1. Introduction
Air–sea interaction has long been recognized as an important factor in the development of mesoscale structures in frontal oceanic regions because the ocean is an enormous source of heat and moisture. Intense thermal gradients frequently form within the marine atmospheric boundary layer (MABL) in coastal regions and in frontal oceanic regions, such as the Gulf Stream (GS) and the Kuroshio Currents, in response to the large differential surface energy fluxes (Bosart et al. 1972; Bosart 1975; Doyle and Warner 1990; Giordani et al. 1998).
Before a complete understanding is achieved of how the MABL fluxes force and respond to marine cyclogenesis and boundary layer fronts in presence of sea surface temperature (SST) fronts, the general problem of the differential surface forcing of the midlatitude MABL must be studied. Warner et al. (1990) found that the surface turbulent fluxes of sensible and latent heat had a significant influence on mesoscale structures in the MABL off the North Carolina coast. Similarly, Rogers (1989) concluded from the FASINEX experiment that a SST discontinuity can create large differences in surface turbulent fluxes of heat, moisture, and momentum. The resulting flux gradients can induce solenoidal mesocirculations in the MABL around the GS such as those found by Raman and Riordan (1988) and Warner et al. (1990); they were able to discern a narrow convergence line along the GS during the GALE experiment. More recently, Kwon et al. (1998) showed that the thermal and turbulent differences present in the MABL during the SEMAPHORE experiment (1993) (Eymard et al. 1996) were due to a SST front. In order to understand the MABL response to SST gradients along coastlines, Walsh (1974), Physick (1976), Arritt (1987, 1993), Wai (1988), and Bechtold et al. (1991), have successfully simulated the circulation in the MABL with 2D boundary layer models. These authors have approached the real case with a superimposition of a large-scale flow that is constant in time and space. Their results showed that besides differential warming by the surface fluxes, atmospheric wind forcings can act on the horizontal thermal gradient and therefore feed back to the circulation itself. Giordani et al. (1998) confirmed these conclusions and have shown that despite SST gradients of only 1°C 100 km−1, a SST front is able to induce significant ageostrophic circulations in the MABL.
Analyzing three dimensional nonhydrostatic simulations, Doyle and Warner (1993) showed that the differential heating induced by the SST front in the GALE experiment, plays an important role in the frontogenesis or frontolysis processes. The SEMAPHORE experiment over the Azores Current showed weaker SST discontinuities in comparison with the ones of the GALE experiment. Nevertheless, even in such weak surface forcings, Giordani and Planton (2000) showed with nonhydrostatic simulations that the ageostrophic circulation and the turbulent heat fluxes were strongly coupled in the first 200 m above the surface. This layer can be assimilated to an internal boundary layer for the flow. This interregulation works in such a fashion to minimize the atmospheric thermal wind imbalance through an adaptation of the atmospheric flow and the surface turbulent heat fluxes.
The existence of a relationship between the cyclogenesis process and the surface and MABL fluxes has been investigated for many years, but this relationship still remains unclear. The climatologies of cyclones (Austin 1941; Petterssen 1941; Colucci 1976) and explosively deepening cyclones (Roebber 1984; Sanders 1986) show distinct frequency maxima near the Atlantic coastline or over the GS. The dynamic mechanisms responsible for this cyclogenesis frequency maximum have not been conclusively identified; however, it is clear that the surface forcing must operate through modification of the MABL. Specifically, the low-level thermal field is destabilized over the warm water, and a low-level baroclinic zone develops at the boundary between the warm GS waters and the cooler Labrador waters. Additionally, surface moisture fluxes are enhanced by large air–sea temperature differences over the warm water. This differential modification of the MABL thermal and moisture fields also forces local ageostrophic circulations that eventually are reflected in the low-level pressure, divergence, and vorticity fields (Wai and Stage 1989; Warner et al. 1990; Sublette and Young 1996). Three-dimensional modeling studies have quantified the importance of some of these particular mechanisms for specific cases. For example, Nuss and Anthes (1987) found that surface fluxes were important in the development of an idealized maritime cyclone, but the specific effect depended on whether the surface heat and moisture fluxes enhanced or diminished the low-level baroclinicity. Anthes et al. (1983) also found in a modeling study of the Queen Elizabeth II storm of 9–10 September 1978 that the fluxes of heat and moisture over the GS were important for the development of this Atlantic storm, even though the large-scale baroclinic forcing was dominant. Bosart (1981) studied the Presidents' Day snowstorm case and he concluded that the role of the surface heating and moistening in cyclogenesis is controversial because of the complex link existing between the surface fluxes, the MABL characteristics (depth, stability, and baroclinic structure), the convection, and the ageostrophic circulation.
The complex influence of the surface turbulent fluxes—and more generally the turbulent fluxes in the MABL—on the cyclone properties and the cyclogenesis, is not a resolved question. This is why this paper aims at studying the sensitivity of deepening cyclones with respect to the SST. The Intensive Observing Period 15 (IOP15) case of the FASTEX experiment (1997) (Joly et al. 1999) was selected for two main reasons: IOP15 was well sampled and the cyclone crossed the SST front associated with the North Atlantic Current (NAC) from the cold to the warm side. A strong decrease of the surface pressure was observed when the cyclone crossed the oceanic front. This study was conducted with a nonhydrostatic mesoscale model well suited to simulate the evolution of baroclinic waves. Three simulations of 36 h each were performed between 1200 UTC 13 February and the 0000 UTC 15 February: the first one with a realistic SST field describing in detail the oceanic fronts associated with the GS and the NAC systems, the second one with a uniform cold SST equal to 3°C, and the third one with a uniform warm SST equal to 15°C. These uniform SSTs correspond to the mean temperatures on the cold and the warm sides of the NAC, respectively. These simulate the extreme surface conditions for a cyclone traveling only on the cold or warm side of the oceanic front. These important changes of environment inducing possible extreme cyclogenesis should help us to better understand the role of SST and surface fluxes in the dynamics of oceanic cyclones.
2. The case study
a. General context
The major objective of the Front and Atlantic Storm Tracks Experiment (FASTEX; Joly et al. 1999) was to document the life cycle of midlatitude cyclones over the Atlantic during January–February 1997. The Couplage avec 1'Atmosphère en Conditions Hivernale experiment (CATCH; Eymard et al. 1999) was the oceanic component of FASTEX. CATCH was performed in the Newfoundland Basin near 47°N, 40°W, which is a region characterized by the confluence between the cold Labrador current and the warm NAC (Rossby 1996). The in situ experimental and operational oceanic data collected during the 2 months of FASTEX provided a particularly high and well distributed spatial density of SST data over the entire domain. These data allowed the production of realistic SST maps of the oceanic front associated with the NAC, which were used in this study. CATCH was devoted to the study of the surface turbulent fluxes under strong winds related to the passage of atmospheric fronts, and the influence of the strong SST gradients (0.4°C km−1) associated with the NAC on fluxes. Since SST gradients are maintained throughout the entire year, even in winter, this surface oceanic structure results in a favorable environment to trigger atmospheric cyclogenesis, explosively deepening cyclones, and frontal waves. Here the strong SST gradients associated with the NAC vigorously enhance the baroclinicity in the MABL. For example, Sanders and Gyakum's (1980) dynamic climatology of explosive cyclogenesis identifies the north wall of the GS as a preferred area for rapid development. This is why the Newfoundland Basin was investigated by aircraft, ships, and drifting buoys.
3. The model
In order to conduct a complete analysis of the sensitivity of the IOP15 cyclone to the prescribed SSTs, numerical simulations were performed with the mesoscale nonhydrostatic model Meso-NH (Lafore et al. 1998). Due to its numerous possibilities, the model is well suited to study baroclinic waves and shallow and deep convection (Mallet et al. 1999). This model was implemented in the FASTEX area where the horizontal SST gradients induced by the NAC are strong. The size of the simulation domain is about 2752 km × 2224 km (25°–60°W and 35°–55°N). The horizontal resolution is 20 km in the central part of the domain, and the vertical grid has 50 levels between the surface and 17 000 m. The vertical resolution adopted is 40 m on average in the lower layers and 900 m near the top of the domain (17 000 m).
