Atmospheric storm surge modeling methodology along the French (Atlantic and English Channel) coast

Abstract

Storm surge modeling and forecast are the key issues for coastal risk early warning systems. As a general objective, this study aims at improving high-frequency storm surge variations modeling within the PREVIMER system (www.previmer.org), along the French Atlantic and English Channel coasts. The paper focuses on (1) sea surface drag parameterization and (2) uncertainties induced by the meteorological data quality. The modeling is based on the shallow-water version of the model for applications at regional scale (MARS), with a 2-km spatial resolution. The model computes together tide and surge, allowing properly taking into account tide-surge interactions. To select the most appropriate parameterization for the study area, a sensitivity analysis on sea surface drag parameterizations is done, based on comparisons of modeled storm surges (extracted with a tidal component analysis) with four tidal gauges, during four storm events, and over about 7.5 years, where the observed water level is processed in the same way as the modeling results. The tested drag parameterizations are a constant one, as reported by Moon et al. (J Atmos Sci 61: 2321–2333, 2007), Makin (Bound-Layer Meteorol 115: 169–176, 2005), and Charnock (J Roy Meteor Soc 81: 639–640, 1955). Charnock’s parameterization, either constant with high value (0.022) or relying on a full statistical description of the sea state, enables to improve storm surges forecast with peak errors 10 cm smaller than those computed with the other drag coefficient formulations. The impact of the meteorological forcing quality is evaluated over January 2012 from the comparison between surges modeled with different meteorological data (ARPEGE, ARPEGE High Resolution and AROME) and observations. For event time scale, storm surge computation is highly improved with ARPEGE High Resolution data. For month time scale, statistics of model accuracy are less sensitive to the choice of meteorological forcing. As a conclusion, the Charnock’s parameterization is advised to model storm surges on the French Atlantic and English Channel coasts, whereas the quality requirements regarding meteorological inputs depend on the time scale of interest. Within the PREVIMER system, aiming at forecasting events, ARPEGE High Resolution data are used.