J1.18 ATMOSPHERE-OCEAN-ICE INTERACTION PROCESSES IN THE GULF OF ST. LAWRENCE : NUMERICAL STUDY WITH A COUPLED MODEL Philippe Gachon 1* , François J. Saucier 1 , and René Laprise 2 1 Maurice Lamontagne Institute, Department of Fisheries and Oceans, Québec, Canada 2 Earth and atmosphere sciences department, University of Québec at Montréal, Québec, Canada 1. INTRODUCTION The * accurate representation of sea ice thickness distribution in numerical model is a critical challenge for global or regional climate simulations (e.g., Allison et al., 2000). In coastal regions of eastern Canada, the winter climate is strongly influenced by the energetic exchanges between the atmosphere and the oceanic surface according to the seasonal sea ice cover. The presence of ice modifies the air-sea exchanges of momentum, heat and mass relative to the open ocean, which in turn influence the regional atmospheric circulation, especially in the Labrador Sea-Hudson Bay sector (Serreze, 1995; Gachon et al., 2001), and also in the Gulf of St. Lawrence (GSL; Gachon and Saucier, 2001). This influence must be studied with particular interest, especially in wintertime, because the Baffin Bay-Labrador Sea-GSL region shows intense cyclonic activity (Serreze et al., 1993; Serreze, 1995) extending from the Labrador Sea/Davis Strait sector toward the Iceland and Norvegian seas. Recent work of Parkinson et al. (1999) indicates that a major part of the Arctic and subarctic basins shows a large decrease (in annual mean) in the sea ice extent over the 1978-1996 period, except in the Baffin Bay-Labrador Sea, Bering Sea, and Gulf of St. Lawrence, where more severe conditions were observed. However, among these regions, only the GSL positive anomaly in sea ice extent is being statistically significant at the 99% level. Moreover, these authors have suggested that the wintertime changes in sea ice extent in the GSL and Labrador Sea do not reveal a close correspondence. In order to increase our knowledge about the natural climatic variability in the northwestern Atlantic sector, the study of the interactions between these two regions, both in the ocean and atmosphere, must be better documented. The aim of this paper is to get a better understanding of the effect of interactive processes between the atmosphere-ocean-sea ice in the GSL on the regional atmospheric circulation, and to demonstrate the importance of the coupling between atmospheric and oceanic models in the GSL region. To realize this, we compare two simulations performed over a period of seven days (between 1 st and 8 th January 1990). Firstly, an off-line coupling (coupled run) every 24 hours has been realized between an atmospheric model (the * Corresponding author address : Philippe Gachon, Maurice Lamontagne Institute, Ocean Sciences Division, Department of Fisheries and Oceans, 850 route de la mer, Mont-Joli, Québec, Canada, G5H 3Z4; e-mail : gachonp@dfo-mpo.gc.ca. Canadian Regional Climate Model, hereafter the CRCM) and the oceanic model of Saucier et al. (2001). Secondly, prescribed constant oceanic conditions in the GSL have been used for the whole run (fixed run). The coupling consists of alternative runs with the CRCM and GSL models, and exchanges of fields between these two models every day. First, we describe briefly the experimental setup of the simulations using the coupled model. We present the results of the two simulations in terms of atmospheric fields’ differences during the simulation. We focus on the differences in air-sea exchanges between individual realizations. 2. EXPERIMENTAL DESCRIPTION, MODEL SETUP The two simulations with the CRCM (recently described in Caya and Laprise, 1999) have been realized with a multiple nesting procedure from 60, to 30 and then 15 km horizontal resolution grid. The NCEP (National Center for Environmental Prediction) analyzes have been interpolated to drive the lateral boundary conditions of the 60 km grid. This 60 km grid covers a 4800 × 4800 km domain centered over the GSL (48°N, 61°W). Outputs of this first simulation (archived every 3h) drive a new CRCM simulation at a higher resolution (30 km) on a smaller grid (3600 × 3600 km). The results of this second simulation (archived every 1h) drive a final CRCM simulation at an even higher resolution (15 km) on a smaller grid (2400 × 2400 km; Fig. 1). The choice of a horizontal resolution of 15 km has been motivated by the preliminary study of Roy et al. (1999) who show that the sea ice drift in the GSL is improved with an increased wind resolution of 35 to 10 km, allowing to resolve the effect of the topography near the GSL. In this finer mesh CRCM grid (Fig. 1), the model timestep is 5 min and includes 30 levels on the vertical (staggered Gal-Chen levels, e.g. terrain-following vertical coordinate, Caya and Laprise, 1999) between the surface and near 20 km high (with 15 levels from the surface up to 2500 m high). This version with the highest resolution is used to produce the results presented here. The two simulations were produced for the same period during the first week of January 1990. The coupling consists of alternative runs with the CRCM and GSL models, with daily exchanges of boundary fields between these two models. First, the oceanic model is run for December and forced by atmospheric fields issued from CRCM outputs at 60 km. On January 1 st , we used the surface temperature (sea surface temperature, and ice surface temperature), and sea ice concentration and thickness archived from this run, to