A complete description of the Meso-NH model may be found in Lafore et al. (1998). We only summarize here the main characteristics of the physical parameterizations and of the treatment of boundary layer conditions adopted for our study. The vertical mixing is parameterized using the 1.5-closure assumption of Bougeault and Lacarrère (1989) applied to the eddy kinetic energy prognostic equation. A microphysical scheme for the boundary layer clouds and the atmospheric ice (Caniaux 1993) is activated because of the presence of “cold” and “warm” clouds in the perturbation. This parameterization includes five prognostic variables: water vapor, cloud liquid water, cloud ice water, precipiting water, and precipiting ice. A subgrid-scale condensation scheme coupled to the turbulence parameterization is used in order to have a realistic behavior of the atmospheric parameters between cloudy and cloudless grid elements. The Meso-NH model uses the Morcrette (1991) radiative scheme, which runs every 30 min for the clear sky columns and every 10 min for the cloudy columns, because there is a strong coupling between radiative and turbulent processes within the clouds. The time step of the Meso-NH model is 30 s.
One of the main difficulties in modeling a limited area of the atmosphere is the treatment of the open lateral boundaries. Newtonian relaxation and the radiative technique are the two schemes that can be used in the model to propagate the information from the lateral boundaries into the inner domain. Several tests have shown that the radiative technique is more suitable than Newtonian relaxation for our case, because it does not generate reflexions of gravity waves on the lateral boundaries or spurious shortwave disturbances. The radiative technique propagates the scalar prognostic variables through the lateral boundaries by means of the advection (Klemp and Wilhelmson 1978), and the normal velocity component used in this advection is computed using a general Sommerfeld equation proposed by Carpenter (1982).
A horizontal numerical diffusion is applied to the prognostic variables in order to filter the numerical waves induced by the temporal scheme. The horizontal diffusion coefficient is prescribed as a characteristic time for the fourth-order numerical diffusion. This time is kept equal to 2 h because it allows the filtering of the numerical noise in both the boundary layer and the free atmosphere.
a. Initialization and lateral boundary conditions
Atmospheric analyses are performed by the French operational weather forecast atmospheric model Action de Recherche, Petite Echelle, Grande Echelle (ARPEGE; Courtier et al. 1991) and are available every 6 h.
Meso-NH was initialized with the ARPEGE atmospheric analysis of 1200 UTC 13 February 1997.ARPEGE atmospheric analyses were linearly interpolated to each time step of the Meso-NH model in order to force the lateral boundaries of the domain.
In order to force the atmospheric model Meso-NH at the bottom, realistic large-scale SST maps were used. ARPEGE SST analyses are not adequate because the major SST Front associated with the NAC is very roughly depicted. Consequently, it was necessary to perform SST reanalyses using in situ measurements in order to more realistically describe the SST features in this region. SST reanalyses were performed in the domain with the same resolution of the model, using all available data collected during the months of January and February (1997). These reanalyses are detailed and validated in Eymard et al. (1999) and Caniaux et al. (2001). The reanalyzed SST finely describe the mesoscale structures of the front linked to the NAC and they capture the SST gradients (0.4°C km−1, Fig. 1), which are much more realistic than in the operational analyses.
4. Synoptic overview
The Northwest Atlantic Basin was well sampled during FASTEX and the IOP15 (13–15 February) was chosen because cyclone maturity was achieved upon crossing the NAC oceanic front from the cold to the warm waters.
The infrared image (Fig. 2) clearly displays the signature of the upper-level wave at 1200 UTC 14 February in the middle Atlantic. The low greatly deepened between 0600 and 1200 UTC, and Fig. 2 shows the standard characteristics of a mature cyclone. Cloud areas in white in Fig. 2 show a great meridional extension of the cold front from 30°N near to 50°N. The warm and humid core seems to be very active and extends from 50°N to 55°N. The discontinuity between the cold front and the wave head near 50°N is due to a dry air intrusion, which occurs in the midtroposphere. The wave-head beyond 50°N is composed of cirrus clouds along the cyclonic-shear side. This kind of cloud pattern is described as a cloud head (Monk and Bader 1988; Cammas et al. 1999) and is indicative of rapid cyclogenesis. Finally, a large dry area extends southerly from southern Greenland to 35°N behind the system. The limit of this area is delineated by a sharp edge of clouds and it has a cyclonic curvature. This limit marks the positions of sharp gradients of both potential temperature and potential vorticity at the tropopause (not shown) and represents the jet stream axis (150 kt).
The low-level structure of the synoptic wave is now analyzed in the simulation using the reanalyzed SSTs. The IOP15 cyclone underwent between 0600 and 1200 UTC 14 February a rapid surface pressure deepening of −3 hPa h−1 when crossing the oceanic front. At 1200 UTC, the low center is located over the warm waters just to the right of the oceanic front symbolized by the isotherm 10°C (Fig. 3a). Its surface pressure reaches 978 hPa and is 10 hPa higher than the low at the end of the simulation (15 February 0000 UTC). After 1200 UTC, the surface pressure fall is much more weak than −3 hPa h−1. At 1200 UTC, the wave is well developed with marked cold and warm fronts in potential temperature (Fig. 3d), specific humidity (Fig. 3e), wind, and rain intensity (Fig. 3c). The occlusion rounds the low and starts with an ascent branch at 46°N, 38°W and ends with a descent branch parallel with the SST front. At 1200 UTC, the descent branch of the occlusion and the oceanic front are thus in phase. The frontal line are less visible in the pressure field but can be located by the breaks in the isobar lines. The first break line starting from the south (44°W) up to the low marks the cold front. The oblique break line before the cold front marks the warm front. Ahead of the perturbation and in the warm sector of the wave, the wind blows from the southwest and southeast in the southern and the northern part of the domain, respectively, and supplies the system with warm and wet air. Behind the cold front, the wind veers rapidly to the northwest advecting cold and dry air.
5. Validation of the real simulation
a. Atmospheric parameters
In this section, the reanalyzed SST and simulated atmospheric parameters Θ (potential temperature), Qa (specific humidity), and Ua (wind intensity), are compared to the in situ data collected by the research vessel R/V Le Suroît, which performed continuous surface measurements in the center of the domain. This comparison covers the 36 h of the simulation and is displayed in Fig. 4. Reanalyzed SSTs are obviously in good agreement with the observed data since they were assimilated in the optimal interpolation system. In the first approximation, we can suppose the ship was fixed in comparison with the atmospheric perturbation. Horizontal profiles of the observed data (Θ, Qa, and Ua) show two discontinuities at 1200 UTC and 1800 UTC 14 February. The first discontinuity at 1200 UTC is due to the crossing of the ascent branch of the atmospheric occlusion with respect to the ship position and the second one at 1800 UTC is induced by the descent branch. These discontinuities are reasonably captured in the simulation. Nevertheless, it seems that the simulated perturbation is moving faster than in reality because the discontinuities are ahead by approximately 3 h in comparison with the observed ones.
A vertical validation of the simulated parameters Θ, Qa, and the two wind components U and V is made through the comparison with the ten high-resolution radio soundings launched from the R/V Le Suroît between 1200 UTC 13 February and 1800 UTC 14 February. The simulated parameters were interpolated in space and time to the ship position and the comparison is displayed in Figs. 5 and 6. The model structures compare very well with the radio-sounding ones up to 9000 m. This is a remarkable result because the model did not assimilate the ship radio soundings and produced a realistic solution only with its physical package. We can therefore ascribe some credit to the model that will be used in the following analysis. Nevertheless it appears that the model develops stronger gradients than the ones observed by the ship, especially for Θ, Qa, and U in the lower layers. Therefore the model tends to simulate a frontal dynamic, which is too strong. Moreover, the Θ structures show that the simulated perturbation is in advance in comparison with the ship observations as previously mentioned.
b. Turbulent and net heat budget surface fluxes
The MABL and the oceanic boundary layer (OBL) interact through the turbulent and radiative fluxes. The surface fluxes are thus important because they couple the MABL and OBL. The determination of the turbulent fluxes at different scales and by different methods was one of the main objectives of the CATCH experiment in order to study the ocean–atmosphere interactions. To achieve this objective, turbulent flux measurements were performed along the R/V Le Suroît trajectory using the dataset of mean atmospheric and surface parameters collected on board the ship. For this determination, the formulation proposed by Dupuis et al. (1997), deduced from an adjustment to the inertio-dissipative fluxes of the SOFIA/ASTEX (June 1992) and SEMAPHORE (October–November 1993), was modified using the CATCH data (Eymard et al. 1999).
