Simulation of Circulation and Ice over the Newfoundland and Labrador Shelves: The Mean and Seasonal Cycle

Abstract

A three dimensional ice–ocean coupled model with a 7 km horizontal resolution has been developed to examine spatial and seasonal variability of hydrography and circulation over the Newfoundland and Labrador Shelves. Daily atmospheric forcing is applied and monthly open boundary forcing is prescribed based on a global ocean assimilation model. Monthly mean results averaged over a simulation period from 1979 to 2010 are evaluated using a variety of temperature, salinity, current, and ice observations. In comparison with observations and previous model results, the present model shows good skill in simulating the inshore and shelf-edge Labrador Current. The model temperature and salinity agree well with observations. Model sea-ice extent compares well with observations. The model mean transport is approximately 7.5 and 0.7 Sv (Sv = 10⁶m³s⁻¹) for the shelf-edge and inshore branches of the Labrador Current, respectively, consistent with observational estimates. The modelled total transport from the coast to the central Labrador Sea is 27.5 Sv from June to September, in good agreement with the observational estimate. The seasonal range for the shelf-edge and inshore branches is 2.0 and 0.6 Sv, respectively, strong in winter and fall and weak in spring and summer. The model mean freshwater transport at the Seal Island and Flemish Cap transects is 0.12 and 0.14 Sv, respectively, consistent with observational estimates, and the range of the seasonal freshwater transport is 0.09 Sv and 0.04 Sv for each transect, respectively, which is approximately in phase with the volume transport.

Résumé

[Traduit par la redaction] Un modèle couplé glace-océan tridimensionnel, d'une résolution horizontale de 7 km, a été développé pour examiner les variabilités spatiale et saisonnière de la circulation et des paramètres hydrographiques au-dessus des plateaux de Terre-Neuve et du Labrador. Nous avons appliqué un forçage atmosphérique quotidien ainsi qu'un forçage mensuel, aux limites du domaine ouvert, selon un modèle global d'assimilation océanique. Nous avons calculé la moyenne des résultats mensuels moyens sur une période de simulation allant de 1979 à 2010, et l'avons évaluée par rapport à diverses observations de température, de salinité, de courant et de glace. En comparaison avec des observations et des valeurs simulées existantes, le modèle présent montre de bons résultats quant à la simulation du courant du Labrador dans la zone côtière et en bordure du plateau. La température et la salinité issues du modèle correspondent bien aux observations. L’étendue de glace marine modélisée est aussi comparable aux observations. Le transport moyen modélisé est d'environ 7,5 et 0,7 Sv (Sv = 10⁶m³s⁻¹) pour les bras du courant du Labrador situés à la limite du plateau et dans la zone côtière, respectivement. Le transport total modélisé allant de la côte au centre de la mer du Labrador s’élève à 27,5 Sv, de juin à septembre. Ces valeurs correspondent bien aux estimations provenant d'observations. Les valeurs saisonnières pour les bras situés dans la zone côtière et en bordure du plateau vont de 2,0 à 0,6 Sv. Elles sont fortes en hiver et à l'automne, et faibles au printemps et en été. Le transport moyen d'eau douce modélisé pour les transects de l’île Seal et du bonnet Flamand s’élève à 0,12 Sv et à 0,14 Sv, respectivement. Il correspond donc aux estimations fondées sur les observations. L’étendue des valeurs de transport saisonnier de l'eau douce est respectivement de 0,09 Sv et de 0,04 Sv, pour chaque transect, ce qui concorde approximativement avec le transport du volume.

KEYWORDS:

• ice–ocean modelling
• Labrador Current
• freshwater transport
• mean and seasonal cycle
• Newfoundland and Labrador Shelves
• NEMO

