northwestern Mediterranean Sea - Springer Link

Ocean Dynamics (2012) 62:213–226 DOI 10.1007/s10236-011-0502-8

Ocean response to strong precipitation events in the Gulf of Lions (northwestern Mediterranean Sea): a sensitivity study Cindy Lebeaupin Brossier · Karine Béranger · Philippe Drobinski

Received: 20 April 2011 / Accepted: 26 October 2011 / Published online: 2 December 2011 © Springer-Verlag 2011

Abstract The Mediterranean Sea is a region of intense air–sea interactions, with in particular strong evaporation over sea which drives the thermohaline circulation. The Mediterranean region is also prone to strong precipitation events characterized by low spatial extent, short duration, and high temporal variability. The impacts of intense offshore precipitation over sea, in the Gulf of Lions which is a spot for winter deep convection, are investigated using four sensitivity simulations performed at mesoscale resolution with the eddy-resolving regional ocean model NEMO-MED12. We use various atmospheric fields to force NEMO-MED12, downscaled from reanalyses with the non-hydrostatic mesoscale Weather Research and Forecasting model but differing in space resolutions (20 and 6.7 km) or in time frequencies (daily and three-hourly). This numerical study evidences that immediate, intense, and rapid freshening occurs under strong precipitation events. The strong salinity anomaly induced extends horizontally (50 km) as vertically (down to 50 m) and persists several days after strong

Responsible Editor: Birgit Andrea Klein C. Lebeaupin Brossier (B) · P. Drobinski Laboratoire de Météorologie Dynamique (CNRS/Ecole Polytechnique/ENS/UPMC), Institut Pierre Simon Laplace, Palaiseau, France e-mail: [email protected] C. Lebeaupin Brossier · K. Béranger Unité de Mécanique, École Nationale Supérieure de Techniques Avancées (ENSTA)—ParisTech, Palaiseau, France

precipitation events. The change in the space resolution of the atmospheric forcing modifies the precipitating patterns and intensity, as well as the shape and the dynamics of the low-salinity layer formed are changed. With higher forcing frequency, shorter and heavier precipitation falls in the ocean in the center of the Gulf of Lions, and due to a stronger vertical shear and mixing, the low-salinity anomaly propagates deeper. Keywords Strong precipitation · Mediterranean Sea · Stratification · Resolution impact

1 Introduction The northwestern Mediterranean (NWE; Fig. 1c), and more specifically the Gulf of Lions, is a region of particularly intense atmospheric forcing which drives the local thermohaline circulation. Indeed, the complex surrounding orography (Massif Central, Alps, and Pyrenees) steers the atmospheric circulation at large scale producing mesoscale flow, which can evolve into intense atmospheric events. The most known effect of the atmosphere forcing on the local oceanic circulation is that of the strong local offshore winds (mistral and tramontane; see Drobinski et al. 2001, 2005; Guénard et al. 2005, 2006) which produce strong heat and momentum exchanges. These exchanges modify the thermohaline characteristics by increasing through evaporation (E) the seawater density in the ocean surface layer and consequently contribute to ocean deep convection (Marshall and Schott 1999; Lebeaupin Brossier and Drobinski 2009; Béranger et al. 2010). Compared to the effect of strong winds and associated heat fluxes, the impact of strong precipitation on

214

Ocean Dynamics (2012) 62:213–226

(a)

(b)

Fig. 1 a MED12 model grid (567 × 264 grid meshes) and bathymetry in meters. b WRF 20-km resolution domain (240 × 130 grid meshes) illustrated by the topography in meters (contours every 1,000 m). c The northwestern Mediterranean (NWE) region (WRF zoomed domain with a 6.7-km resolution) illustrated by the topography in meters (contours every 500 m) and

the main local atmospheric and oceanic circulation patterns: regional winds in pink arrows; main currents in dark blue arrows; box with green lines indicates the MEDOC box (4.25◦ E–5.45◦ E; 41.5◦ N–42.4◦ N). The green square indicates the diagnosed mooring location at 4.92◦ E–41.98◦ N

the local thermohaline characteristics in the winter convection area has received much less attention. Indeed, the NWE area is very often affected by strong precipitation (SP) events over land and sea that are triggered by mountain-induced lifting of impinging marine moist air masses. As they generally occur during the fall and winter, they directly affect the pre-conditioning phase, just before deep convection with dense water formation.