The simulated sensible (H), latent (LE), and momentum (τ) turbulent fluxes and net heat budget (G) temporal runs agree generally well when compared to the ship ones (not shown). The observed discontinuities associated with the occlusion are especially well captured by the simulation (not shown). The scores of the simulated runs compared to the ship ones (Table 1), show reduced bias but high root-mean-square errors. The high dispersions are induced by the time lag between the simulated and observed perturbations but also by the different algorithms used to derive the model and the CATCH (ship) bulk surface turbulent fluxes.
6. Sensitivity to the SST
The strategy adopted to study the cyclogenesis sensitivity to the SSTs was to perform three simulations of 36 h each: the first one with the reanalyzed SST of the second half of February (Fig. 1, real experiment: ER); the second one with the uniform warm SST equal to 15°C (warm experiment: EW) and the third one with the uniform cold SST equal to 3°C (cold experiment: EC). These three SST fields were kept constant from 1200 UTC 13 February (the start of the simulation) up to 0000 UTC 15 February (the end of the simulation). This is a reasonable approximation given the slow evolution of the SST during 36 h. The warm and cold SST fields correspond to the mean temperatures on the warm and cold sides of the NAC, respectively. So, the EW and EC experiments simulate the bottom boundary conditions for a cyclone tracking on the warm and cold side of the NAC. Cyclone EW and EC are unlikely to occur but these cases are interesting because they represent two possibilities of extreme cyclogenesis in response to the extreme SSTs present in the Northwestern Atlantic.
a. Sensitivity of the low center trajectories, surface pressures, and precipitation
Figure 7 displays the low center trajectories obtained in ER, EC, and EW, respectively. One can see that the ER trajectory sharply changes direction from the northeast to the south when the low arrives in the northwest corner of the NAC at 50°N and leads the ER low to travel above warm waters. On the other hand, EC and EW trajectories are almost straight but finally the ER and EW lows reach the same area. So, in only 36 h, the ER, EC, and EW low trajectories are already different.
Figure 8 displays the SSTs and the pressures along the trajectories of the ER, EC, and EW lows as a function of time. The ER surface pressure trend increases significantly when the perturbation crosses the oceanic front, whereas the EC and EW pressure trends are quite monotonic. In EW, the falling rate of the pressure is 1.6 times greater than in EC. In ER, this rate is close to the EC one before crossing the SST front. After crossing the oceanic front, the ER surface pressure almost overtakes the EW one. This is a remarkable behavior because this crossing lasts only 7 h during which the pressure falls 17 hPa.
The total precipitation usually is split into two components: the stratiform and the convective precipitation. The stratiform and convective precipitation are distinguished by the rainfall rate. Generally, the threshold between the stratiform and the convective regime is kept equal to 5 mm h−1 in the middle latitudes (Ramos-Buarque and Sauvageot 1997). Table 2 shows the averaged precipitation over the simulated domain accumulated during the entire simulation in the three SST cases. In comparison with ER and EW, EC exhibits the weakest stratiform, convective, and total precipitation. In comparison with ER, the heavy precipitation obtained in EW is mainly produced by the convective part and not by the stratiform part. Consequently, warm SSTs significantly enhance the convective activity of the wave.
b. Structure of the atmospheric perturbation at 1500 UTC 14 February
The wind and the temperature fields at 20 m displayed in Fig. 9 show substantial differences in ER, EW, and the EC. In ER, the wave is well developed with marked temperature and wind discontinuities in the cold front, the warm front, and the occlusion. In EC, the temperature and wind horizontal gradients and the wind intensities are weaker than in ER. In EW, the cold atmospheric front develops weaker activity than in ER and EC, but has an occlusion and a much more intense low than in the other two cases.
In EC, the cold SSTs induce a decrease of the warm advection in front of the perturbation and in the warm sector in comparison with ER. So, the differential thermal advection on both sides of the cold front decrease too and thus reduce the cold front intensity.
In EW, the cold atmospheric front is almost nonexistent, but conversely as in EC, the warm SSTs act in such a way as to disrupt the cold advection behind the atmospheric cold front and therefore the horizontal temperature gradient. This cold advection is enhanced in ER because of the presence of cold SSTs in the northern part of the domain. In fact, the warm SSTs induce a decrease of the differential thermal advection across the cold front that generates its degeneration. Finally, the central part of the wave is more intense in EW than in ER because the stronger warm advection in the warm sector of the perturbation induces a vigorous extension in latitude of the atmospheric wave. In EW, only the occlusion persists at high latitudes and is intensified thanks to the strong differential thermal advections built by the cold and warm advections in the western and eastern parts of the occlusion, respectively. Convection is observed to beak out in the occlusion (not shown) where the cold air is rapidly being warmed, moistened, and destabilized by oceanic sensible and latent heat fluxes. The perturbation seems to have tropical cyclone–like characteristics in the form of the large vortex and very strong surface winds in association with the outbreak of the convection. Unlike a tropical cyclone, the distribution of convection is very asymmetric and confined in the western part (i.e., in the occlusion) of the vortex (not shown) in the region of strongest surface winds.
c. Surface net heat budget, turbulent heat, and momentum fluxes
1) Horizontal distribution
In this section, the surface net heat budget (G) and the surface turbulent momentum flux (τ) are analyzed in ER, EC, and EW, 1200 UTC 14 February, when the waves are well developed (Fig. 10).
The surface net heat flux (G) is the budget between the radiative and turbulent heat fluxes, and its spatial variability is subjected to the horizontal structure of the atmospheric parameters. The radiative G component is mainly linked to the air temperature and the cloud and rain spatial distribution, whereas the turbulent G component is mainly associated to the wind intensity and the difference between the SST and air temperature (ΔsurfT = SST − Ta).
In ER, Fig. 3f shows that the surface buoyancy fluxes are strong (≃800 W m−2) in the region of the cold-air outbreak behind the cold atmospheric front and west of the occlusion. This is mainly because of the strong positive ΔsurfT and wind-inducing strong turbulent heat fluxes in this region. This leads to strong positive values of G (1400 W m−2) (Fig. 10a) meaning that the ocean supplies a large amount of energy to the atmosphere.
In the warm sector and in the region of the ER occlusion, in the warm, moist southerly and southwesterly flow, negative values of G (−400 W m−2) are found meaning that the atmosphere supplies energy to the ocean (Fig. 10a). In this region, negative values of ΔsurfT induce negative turbulent heat fluxes (Fig. 3f) and explain the negative values of G.
Variability of these fluxes are quite comparable to those reported by Petterssen et al. (1962) for various categories of Atlantic cyclones. A big difference, however, is that Petterssen et al. (1962) typically found the largest heat fluxes only in the wake of a cyclone passage, whereas in the present case the large values are also found in the occlusion (see Figs. 3f and 10a). The strong variations of G are found across the occlusion because of strong variations of the surface buoyancy flux (Fig. 3f) induced by the SST front and the strong variations of air temperature and specific humidity: this is a particular configuration because the occlusion and the oceanic front are in phase.
In EC (Fig. 10b), weaker G frontal discontinuities than in ER (Fig. 10a) are found across the cold front and the occlusion. Negative values (−400 W m−2) are found over a large area extending from the south to the northeast of the domain and weak positive values (200 W m−2) of G are found in the northwest quarter of the domain where the atmosphere is colder than the SST. Consequently, the EC perturbation gives up energy to the ocean almost everywhere.
Here EW (Fig. 10c) is dramatically different from EC (Fig. 10b): G is positive everywhere with strong values in front of the occlusion where the ocean supplies the atmosphere with energy up to 2600 W m−2. These values are stronger than in ER because warmer SSTs produce stronger positive turbulent heat fluxes. The occlusion is very well marked with large G gradients because it represents the limit between the warm air advected from the south of the domain and the cold air displaced far to the north.