1 Introduction

The Newfoundland and Labrador Shelves are both located in the northwest Atlantic Ocean. The coast of the Labrador Shelf is relatively straight, extending from Hudson Strait to the Strait of Belle Isle (Fig. 1). Connecting with the Labrador Shelf at the Strait of Belle Isle, the Newfoundland Shelf extends southward and then westward to Cabot Strait, with an irregular coastline and several large embayments. The topography of the Labrador Shelf is relatively deep with depths greater than 200 m over most of the area. Compared with the Labrador Shelf, the Newfoundland Shelf is generally shallow due to the presence of the Grand Banks of Newfoundland, which are 80 m deep on average. Labrador Shelf water is predominantly influenced by the Arctic outflow through the Canadian Arctic Archipelago, as well as the North Atlantic subpolar gyre and its western boundary current, the Labrador Current, which flows equatorward (Lazier & Wright, 1993). Water over the Newfoundland Shelf is affected by the Labrador Current and the poleward North Atlantic Current along the continental rise. Sea ice covers the Labrador Shelf and the northern Newfoundland Shelf in winter. Its advection, formation, and melting could substantially influence the ocean–atmosphere heat budget, ocean temperature, as well as salinity, ocean circulation and, therefore, marine ecosystems and fisheries. The presence of sea ice can also influence marine navigation, waves, and coastal erosion in this region.

Fig. 1 Map showing the Labrador and Newfoundland Shelves and adjacent northwest Atlantic Ocean as well as the model boundaries (thick solid lines). The isobaths displayed are 200, 1000, and 3000 m. SBI is Strait of Belle Isle, and GSL is the Gulf of St. Lawrence. HS is Hudson strait, and CS is the Cabot Strait. FP and FC are Flemish Pass and Flemish Cap, respectively. AC is the Avalon Channel. NAC stands for the North Atlantic Current and LC for the Labrador Current.

Efforts have been made to examine the spatial and seasonal circulation variability over the Labrador and Newfoundland Shelves (e.g., Greenberg & Petrie, 1988; Han, 2005; Han et al., 2008; Han, Ma, deYoung, Foreman, & Chen, 2011; Mertz, Narayanan, & Helbig, 1993; Petrie & Buckley, 1996; Tang, Gui, & Peterson, 1996; Urrego-Blanco & Sheng, 2012; Wu, Tang, & Hannah, 2012). Based on current meter observations in the fall of 1986, Petrie and Buckley (1996) estimated the volume transport and freshwater transport through the Flemish Pass and described the short-term variability in freshwater transport from fall to early winter. Mertz et al. (1993) estimated the mean freshwater transport along the Seal Island and Flemish Cap transects (Fig. 2) based on observations. In the late 1990s, Fisheries and Oceans Canada launched the Atlantic Zone Monitoring Program (AZMP; www.Meds-sdmm.dfo-mpo.gc.ca/isdm-gdsi/azmp-pmza/index-eng.html) to regularly collect physical, chemical, and biological data at fixed stations and transects. Han et al. (2008) developed a high-resolution model that resolves the Labrador Current well from the southern Labrador Shelf to the Newfoundland Shelf. These authors evaluated monthly model results using observations and estimated the seasonal cycle of transport for the Labrador Current. Han et al. (2011), using a finite volume coastal ocean model driven by the same climatological forcing, reproduced improved monthly ocean currents and examined seasonal and intraseasonal ocean dynamics in 1999. Urrego-Blanco and Sheng (2012) simulated seasonal and interannual variability over the eastern Canadian continental shelf by applying a spectral nudging method and a semi-prognostic method. Ikeda, Yao, and Yao (1996) developed a coupled ice–ocean model, showing overall agreement with observations for seasonal and interannual variations. Yao, Tang, and Peterson (2000) examined the seasonal cycle of ice formation and melting offshore of Labrador with an ice–ocean coupled model driven by monthly climatological atmospheric forcing. Zhang, Sheng, and Greatbatch (2004) also simulated the climatological seasonal cycle of sea ice using a one-third degree ice–ocean model, demonstrating the effect of ice heat capacity in delaying springtime sea-ice melt. These numerical studies have improved knowledge of the seasonal dynamics of circulation and ice, including the volume transport of the Labrador Current. However, the seasonal cycle of the freshwater transport of the Labrador Current has not been examined. Besides, the aforementioned ice–ocean models have insufficient spatial resolution to resolve the Labrador Current, with limited validation of model circulation fields.

Fig. 2 Sites of moored measurements from 1979 to 2010 used in the model evaluation. The 200, 1000, and 3000 m isobaths are also shown. The two blue plus signs (sites A and B) indicate the locations where the model and observational current profiles are compared. The solid blue squares are monthly mean current observations at different depths. The open red circles and the red crosses are observational temperature sites at different depths over the Newfoundland Shelf and the Labrador Shelf, respectively. The solid red square indicates the location of Station 27 where temperature and salinity were observed. ST, BT, and FT are the Seal Island, Bonavista, and Flemish Cap Transects, respectively. The thick black line is the model boundary.

