal SST (2-m temperature) RMSE is reduced from as much as 25°–1°C (7°–1°C), contributing to a more accurate propagation of the sea-breeze front. Techniques described in this work may be important for mesoscale simulations and forecasts of other coastal phenomena.