When considering the surface wind stress (τ), the maxima observed in ER, EC, and EW (Figs. 10d, e, f) are all found along the occlusion, where the instability is maximum. However these maxima are very different in the three simulations, and this reflects the strong influence of the SST on the vertical stability. Lower values (1.5 N m−2) are observed in EC as compared with ER and EW (5 N m−2). In ER and particularly in EW, the warm SSTs significantly enhance the vertical instability and the surface wind compared to EC. Moreover, horizontal gradients of τ are very intense in ER and EW between the warm air of the cyclone core and the cold air in front of the occlusion. Koracin and Rogers (1990) investigated the response of the MABL to ocean forcing and observed that cold SSTs stabilize the lower layers of the atmosphere and decouple the low-level wind from the wind aloft.
It thus appears that both i) the energetic loss in the perturbation warm sector and ii) the energetic gain behind the cold atmospheric front and in front of the occlusion generate horizontal gradients of G and τ that are strongly dependent on the underlying SST. Kuo et al. (1991) came to the same conclusion in studying the surface energy fluxes during the rapidly deepening stage of an explosive cyclone in the western Atlantic.
An important question arising from the present study is how the surface heat and momentum fluxes contribute to the frontogenesis and cyclogenesis in the three simulations ER, EC, and EW. This question was investigated from different points of view by several authors. For instance, Danard and Ellenton (1980) studied the effect of surface fluxes on a cyclone development and showed that its deepening rate is directly linked to the Laplacian of the heat transfer, with upward motion and low-level cyclonic development in the cold air outbreak behind the cold atmospheric front and downward motion and low-level anticyclonic development in the warm sector of the perturbation, as expected in our cases. This vertical motion pattern is unlikely to generate cyclone intensification as stressed by baroclinic instability theory (Haltiner 1967). Nevertheless, the problem can not be resolved by the single analysis of the surface fluxes distribution. Indeed, Bosart (1981) studied the memorable Presidents' Day snowstorm case of the middle Atlantic states, and concluded that the role of the surface heating in cyclogenesis is controversial and needs further investigation.
Although the usual pattern of low-level heating and cooling was not conducive to cyclone intensification, Mak (1998) showed that surface heat fluxes can nevertheless excite strongly unstable short waves and that surface heat fluxes can play a role in diabatic destabilization. The diabatic destabilization by surface heat flux could thus be an alternative mechanism for initiating small marine cyclones without upper-level forcing. Roebber (1989) confirmed that the larger the surface heating size and intensity were, the shorter the wavelength of maximum instability could be. Nevertheless, Roebber (1989) stressed the importance of the vertical profile of heating on the intensification rate of the cyclone. In the first approximation, concentration of heating in the lowest levels, (i.e., for low boundary layer depths) is the most favorable situation for cyclogenesis. Based in both analytical and numerical works, Mak (1998) extended this result by showing that additional unstable short modes could be induced for different combinations of static stability and heating profiles. Nevertheless, Mak (1998) found that the instability properties were somewhat more sensitive to the heating profile than to the static stability profile.
2) Vertical distribution of fluxes
The vertical distributions of the buoyancy and momentum fluxes across the occlusion at the latitude 50°N (1200 UTC) are shown in Fig. 11 for the three simulations. This configuration is particularly interesting because of the phasing of the occlusion and the oceanic front (Fig. 9). Figure 11 is a vertical cross section, starting at 48°W, west of the occlusion and ending at 38°W, east of the occlusion, at the latitude 50°N.
The role of SST on the intensity and on the spatial distribution of the turbulent fluxes in the atmosphere is particularly well marked along this cross section.
In ER, maxima and horizontal variations of the turbulent fluxes are well marked between 230 and 500 km (Fig. 11a), where the strongest SST gradients are equal to 0.35°C 10 km−1.
In EC (Fig. 11b), the smallest values of intensity and horizontal gradients are found because the cold ocean is a sink of energy for the atmospheric perturbation and tends to stabilize the MABL. In EC, the fluxes increase far from the occlusion, where the air becomes colder than the SST.
In EW (Fig. 11c), a very sharp turbulent flux front can be identified. This horizontal gradient is much stronger in EW than in ER, although there is no SST front. It is very important to notice that, in ER and in EW, different processes create the horizontal flux gradients. In ER, these gradients are both induced by differential thermal advection and the SST front. In EW, only the differential thermal advection is acting and results in stronger flux variations although the SST is uniform.
The boundary layer is deeper in EW (1200 m) than in ER (800 m) and in EC (300 m) because the warm SSTs destabilize the MABL. This destabilization explains the convective activity and the strong turbulent fluxes above the MABL up to 2200 m in EW. This is in agreement with stronger convective precipitation in EW (Table 2). Moreover, Fig. 11 shows that the vertical divergence of the buoyancy and momentum fluxes are all the more for high SSTs. Consequently, the thermodynamic and the dynamic variables should evolve more rapidly in EW than in ER and in EC.
According to the results of Roebber (1989) and Mak (1998), stronger cyclogenesis would occur in ER than in EW because of the lower boundary layer depth in ER. However, this effect is largely offset by the destabilization of the boundary layer, which is stronger in EW than in ER.
This led us to conclude that: i) all the processes involved in cyclogenesis need to be identified and ii) the link between each process and the atmospheric circulation needs to be mathematically formulated and analyzed. The next section is devoted to the study of these two questions in terms of frontogenesis and cyclogenesis.
7. Diagnoses of frontogenesis and cyclogenesis
a. The frontogenetic function
Note that the differential diabatic heating (Qth) depends on the vertical profile of heating (
In order to answer the question of how FQth, FQtw, and FQt are sensitive to the SST, we now consider the synoptic maps of the buoyancy and the advection components of the frontogenetic function in ER, EC, and EW (Fig. 12). These forcings were averaged over the first 1000 m at 1500 UTC. As the sources of frontogenesis are very different throughout the wave, we will first consider the cold front and second the occlusion.
1) The cold front
In the ER cold atmospheric front, FQth is largely negative (Fig. 12a) meaning that the buoyancy fluxes induce differential diabatic heating, which breaks the baroclinicity and therefore weakens the frontogenesis. The same conclusion was found by Petterson et al. (1962) and by Kuo et al. (1991). Here FQth is largely offset by FQtw and their budget (FQt) is positive, which implies frontogenesis: consequently, the dynamic of the front increases.
In EC (Fig. 12b), FQth is weak in comparison with ER and only FQtw has a significant contribution to FQt. Nevertheless, FQt is not so intense and organized as in ER so that the cold atmospheric front will evolve toward weaker activity than in ER.
In EW (Fig. 12c), both FQth and FQtw are weak compared to ER (Fig. 12a) and EC (Fig. 12b). Hence, the budget FQt will not increase the baroclinic intensity of the cold front and we can expect a rapid collapse of the cold atmospheric front.
2) The occlusion
In the occlusion, the sources of frontogenesis are quite different. In ER (Fig. 12a), FQth is largely positive and is a source of frontogenesis. In this region, the three-dimensional distribution of the buoyancy fluxes form a baroclinic zone in the boundary layer, which reinforces the occlusion intensity. Note that this occurs when the occlusion is located just above the SST front. Also note that the maximum values of FQth clearly have the same shape as the SST front (Fig. 12a). This result is very similar to that of Nuss and Anthes (1987), who found that a SST pattern that is in phase with the air temperature wave tends to enhance cyclogenesis. From a statistical point of view, Sanders and Gyakum (1980) established that explosive cyclogenesis occurs preferentially where SST gradients are the strongest. These results stress the specific role of SST gradients on the frontogenetic diabatic forcings. Here FQtw is also a source of strong frontogenesis in the occlusion, such that both FQtw and FQth have a contribution in the development of an intense occlusion.
In EC (Fig. 12b), the cold SSTs have a dramatic effect on the frontogenetic sources, which are much smaller than in ER. Consequently, the EC occlusion evolves in the same way as the EW cold front. It is clear that the weakness of the occlusion (as in the EW cold front) is due to small FQtw induced by weak warm (cold in the EW cold front) advection.