In this study, a high-resolution three-dimensional prognostic model, based on the Nucleus for European Modelling of the Ocean (NEMO; Madec, Delecluse, Imbard, & Levy, 1998), is developed over the Newfoundland and Labrador Shelves, to provide a new modelling capacity for this region. The main objectives are to validate the model's ability in reproducing temperature, salinity, currents, and ice and to investigate seasonal and spatial variability of the volume and freshwater transport associated with the Labrador Current. In Section 2, the model and its configuration, open boundary conditions, atmospheric forcing data, initial conditions, and solution procedure are described. In Section 3 the ability of the model is evaluated based on temperature and salinity data, current data and sea-ice data. Subsurface circulation over the entire domain is examined, with a focus on the Labrador Current. In Section 4 the seasonal cycle of volume and freshwater transport across the Seal Island and Flemish Cap transects is investigated. A brief summary and discussion is provided in Section 5.

2 Ice-ocean coupled model and forcing

a Model Configuration

The numerical ocean model used in the present study is based on the NEMO, version 2.3, modelling system (Madec et al., 1998), using version 9 of the Océan Parallélisé system (NEMO-OPA9) and version 2 of the Louvain-la-Neuve Ice Model (NEMO-LIM2); NEMO-OPA9 is a primitive equation, finite difference, ocean circulation model with a free sea surface and a z-coordinate system in the vertical direction. The vertical mixing scheme is parameterized from the turbulent closure scheme of Gaspar, Gregoris, and Lefevre (1990). The model external and internal time steps in this study are 6 and 90 s, respectively.

The NEMO-LIM2 system is a two-level thermodynamic-dynamic sea-ice model (Bouillon, Morales Maqueda, Legat, & Fichefet, 2009; Fichefet & Morales Maqueda, 1997). The ice model is coupled with the OPA ocean circulation model through the exchange of momentum, heat, water, and salt. Ice and ocean exchange momentum through an ice–ocean stress in a quadratic bulk formula. The oceanic heat flux from ocean to ice is a function of ocean temperature and turbulent mixing.

The model domain covers the Newfoundland Shelf including the Grand Banks, the Labrador Shelf, and the adjacent deep northwest Atlantic Ocean (Fig. 1). The ocean circulation model has a rotated rectilinear grid with a resolution of 6–8 km. The vertical resolution ranges from 6 m at the surface to 250 m at the bottom with a total of 46 levels. As a result, the surface layer point is located approximately 3 m below the actual water surface. The model uses topography from the Canadian Hydrographic Service with 7 km resolution covering the shelf regions and topography from the Earth Topography 5-minute gridded elevation data (ETOPO5; http://www.ngdc.noaa.gov/mgg/global/etopo5.HTML) for the remainder of the ocean area.

b Atmospheric Forcing and Open Boundary Treatment

The atmospheric forcing is from the North American Regional Reanalysis (NARR) dataset (Mesinger et al., 2006; www.esrl.noaa.gov/psd). The NARR project is an extension of the National Centers for Environmental Prediction (NCEP) Global Reanalysis that is run over North America. This model uses the NCEP Eta Model with a horizontal resolution of 32 km and 45 vertical layers together with the Regional Data Assimilation System (RDAS). This assimilation method significantly improves the accuracy of historical air temperature, winds, and precipitation simulation. The model ingests daily downward shortwave and longwave radiation, wind speed, air temperature, specific humidity, and precipitation. A constant albedo value of 0.066 is used to approximate the average cloudy conditions over the study region based on the suggestion of Pegau and Paulson (2001).

Model open boundary conditions are from the Simple Ocean Data Assimilation (SODA) global model (Carton & Giese, 2008) and from Hu and Myers (2014). Sea levels and three-dimensional currents linearly interpolated from the SODA monthly model output were prescribed along the open boundary points at each time step. A relaxation sea-ice open boundary condition was applied along the northern open boundary, with sea-ice concentration and thickness from Hu and Myers (2014). Their domain covers the Arctic and part of the Atlantic Ocean, including the northern open boundary and part of the eastern open boundary of the present model. Their modelling period is from 1970 to 2100, and they validated model ice concentration and thickness using observations in the Canadian Arctic Archipelago.

c Solution Procedure and Statistics Used for Model Evaluation

The model was initialized from rest and run for two years with the forcing for 1979 repeated in the second year. The model was then run from 1980 to 2010.