Regarding the strong inter-annual variability of the deep convection and of the dense water characteristics in the northwestern Mediterranean, one question is the role of strong precipitation events over sea and in the coastal area on their variability. SP events are crucial for salinity as they bring a large amount of freshwater in surface during a very short period. Moreover, they break mixing and a very-stratified layer with low


215

Fig. 1 (continued)

(c)

Massif Central

Alps Mistral

Gulf of Genoa Tramontane

Gulf of Lions

Pyrenees

Liguro-Provencal current

Corsica

Sardinia

Balearic Islands

salinity is immediately formed (Lebeaupin Brossier et al. 2009a, b). These SP events are also generally associated with intense evaporation over the sea because of strong onshore winds and a strong air–sea thermal contrast upstream (Ducrocq et al. 2008; Nuissier et al. 2008). This creates a very strong horizontal gradient in the freshwater flux seen by the ocean and, consequently, produces significant vertical and horizontal gradients for the upper ocean layer salinity. Because precipitation varies strongly in space and time, an accurate representation is crucial considering the fine-scale ocean response. Short-term and local effects in terms of stratification increase due to intense ocean freshwater gain were preliminary addressed by Lebeaupin Brossier et al. (2009a, b) using a simple 1D ocean model in the NWE area. In this study, the investigation goes one step forward. Indeed, sensitivity experiments using a 3D regional oceanic model are performed to evaluate the thermohaline anomalies induced by SP events in the Gulf of Lions and their propagation. Section 2 presents the ocean model and the four atmospheric forcings used. The ocean response under SP events is examined in Section 3. Section 4 presents an evaluation of the sensitivity of this response to the space and time resolutions of the atmospheric forcing, before concluding remarks in Section 5.

2 Experimental design We describe hereafter the regional ocean model of the Mediterranean Sea and the different atmospheric fields used. More details about the experimental design could be found in Lebeaupin Brossier et al. (2011).

2.1 Ocean general circulation model The numerical code for the ocean is NEMO (Madec 2008) and is used in a regional Mediterranean configuration with a 1/12◦ horizontal resolution (Fig. 1a). In the vertical, 50 unevenly spaced Z -levels are used. The vertical level thickness is 1 m in surface and nearly 4 m at 30-m depth. This eddy resolving model is referred as MED12 hereafter. The horizontal eddy viscosity coefficients is fixed to −1.25 × 1010 m2 s−1 for the dynamics (velocity) with the use of a bi-Laplacian operator. The TVD scheme is used for the tracer advection in order to conserve energy and enstrophy (Barnier et al. 2006). The vertical diffusion is performed by the standard turbulent kinetic energy model of NEMO (Blanke and Delecluse 1993), and in case of instabilities, a higher diffusivity coefficient of 10 m2 /s is used.

216


The initial conditions for 3D potential temperature and salinity fields are provided by the monthly climatology of Levitus et al. (2005). The water exchanges with the Atlantic Ocean are parameterized with a 3D relaxation to this climatology between 11◦ W–7.5◦ W. The runoffs and the Black Sea water input are prescribed from a climatology (Beuvier et al. 2010), i.e., for each river, a freshwater input is applied in surface in one grid-point. The filtered free surface of Roullet and Madec (2000) is used, and to keep the sea volume constant, the freshwater deficit of the Mediterranean Sea is reported at each time step as Atlantic Water in the buffer zone (7.5◦ W–11◦ W). MED12 is forced at the air–sea interface by heat and freshwater fluxes and wind stress. Specifically, the freshwater flux is the net budget of precipitation minus evaporation (P − E). Thus, the ocean gains water if P is higher than the local E value. 2.2 Atmospheric forcings To obtain atmospheric forcings, we use the nonhydrostatic Weather Research and Forecasting (WRF) atmospheric model (Skamarock et al. 2008) over an Europe–Mediterranean domain with a 20-km resolution (28◦ N–49.5◦ N; 13◦ W–42◦ E] (Fig. 1b) and in a NWE configuration (39.5◦ N–46◦ N; 1◦ E–9.5◦ E) of 6.7km resolution (Fig. 1c) to downscale National Centers for Environmental Prediction (NCEP) reanalyses from 1 August 1998 to 31 July 1999. Thus, the 20-km domain is driven by the NCEP reanalyses and drives the NWE domain in a one-way nesting. The output frequency is either three-hourly or daily, with the daily averaged fields computed from the three-hourly fields. During the winter 1998–1999, intense evaporation occurred leading to dense water formation and deep convection in the northwestern Mediterranean (Canals et al. 2006; Béranger et al. 2010). In the atmospheric model, the precipitation is solved by the new Kain–Fritsch implicit convection parame-