In EW (Fig. 12c), FQt is strongly frontogenetic due to the contribution of large positive FQtw, despite the strongly frontolytic effect of FQth. From the three simulations, the strongest baroclinicity is obtained in the EW occlusion and is due to the largest contribution of FQtw. Strong FQtw can be ascribed to the large amount of latent heat release associated with well-developed convection. The mechanism invoked here is the warm SST, which destabilizes the MABL by lowering the effective static stability and triggers deep convection. This convection enhances the low-level convergence, which increases the baroclinicity in the MABL.
In conclusion, the SSTs affect not only the buoyancy flux source of frontogenesis but also the kinematic source, meaning that the SST acts indirectly on the temperature and the horizontal wind shear. The atmospheric response is very different in the occlusion and in the cold front of ER, EC, and EW because the SSTs simultaneously modify the atmospheric stability, the convection, and the low-level baroclinicity, particularly when an oceanic front is present.
The effect of the diabatic processes on cyclogenesis is due to the complex interaction between the diabatic transfers and the dynamic flow. The next section is an attempt to further explain this connection.
b. Diagnoses of the sources of vertical velocity
Regions of Qi convergence (Di > 0) (divergence, Di < 0) are associated with regions of ascent (descent). Radiative diabatic and nonstationary processes (see Giordani and Planton 2000) have not been taken into account in the production of vertical velocity, and these terms are generally neglected (Keyser and Pecnick 1987). This equation is a powerful tool to unravel the main physical processes active in the boundary layer and to determine how the forcings (Di) are affected by the SSTs.
The sensitivity of the w forcings to the SST was investigated in simulations ER, EC, and EW over the entire simulated area and proved that the thermal (Dth and Dtw) and dynamical (Ddm and Ddag) w forcings were affected by the SSTs throughout the wave in the same way as the frontogenetic sources (see Fig. 12). The most important point is that the four components Dth, Dtw, Ddm, and Ddag of Dt are nearly of the same magnitude (not shown), meaning that several processes are active and can be used to explain the dynamical structure of the frontal wave.
Analysis of the thermal and dynamical forcings is done in Figs. 13, 14, and 15, which represent the same vertical cross section as in Fig. 11 through the occlusion. This configuration is particularly interesting to use in order to quantify the role of the oceanic front on the w forcings and consequently on the ageostrophic circulation in the occlusion.
In ER (Fig. 13), the potential temperature (Θ) (Fig. 13a), the specific humidity (Qa) (Fig. 13b), and the vertical velocity (w) (Fig. 13c) increase across the occlusion, which is in phase with the oceanic front (see Figs. 3a,b). Here, the three-dimensional distribution of the buoyancy fluxes (Fig. 11a) produces positive Dth (Fig. 13e), which is a significant component of Dt (Fig. 13d). Here Dth intensifies the occlusion like in a pure sea-breeze case. The most important terms of w production in the occlusion are Dtw and Ddag (Figs. 13f and 13h). This confirms the importance of differential advections and ageostrophic circulations in frontal dynamics (Hoskins and Bretherton 1972). The momentum forcing (Ddm) (Fig. 13g) is not very important in comparison with the kinematic (Dtw) (Fig. 13f) and the ageostrophic (Ddag) (Fig. 13h) forcings, meaning that the momentum fluxes are horizontally and vertically homogeneous. The sign of Ddm suggests vorticity spindown and spinup near the surface and at the MABL top (≃800 m), respectively. Finally note that, although the turbulent forcings Dth and Ddm are weaker than Dtw and Ddag, they contribute significantly to Dt.
In EC, the cold SST stabilizes the MABL by increasing the static stability and generates strong negative fluxes in the perturbation core (Fig. 10b). Dramatic decreases of the temperature, humidity, vertical velocity (Figs. 14a–c), wind, and buoyancy flux (Fig. l1b) gradients are observed and lead to smaller w forcings in EC (Figs. 14d–h) than in ER (Figs. 13d–h) across the occlusion. This configuration confirms that the convective activity and the warm advection in the warm sector of the perturbation collapse in EC.
In EW, the different structures of temperature, humidity (Figs. 15a,b), wind, and fluxes (Fig. 11c) contribute to very different forcing patterns than in ER and EC. The noticeable point here is that there are the strong values of Dth in the EW MABL even in the absence of an oceanic front (Fig. 15e). However, values of Dth in ER (Fig. 13e) are still greater than in EW (Fig. 15e). This is quite surprising as would be expected from the following analysis. In ER, the variation of surface fluxes are expected to be greater than in EW, because of the absence of a SST front in EW. However, this is not the case (see Figs. 11a,c), because in EW the warm advection and convective activity are much important than in ER and contribute to enhance flux variations across the occlusion. Consequently, one would expect Dth to be stronger in EW than in ER: this is clearly contrary to what is observed in Figs. 13e and 15e. The main reason is that in ER, both the SST front and the vertical distribution of the vertical fluxes contribute to stronger Dth than in EW. This point stresses the importance of the vertical distribution of fluxes in MABL, as already mentioned by Roebber (1989).
In EW, the thermal forcing Dtw (Fig. 15f) and the dynamical forcings Ddm (Fig. 15g) and Ddag (Fig. 15h) are stronger than in ER (Fig. 13). We infer that the large amount of latent heat release in the EW occlusion enhances these forcings. Kuo et al. (1991) observed a cyclone deepening by inducing strong low-level convergence and vorticity spinup by latent heat release (here these processes are represented by Dtw).
The ageostrophic forcing Ddag appears significantly stronger in EW (Fig. 15h) than in ER (Fig. 13h) specially in the low-level (≃400 m): the frontal ageostrophic circulation is effectively due to well-developed convection, that is by increased latent heat release. Mallet et al. (1999) obtained a similar result, showing that after the initial baroclinic growth, the FASTEX IOP17 cyclone evolved toward a purely diabatic regime and the frontal ageostrophic circulation of this case was mainly supported by condensation heating. This states that latent heat release and baroclinicity are in tight interaction as shown by Chen and Dell'Osso (1987) and Kuo and Reed (1988). Finally, the w forcing Ddm is the less sensitive to SSTs, which means the turbulent momentum fluxes do not play a crucial role in the IOP15 cyclogenesis.
The different structures of baroclinicity found in the ER, EC, and EW occlusions have shown considerable impact on the low-level jet (LLJ) through the thermal wind balance. Intensities of the ER, EC, and EW LLJs reached 38 m s−1 at 700 m, 26 m s−1 at 400 m, and 40 m s−1 at 1000 m above the surface, respectively (not shown). The dropsonde data and the three-dimensional variational reanalysis performed at Météo-France show a LLJ of 33 m s−1 at about 500 m and confirm the LLJ found in ER (not shown).
8. Sensitivity of the w forcings to the buoyancy flux
In order to capture the main differences of development between two pairs of cyclones [i.e., ER and EC (ERC) on one side and ER and EW (ERW) on the other side], we analyze the sensitivity of the w forcings to the surface buoyancy flux Hb. As already mentioned, Hb was chosen rather than the SST because the atmosphere responds directly to the surface energy flux and indirectly to the SST.
In ERC and ERW (Figs. 16a,b), Dtw and Ddag are the most sensitive to Hb and their evolutions allow the definition of three distinct regimes.
In ERC, these regimes coincide with the bifurcations of Dtw and Ddag on 2300 UTC 13 February and on 1200 UTC 14 February.
The first bifurcation occurs simultaneously in ERW and in ERC, but it is important to note that in ERW, the second occurs on 0700 UTC 14 February, that is 5 h before that of ERC. This shift suggests that the EW cyclone evolves more rapidly than EC, because the strong positive surface heat fluxes of EW have provided the MABL with moisture and heat in the low levels during the first regime. This result is similar to that of Fantini (1990), who found that the effect of fluxes was to build up the potential for rapid development in the precyclogenesis stage. Once this potential is realized, rapid cyclogenesis can develop.