Table 1 Ten-meter wind speed (meters per second), precipitation and evaporation (millimeters per day) characteristics in the MEDOC box (Gulf of Lions) (4.25◦ E–5.45◦ E; 41.5◦ N–42.4◦ N),

Mean Maximum Standard deviation

terization (Kain 2004) and the single-moment threeclass microphysics scheme (Hong et al. 2004). In the atmospheric run with the finer space resolution, the microphysics is more efficient to explicitly represent cumulus and precipitation clouds (Gomes and Chou 2010). Table 1 summarizes the differences in wind speed, precipitation, and evaporation above the convective area of the Gulf of Lions, hereafter the MEDOC box (41.5–42.4◦ N; 4.25–5.45◦ E—Fig. 1c). In the MEDOC box, with the 20-km resolution forcing, the annual mean wind speed is 7.61 m/s. The annual evaporation budget is 1,424 mm. The annual P budget is 331 mm, with 142 mm for the fall season, 106 mm in winter, and 69 mm in spring. The summer P budget (14 mm) represents only 4% of the annual budget. In the 6.7-km resolution forcing, both annual mean and maximum of wind and evaporation are increased (Table 1). The larger standard deviation values indicate a larger variability of these fields in the zoomed domain. For precipitation, the budget is increased over the MEDOC area, whereas the maximum is decreased. The amount of precipitation is lower. Note that, on the contrary, for the whole NWE area, the precipitation budget is decreased whereas extreme rainfall values are increased (Lebeaupin Brossier et al. 2011).

2.3 Sensitivity experiments Numerical simulations are run in perpetual mode, i.e., a single-year is repeated. After a spin-up of 8 years in perpetual mode with MED12 forced by the daily 20-km WRF atmospheric fields, four sensitivity simulations are performed using previous atmospheric fields during two additional years. – –

The control experiment (CTL) is driven by daily 20km WRF fields, as in the spin-up. The ZOOMGOL experiment: MED12 is driven by daily WRF fields built by merging the 6.7-km fields

between 1 August 1998 and 31 July 1999, in the 20-km (CTL and 3HFREQ experiments) and 6.7-km resolution (ZOOMGOL and HIGHRES) forcings (three-hourly data)

20km WSPD (m/s)

P (mm/day)

E (mm/day)

6.7km WSPD (m/s)

P (mm/day)

E (mm/day)

7.61 19.71 3.83

0.91 108.40 4.82

3.90 19.30 3.83

7.75 20.15 3.96

0.94 86.94 4.76

3.96 20.95 3.92

WSPD wind speed, P precipitation, E evaporation


217

Table 2 The annual water budgets in the NWE area and in the MEDOC box in the 20-km (CTL and 3HFREQ experiments) and 6.7-km (ZOOMGOL and HIGHRES) resolution forcings Water budget P − E (mm)

20 km

6.7 km

NWE MEDOC box

−1,009 −1,093

−1,038 −1,104

– –

km fields in the NWE area and of 20-km fields elsewhere, taking care of the Davies-type zone. In the following, only the last year of the four companion experiments, referred hereafter as year 10, will be examined.

in the NWE area and of 20-km fields elsewhere, taking care of the Davies-type zone. The 3HFREQ experiment: MED12 is forced by 20km three-hourly WRF fields. The HIGHRES experiment: MED12 is forced by three-hourly WRF fields built by merging the 6.7-

3 Impacts of strong precipitation on the ocean stratification At the Mediterranean basin scale, the P − E budget is of the order of −1 m, which is in agreement with the several estimates reported in the literature (Romanou

SP2

3h - 20km 3h - 6.7km daily - 20km daily - 6.7km

105 PRECIPITATION

90

(a)

SP5

75

(mm/day)

SP3

60

SP11 SP4

45

SP1

SP7

SP6

SP8 SP12 SP9

SP10

30

15

0

15 EVAPORATION

SEP

OCT

NOV

DEC

JAN

FEB

MAR

APR

MAY

(b)

Fig. 2 Evolution of a the four freshwater forcings (precipitation and evaporation rates—millimeters per day) averaged over the MEDOC box (see Fig. 1c) between September 1998 and May

1999 and b averaged vertical salinity profiles over the MEDOC box between September and May of year 10 in CTL (bottom panel)

218

Ocean Dynamics (2012) 62:213–226 4

surface salinity maximum decrease (psu)

0

7 10

-0.1

9

5

8

1

4

3

12

0 3

duration 10

11

6

1 5

9

11

7

8

12

-0.1 6

-0.2

-0.2

-0.3

-0.3

6hrs 12hrs 15hrs 18hrs 21hrs 30hrs 36hrs 39hrs 45hrs

-0.4

-0.4 2

2

0

10

30 0

20

mean freshwater flux (mm/day)