During the first regime,
During the second regime of ERC (between 2300 UTC 13 February and 1200 UTC 14 February) and ERW (between 2300 UTC 13 February and 0700 UTC 14 February), the sensitivity of the forcings to Hb is moderate: the perturbations are still in a development stage and we do not observe a strong difference.
During the third regime (beyond 0700 UTC 14 February for ERW and 1200 UTC for ERC), Dtw and Ddag strongly depend on Hb. During this period, the perturbations reach their mature stages and the circulations, thermodynamic structures, and associated forcings diverge strongly.
Sensitivity of Dth and Ddm to Hb differ from that of Dtw and Ddag and increase roughly linearly with time. This means that the modifications of the surface buoyancy flux (Hb) induced by the different SST fields have a greater impact on the thermal and ageostrophic forcings Dtw and Ddag than in Dth. This explains why an ageostrophic circulation (or w anomaly) induced by an anomaly of surface buoyancy flux (Hb), generates stronger forcings Dtw and Ddag than turbulent forcings Dth and Ddm.
It is obvious that the higher sensitivities of
This behavior is the inverse of the one found at the box scale. Nevertheless, these two results are not in contradiction but show that the high response of the w forcing to Hb at small scales tends to be canceled at large scales: they finally produce the same mean effects in terms of vertical velocity during the first stage of the perturbation development.
Therefore, the scale at which the surface turbulent flux acts on frontogenesis is important. Kuo et al. (1991) mentioned the role of the surface turbulent fluxes during the first stage of a cyclone development, but they did not define at which scale this role is significant.
We thus conclude that the forcings Dtw and Ddag are very sensitive to Hb at small scales, in particular with higher growth rates during the first regime. This result indicates that the circulations and the thermodynamic organizations in the three systems differ rapidly during the incipient regime. At large scales (the scale of the system), these divergences merge only during the second regime, 12 h after the cyclone birth.
9. Summary and conclusions
In this paper, the sensitivity of the frontogenesis and cyclogenesis of the FASTEX IOP15 cyclone to the SST was studied. Three nonhydrostatic mesoscale simulations were performed: the first one with a realistic SST field (ER simulation) finely describing the oceanic front associated with the NAC system, and two other simulations using extreme SST fields: the first one with a uniform cold SST equal to 3°C (EC simulation) and the second one with a uniform warm SST equal to 15°C (EW simulation). The probability for a cyclone to travel only on the cold or the warm side of the NAC is certainly very weak, making EC and EW cyclones unlikely to occur and as a result, statistically unrealistic. But such extreme events cannot be excluded and their physics would be simulated as described in EC and EW. Here EC and EW were considered as an envelope of an ensemble of cyclogenesis sensitive to SST occurring during winter in the northwestern Atlantic.
The atmospheric parameters (Θ, Qa, Ua), the surface pressure, the precipitation, the turbulent fluxes, the low center trajectories, and the vertical velocity have shown very different behavior with respect to the SST. In order to understand how the SST fields affect the internal mean structures of a synoptic perturbation, a detailed analysis of the frontogenetic function and the sources of the generalized form of the w equation were performed.
In ER, the component of the frontogenetic function associated with differential thermal advections (FQtw) is an important source of baroclinicity in the frontal lines. The component associated with the differential turbulent heat fluxes (FQth) is frontolytic and tends to break the frontal dynamic, except in the occlusion when it is phased with the oceanic front.
In EC, the origin of frontogenetic sources are the same as in ER, but their intensities are much weaker. This is due to weaker warm advection in the warm sector of the perturbation and to the MABL stabilization over cold SSTs. These two processes weaken the cold front and occlusion. Nevertheless, the decrease of the frontogenetic forcings from ER to EC is not homogeneous because it mainly affects the occlusion and the wave head rather than the cold front.
In EW, the frontogenetic sources collapse in the cold front because the differential thermal advections are broken by the warm SST: this implies a vanishing of the cold front. On the other hand, in the occlusion and in the perturbation core, strong convection takes place and enhances the low-level convergence. Consequently, the frontogenesis is highly rainforced by the differential thermal advections but not by the differential turbulent heat fluxes as in ER.
The frontal dynamic sensitivity to the SST was refined by analyzing the forcings of the generalized form of the w equation in the three simulations. This tool established the complex link between diabatic and dynamic (circulation) processes, which gave an added value compared to the frontogenetic function.
The smallest w forcings were obtained in EC because cold SSTs induce stabilization of the MABL and weak differential heating by the fluxes and the advection. The greatest differences between ER and EC, and ER and EW, especially occur in the cold front and the occlusion.
Across the cold atmospheric front, the total source of w remains strong in ER whereas in EC and particularly in EW, they collapse rapidly. This behavior is confirmed by the frontogenetic function.
Across the occlusion, the strongest w total source is found in EW because the forcings were stronger in comparison with ER, except for the buoyancy flux forcing Dth. Even without an oceanic front, the EW cyclone develops strong Dth because the horizontal variability of the turbulent heat fluxes is driven by vigorous atmospheric advection and convective activity. In ER, the occlusion and the oceanic front are in phase and the forcing Dth is driven by the thermal advection and also by the SST gradients associated with the NAC. Finally, Dth is stronger in ER than in EW because the heating depth is smaller in ER than in EW: a heat confining effect is occuring here.
In order to quantify the sensitivity of the vertical velocity sources to the surface buoyancy flux (Hb), an analysis of the relative variations of the w forcings to Hb between ER and EC (ERC), and ER and EW (ERW), was performed at the scales of the simulated domain and at the resolution of the model (20 km). This study showed that at small and large scales, the thermal wind forcing (Dtw) and the thermal wind imbalance forcing (Ddag) are the most sensitive to the surface buoyancy flux variation induced by the SST change. This result means that the feedback of the atmospheric circulation induced by the surface is greater on the kinematic forcings than on the turbulent forcings.
An original point emerges from this analysis: ER, EC, and EW cyclones are identical in their first 12 h at the box scale, but at small scales their internal structures diverge strongly and rapidly. In their incipient stages of development and at large scales, the cyclones are insensitive to a surface buoyancy flux change because the small-scale w forcing differences cancel at large scales. After 12 h, increases of the w forcing sensitivity to Hb at large and small scales means that the distributions of the mass and wind fields become more and more differently structured in the three systems. At small scales, the growth rates of the differences in ERC (i.e., between ER and EC) and ERW (i.e., between ER and EW) are much greater in the incipient stage than in the second and third stages of the cyclone development. Finally, the EW cyclone diverges more rapidly from the ER than the EC one. This behavior was attributed to the strong positive surface heat fluxes because they have preconditioned the MABL by moistening and heating the low levels during the incipient phase of the cyclone development. Nevertheless, to reinforce this physical interpretation, an ensemble of simulations should be made with perturbed initial atmospheric fields in order to quantify the model error part in the w forcings sensitivity to the SST and surface buoyancy flux.
Given that the cyclone response is known under drastic SST conditions, in the future it would be interesting to study cyclogenesis under more realistic SST perturbations. For instance, a reasonable change in i) magnitude of the SST frontal gradient, ii) location of the SST front, and iii) SSTs temporal evolution are three possible axes that could be pursued. It would thus be possible to understand how these “reasonable” SST cases could take place compared to our extreme cases by analyzing the relative deviation of sources of vertical velocity (i.e., ageostrophic circulation). It should thus be possible to see whether the same processes are active or not compared to our extreme cases.
Acknowledgments
We are most grateful to Youcef Amar for his valuable technical assistance and to Aaron Boone for his help in revising this manuscript. We wish to thank the anonymous reviewers for their comments, which helped to clarify and improve this paper.
This work was supported by the Institut National des Sciences de l'Univers.
REFERENCES
Anthes, R. A., Y-H. Kuo, and J. R. Gyakum, 1983: Numerical simulation of a case of explosive marine cyclogenesis. Mon. Wea. Rev., 111 , 1174–1188.
Arritt, R. W., 1987: The effect of water surface temperature on lake breezes and thermal internal boundary layers. Bound.-Layer Meteor., 40 , 101–125.
——,. 1993: Effects of large-scale flow on characteristic features of the sea breeze. J. Appl. Meteor., 32 , 116–125.
Austin, J. M., 1941: Favorable conditions for cyclogenesis near the Atlantic coast. Bull. Amer. Meteor. Soc., 22 , 270–272.
Bechtold, P., J. P. Pinty, and P. Mascart, 1991: A numerical investigation of the influence of large-scale winds on sea-breeze- and inland-breeze-type circulations. J. Appl. Meteor., 30 , 1268–1279.
Bosart, L. F., 1975: New England coastal frontogenesis. Quart. J. Roy. Meteor. Soc., 101 , 957–978.
——,. 1981: The Presidents' Day snowstorm of 18–19 February 1979: A subsynoptic scale event. Mon. Wea. Rev., 109 , 1542–1566.
——, Vaudo, C. J., and H. J. Helsdon, 1972: Coastal frontogenesis. J. Appl. Meteor., 11 , 1236–1258.
Bougeault, P., and P. Lacarrère, 1989: Parameterization of orography-induced turbulence in a meso-beta scale model. Mon. Wea. Rev., 117 , 1872–1890.
Cammas, J. P., and and Coauthors, 1999: FASTEX IOP17 cyclone: Introductory synoptic study with field data. Quart. J. Roy. Meteor. Soc., 125 , 3393–3414.
Caniaux, G., 1993: Paramétrisation de la glace dans un modèle non-hydrostatique de nuage: Application à une ligne de grain tropicale. Ph.D. thesis, Université Paul Sabatier, Toulouse, France, 256 pp.
——, Prieur, L., H. Giordani, F. Hernandez, and L. Eymard, 2001: Observation of the circulation in the Newfoundland Basin in winter 1997. J. Phys. Oceanogr., 31 , 689–710.
Carpenter, K. M., 1982: Radiation conditions for lateral boundaries of limited area numerical models. Quart. J. Roy. Meteor. Soc., 110 , 717–719.
Chen, S. J., and L. Dell'Osso, 1987: A numerical case study of east Asian coastal cyclogenesis. Mon. Wea. Rev., 115 , 477–487.
Colucci, S. J., 1976: Winter cyclone frequencies over the eastern United States and adjacent western Atlantic. Bull. Amer. Meteor. Soc., 57 , 548–553.
Courtier, P., C. Freydier, J. F. Geleyn, F. Rabier, and M. Rochas, 1991: The ARPEGE Project at Météo-France. Proc. ECMWF Seminar, Sept. 1991, Reading, United Kingdom, European Centre for Medium-Range Weather Forecasts, 193–231.
Danard, M. B., and G. E. Ellenton, 1980: Physical influences on East Coast cyclogenesis. Atmos.–Ocean, 18 , 65–82.
Doyle, J. D., and T. T. Warner, 1990: Mesoscale coastal processes during GALE IOP2. Mon. Wea. Rev., 118 , 283–308.
——, and ——,. 1993: Nonhydrostatic simulations of costal mesobeta-scale vorticies and frontogenesis. Mon. Wea. Rev., 121 , 3371–3392.
Dupuis, H., P. K. Taylor, A. Weill, and K. Katsaros, 1997: The inertial dissipation method applied to derive turbulent fluxes over the ocean during the SOFIA/ASTEX and SEMAPHORE experiments with low to moderate wind speeds. J. Geophys. Res., 102 , 115–129.
Eymard, L., and and Coauthors, 1996: Study of the air–sea interactions at the mesoscale: The SEMAPHORE experiment. Ann. Geophys., 14 , 986–1015.
——, and Coauthors, 1999: Surface Fluxes in the North Atlantic Current during the CATCH/FASTEX. Quart. J. Roy. Meteor. Soc., 125 , 3563–3599.
Fantini, M., 1990: The influence of heat and moisture fluxes from the ocean on the development of baroclinic waves. J. Atmos. Sci., 47 , 840–855.
Giordani, H., and S. Planton, 2000: Modeling and analysis of ageostrophic circulation over the Azores oceanic front during the SEMAPHORE experiment. Mon. Wea. Rev., 128 , 2270–2287.
——, ——, Bénech, B., and B. H. Kwon, 1998: Atmospheric boundary layer response to sea surface temperatures during the SEMAPHORE experiment. J. Geophys. Res., 103 , (C11). 25047–25060.
Haltiner, G. J., 1967: The effects of sensible heat exchange on the dynamics of baroclinic waves. Tellus, 19 , 183–198.
Hoskins, B. J., and F. P. Bretherton, 1972: Atmospheric frontogenesis models: Mathematical formulation and solution. J. Atmos. Sci., 29 , 11–37.
——, Draghici, I., and H. C. Davies, 1978: A new look at the w-equation. Quart. J. Roy. Meteor. Soc., 104 , 31–38.
Joly, A., and and Coauthors, 1999: Overview of the field phase of the Front and Atlantic Storm-Track Experiment (FASTEX) project. Quart. J. Roy. Meteor. Soc., 125 , 3131–3163.
Keyser, D., and M. J. Pecnick, 1987: The effect of along temperature variation in a two-dimensional primitive equation model of surface frontogenesis. J. Atmos. Sci., 44 , 577–604.
Klemp, J. B., and R. B. Wilhelmson, 1978: The simulation of three-dimensional convective storm dynamics. J. Atmos. Sci., 35 , 1070–1096.
Koracin, D., and D. P. Rogers, 1990: Numerical simulations of the response of the marine atmosphere to ocean forcing. J. Atmos. Sci., 47 , 592–611.
Kuo, Y. H., and R. J. Reed, 1988: Numerical simulation of an explosively deepening cyclone in the eastern Pacific. Mon. Wea. Rev., 116 , 2081–2105.
——, ——, and Low-Nam, S., 1991: Effects of surface energy fluxes during the early development and rapid intensification stages of seven explosive cyclones in the western Atlantic. Mon Wea. Rev., 119 , 457–476.
Kwon, B. H., B. Bénech, D. Lambert, P. Durand, A. Druilhet, H. Giordani, and S. Planton, 1998: Structure of the marine atmospheric boundary layer over an oceanic thermal front: SEMAPHORE experiment. J. Geophys. Res., 103 , (C11). 25159–25180.
Lafore, J. P., and and Coauthors, 1998: The Meso-NH atmospheric simulation system. Part I: Adiabatic formulation and simulations. Ann. Geophys., 16 , 90–109.
Mak, M., 1998: Influence of surface sensible heat flux on incipient marine cyclogenesis. J. Atmos. Sci., 55 , 820–834.
Mallet, I., J. P. Cammas, P. Mascart, and P. Bechtold, 1999: Effects of cloud diabatic heating on the early development of the FASTEX IOP17 cyclone. Quart. J. Roy. Meteor. Soc., 125 , 3439–3467.
Monk, G. A., and M. J. Bader, 1988: Satellite images showing the development of the storm of 15–16 October 1987. Weather, 43 , 130–135.
Morcrette, J. J., 1991: Radiation and cloud radiative properties in the ECMWF forecasting system. J. Geophys. Res., 96 , 9121–9132.
Nuss, W. A., and R. A. Anthes, 1987: A numerical investigation of low-level processes in rapid cyclogenesis. Mon. Wea. Rev., 115 , 2728–2743.
Petterssen, S., 1941: Cyclogenesis over southeastern United States and the Atlantic coast. Bull. Amer. Meteor. Soc., 22 , 269–270.
——, Bradbury, D. L., and K. Pedersen, 1962: The Norwegian cyclone models in relation to heat and cold sources. Geophys. Publ., 24 , 243–280.
Physick, W., 1976: A numerical model of the sea-breeze phenomenon over a lake or gulf. J. Atmos. Sci., 33 , 2107–2135.
Raman, S., and A. J. Riordan, 1988: The genesis of Atlantic Lows Experiment: The planetary-boundary-layer subprogram of GALE. Bull. Amer. Meteor. Soc., 69 , 161–172.
Ramos-Buarque, S., and H. Sauvageot, 1997: The estimation of rainfall in the sahelian squall line by the area-threshold method. Atmos. Res., 43 , 207–216.
Roebber, P. J., 1984: Statistical analysis and updated climatology of explosive cyclones. Mon. Wea. Rev., 112 , 1781–1794.