10

20

30

40

total amount of P-E (mm)

Fig. 3 Surface salinity decrease function of the mean freshwater flux (left panel) and of total amount of freshwater (right panel) for each SP event in CTL at grid point 4.92◦ E–41.98◦ N (see green square in Fig. 1c)

et al. 2010). It corresponds to a water loss for the ocean, with a maximum in spring and a minimum in fall (see Lebeaupin Brossier et al. 2011). The P − E budgets for

the NWE area and the MEDOC box are reported in Table 2 and are also nearly −1 m with larger absolute value in the fine space resolution forcing.

(a) CTL

(b) ZOOMGOL

(c) 3HFREQ

(d) HIGHRES

38.1

38.15

38.2

38.3

38.4

Fig. 4 Hovmuller diagrams at 4.92◦ E–41.98◦ N in each experiment between 21 September and 20 November of year 10: density (gray color, grams per kilogram), temperature (white contours every 1◦ C), and salinity (color contours, practical salinity unit)


MEDOC box, this halocline is maintained by successive fall SP events and persists until 10 November at about 40 m depth (i.e., 45 days after the first SP event). It is then disrupted and mixed downstream during a moderate convection event (Figs. 2b and 4a). In December and January, low-salinity anomalies are also noticed under SP events (Figs. 2b and 3) and associated with weak mixed-layer cooling. Figure 5 presents the dense water (with density over 29 g/kg) and light water (density under 28 g/kg) trends in the Gulf of Lions in winter (December–January– February). During this period, the amount of dense water increases (positive trends) with strong evaporation (Figs. 2a and 5a). During SP events (SP6, SP7, and SP8), the dense water formation is stopped and even its volume decreases. The volume of dense water mainly decreases due to export. But, at the same time, positive trends of light water during SP events also indicate that dense water is partly replaced by lighter water because of the strong freshwater amount brought at the surface. Although the low-salinity layer is thin (40 m— Fig. 2b), these SP events in winter modify rapidly the thermohaline characteristics of the ocean mixed layer

(a) dense water (>29g/kg) 6e+11

m3/day

4e+11

CTL ZOOMGOL 3HFREQ HIGHRES

2e+11

0

DEC SP6

JAN SP7

FEB

SP8

0

m3/day

The study focuses on the SP events over the MEDOC box. This area is characterized by a cyclonic circulation, propitious to deep convection, that could moreover be a trap for ocean anomalies under intense weather events. SP events are selected with a criteria corresponding to precipitation rates over 30 mm/day, i.e., 30 times the annual mean value corresponding to the 99.9 percentile of the precipitation rate distribution. Twelve SP events are identified and referred as SP1 to SP12 (Fig. 2a). No SP event is simulated in summer (not shown). The duration of these events ranges between 6 and 45 h, with a mean value of 23 h. However, during these SP events, the most intense precipitation over 30 mm/day never last more than 12 h. The strongest SP events occur in fall (SP1 to SP5— Fig. 2a) inducing a significant freshwater gain for the ocean. The most intense event, SP2 (with maximum precipitation rate of 108 mm/day), contributes alone to a rainfall amount of 42 mm (30% of the fall P budget). During winter, SP6 to SP8 are characterized by weaker precipitation rates (less than 45 mm/day—Fig. 2a) and are also mainly associated with strong evaporation. SP events also occur in spring: SP11 and SP12 for example induce accumulated rainfall amounts of 15 and 11 mm on average in the MEDOC box, respectively. The ocean response to these SP events is first examined in CTL. Each SP event induces an immediate and strong freshening of the ocean surface. The freshening is first estimated as the differences in sea surface salinity (SSS) between the beginning and the end of the event. Figure 3 presents the SSS anomaly for each SP event as a function of the mean freshwater (E − P) flux and the accumulated amount of water received during each event. It first shows evidence that events lasting less than 24 h induce a small freshening between 0.01 and 0.05 psu, whereas events lasting more than 1 day induce a freshening between 0.1 and 0.16 psu. The highest SSS decrease is obtained with SP2 giving a freshening of 0.45 psu (Fig. 3). However, the SSS anomaly may not only be due to the ocean freshwater gain but depends also on the ocean stratification. Due to heat loss (cooling) or turbulent energy gain (winds), the negative salinity anomaly “sinks” due to vertical entrainment and becomes thicker as it is diluted. For example, on 24 September during SP2, the surface layer of homogeneous low salinity (