——,. 1989: The role of surface heat and moisture fluxes associated with large-scale ocean current meanders in maritime cyclogenesis. Mon. Wea. Rev., 117 , 1676–1694.
Rogers, D. P., 1989: The marine boundary layer in the vicinity of an ocean front. J. Atmos. Sci., 46 , 2044–2062.
Rossby, H. T., 1996: The North Atlantic Current and surrounding waters: At the crossroads. Rev. Geophys., 34 , 463–481.
Sanders, F., 1986: Explosive cyclogenesis in the west-central North Atlantic Ocean, 1981–1984. Part I: Composite structure and mean behavior. Mon. Wea. Rev., 114 , 1781–1794.
——, and Gyakum, J. R., 1980: Synoptic-dynamic climatology of the bomb. Mon. Wea. Rev., 108 , 1589–1606.
Sublette, M., and G. S. Young, 1996: Warm-season effect of the Gulf Stream on mesoscale characteristics of the atmospheric boundary layer. Mon. Wea. Rev., 124 , 653–667.
Wai, M. M., 1988: Modeling the effects of the spatially varying sea-surface temperature on the marine atmospheric boundary layer. J. Appl. Meteor., 27 , 5–19.
——, and Stage, S. A., 1989: Dynamical analyses of marine atmospheric boundary layer structure near the Gulf Stream oceanic front. Quart. J. Roy. Meteor. Soc., 115 , 29–44.
Walsh, J. E., 1974: Sea breeze theory and applications. J. Atmos. Sci., 31 , 2012–2026.
Warner, T. T., M. N. Lakhtakia, J. D. Doyle, and R. A. Pearson, 1990: Marine atmospheric boundary layer circulations forced by Gulf Stream sea surface temperature gradients. Mon. Wea. Rev., 118 , 309–323.
Reanalyzed SSTs (°C) from the second half of Feb 1997
Citation: Monthly Weather Review 129, 6; 10.1175/1520-0493(2001)129<1273:SOCTSS>2.0.CO;2
Infrared composite image of Goes and Meteosat for 1200 UTC 14 Feb 1997. Symbol L represents the IOP15 low center
Citation: Monthly Weather Review 129, 6; 10.1175/1520-0493(2001)129<1273:SOCTSS>2.0.CO;2
Simulated (a) surface pressure (hPa), (b) wind at 20 m (m s−1), (c) surface rainfall rate (mm h−1), (d) potential temperature at 20 m (Θ) (°C), (e) specific humidity at 20 m (Qa) (g kg−1), and (f) surface buoyancy flux (BUO) (W m−2), for 1200 UTC 14 Feb. The oceanic front is symbolized by the isotherm 10°C (solid line) in the surface pressure field
Citation: Monthly Weather Review 129, 6; 10.1175/1520-0493(2001)129<1273:SOCTSS>2.0.CO;2
Simulated (solid line) and observed (star) temporal runs of the (a) SST (°C), (b) the potential temperature (°C), (c) the specific humidity (g kg−1) and (d) the wind speed (m s−1) between 1200 UTC 13 Feb and 0000 UTC 15 Feb. The simulated parameters are interpolated to the ship position.
Citation: Monthly Weather Review 129, 6; 10.1175/1520-0493(2001)129<1273:SOCTSS>2.0.CO;2
Potential temperature (Θ) (°C) and specific humidity (Qa) (g kg−1) (a) observed (ship) and (b) simulated (model) as a function of time. Time is in Julian days and starts 1200 UTC 13 Feb (44.5) and ends 1800 UTC 14 Feb (45.75)
Citation: Monthly Weather Review 129, 6; 10.1175/1520-0493(2001)129<1273:SOCTSS>2.0.CO;2
(a) Observed (ship) and (b) simulated (model) components U, V of the wind (m s−1), and the wind speed (m s−1) as a function of time. Time is in Julian days and starts 1200 UTC 13 Feb (44.5) and ends 1800 UTC 14 Feb (45.75)
Citation: Monthly Weather Review 129, 6; 10.1175/1520-0493(2001)129<1273:SOCTSS>2.0.CO;2
Low center trajectories in ER, EC, and EW between 1200 UTC 13 Feb and 0000 UTC 15 Feb
Citation: Monthly Weather Review 129, 6; 10.1175/1520-0493(2001)129<1273:SOCTSS>2.0.CO;2
The ER (solid line), EC (dashed line), and EW (long-short dashed line) (a) SST (°C) and (b) surface pressure (hPa) along the low center trajectories between 1200 UTC 13 Feb and 0000 UTC 15 Feb
Citation: Monthly Weather Review 129, 6; 10.1175/1520-0493(2001)129<1273:SOCTSS>2.0.CO;2
The ER, EW, and EC wave structure for 1500 UTC 14 Feb. Super-imposition of the wind direction and intensity at 20 m (arrow), the potential temperature at 20 m (dashed line), and the oceanic front in ER (10°C) isotherm is represented by the thick solid line)
Citation: Monthly Weather Review 129, 6; 10.1175/1520-0493(2001)129<1273:SOCTSS>2.0.CO;2
Surface net heat budget (G) (W m−2) in (a) ER, (b) EC, (c) EW, and surface stress (τ) (N m−2) in (d) ER, (e) EC, and (f) EW, for 1200 UTC 14 Feb
Citation: Monthly Weather Review 129, 6; 10.1175/1520-0493(2001)129<1273:SOCTSS>2.0.CO;2
Vertical section across the occlusion (48°–38°W at the latitude 50°N) of the turbulent buoyancy flux (W m−2) and turbulent momentum flux (N m−2) in (a) ER, (b) EC, and (c) EW, for 1200 UTC 14 Feb
Citation: Monthly Weather Review 129, 6; 10.1175/1520-0493(2001)129<1273:SOCTSS>2.0.CO;2
Vertical average (0, 1000 m) of the frontogenetic function FQt and its components FQth and FQtw (K2 100 km−2 day−1) in (a) ER, (b) EC, and (c) EW for 1500 UTC 14 Feb.
Citation: Monthly Weather Review 129, 6; 10.1175/1520-0493(2001)129<1273:SOCTSS>2.0.CO;2
The ER vertical distributions of (a) Θ (°C), (b) Qa (g kg−1), (c) w (cm s−1), and the w forcings (d) Dt, (e) Dth, (f) Dtw, (g) Ddm, (h) Ddag, (K 100 km−2 day−1), across the occlusion (50°N) 1200 UTC 14 Feb
Citation: Monthly Weather Review 129, 6; 10.1175/1520-0493(2001)129<1273:SOCTSS>2.0.CO;2
The EC vertical distributions of (a) Θ (°C), (b) Qa (g kg−1), (c) w (cm s−1), and the w forcings (d) Dt, (e) Dth, (f) Dtw, (g) Ddm, (h) Ddag, (K 100 km−2 day−1) across the occlusion (50°N) 1200 UTC 14 Feb
Citation: Monthly Weather Review 129, 6; 10.1175/1520-0493(2001)129<1273:SOCTSS>2.0.CO;2
The EW vertical distributions of (a) Θ (°C), (b) Qa (g kg−1), (c) w (cm s−1), and the w forcings (d) Dt, (e) Dth, (f) Dtw, (g) Ddm, (h) Ddag, (K 100 km−2 day−1), across the occlusion (50°N) 1200 UTC 14 Feb
Citation: Monthly Weather Review 129, 6; 10.1175/1520-0493(2001)129<1273:SOCTSS>2.0.CO;2
Bias S1 and rms error S2 (K J−1 × 1012) temporal runs between (a) ER and (c) EC (erc) and between (b) ER and, (d) EW (erw) for Dth (solid line), Dtw (dot-dashed line), Ddag (short-dashed line), and Ddm (long-dashed line)
Citation: Monthly Weather Review 129, 6; 10.1175/1520-0493(2001)129<1273:SOCTSS>2.0.CO;2
Scores of the surface turbulent fluxes and the surface net heat budget simulated with the reanalyzed SSTs in comparison with the ship fluxes collected during the IOP15 (13–15 Feb 1997)
Spatial averages of the ER, EC, and EW cumulative precipitation at the end of the simulations