To evaluate the model solutions quantitatively, the model solutions were compared with various measurements made between 1979 and 2010 (Fig. 2). In addition to the root-mean-square (RMS) difference, the velocity difference ratio (VDR), defined as the ratio of the sum of the squared magnitudes of the vector velocity differences to the sum of the squared magnitudes of the observed velocities, was examined:(1) where is the horizontal model velocity and is the horizontal observational velocity. Lower values indicate better agreement, with being exact agreement and being fair comparison.

Another index is the speed difference ratio (SDR) defined as the ratio of the sum of the squared speed difference to the sum of the squared magnitudes of the observed velocities, that is,(2)

The third index is , which is(3) where is the variance, and is a variable and can be sea level or a current velocity component; and represent observation and model, respectively. The fourth index is the correlation coefficient () between the model and observations.

3 Model validation

a Model Circulation and Ice Concentration

The model monthly mean (January and July) circulation at 20 m depth (Fig. 3) over the Labrador and Newfoundland Shelves averaged over 1979 to 2010 is dominated by the equatorward flowing Labrador Current, similar overall to the climatological monthly mean circulation patterns in previous model results (Han et al., 2008, 2011; Tang et al., 1996; Wu et al., 2012). The combination of the West Greenland Current, the Baffin Island Current, and the outflow from Hudson Strait forms two branches of the Labrador Current: an inshore branch near the coast and an offshore branch along the shelf edge and continental slope. At around 54°N, there is a clear onshore flow along the 200 m isobath due to topographic steering, strong in winter but weak in summer. At around 49°N, part of the inshore branch flows through the Avalon Channel heading to the southwestern Newfoundland Shelf while the remainder turns eastward to merge with the shelf-edge branch. The shelf-edge branch separates into two parts at the northeast corner of the Grand Banks, one following the shelf break through the Flemish Pass and the other flowing around the Flemish Cap. Between the Flemish Pass and the tail of the Grand Banks, the shelf-edge Labrador Current retroflects to join the North Atlantic Current.

Fig. 3 32-year average model currents at 20 m below the surface in January and July, representing winter and summer. The 200, 1000, and 3000 m isobaths are also depicted in red, green, and black, respectively.

There is substantial seasonal change in the strength of the inshore and offshore Labrador Current, strong in fall and winter (as represented by January) and weak in spring and summer (as represented by July) (Fig. 3, for details see Section 4a). Model circulation at 20 m above bottom (Fig. 4) shows that the near-bottom Labrador Current is substantial, suggesting the importance of the barotropic component in the flow. The near-bottom current is also strong in fall and winter and weak in spring and summer.

Fig. 4 Sub-sampled model currents at 20 m above bottom in January and July. The 200, 1000, and 3000 m isobaths are also depicted in red, green and black, respectively.

The model ice concentration and areal extent in December and March are shown in Fig. 5. In December, sea ice occurs over the inner Labrador Shelf. By March, sea ice covers the Labrador Shelf and the northeastern Newfoundland Shelf, reaching the northern Grand Bank. The offshore extent is approximately along the shelf edge.

Fig. 5 Model sea-ice concentration of 0.1 south of 55°N in December and March, for the year with the median modelled sea-ice extent for 1979 to 2010. The observed areal sea-ice boundary with a sea-ice concentration of 0.1 (thick grey line) is also shown, for the year with the median observed sea-ice extent over the same period. The 1000 m isobath (solid line) is depicted.

b Temperature and Salinity at Station 27

Station 27 (see Fig. 2) is located off St. John's at 47.55°N and 52.59°W, with a water depth of 176 m. Hydrographic data have been collected at this station since 1946. Since the late 1990s, temperature and salinity data at Station 27 have regularly been collected by the Northwest Atlantic Fisheries Centre through the AZMP. Both model and observed temperature and salinity at Station 27 from 1979 to 2010 are averaged into monthly means (Fig. 6). The surface temperature has a seasonal range of 15°C, from -1°C in winter to 14°C in summer (Fig. 6a). The RMS difference between observations and model values is 1.0°C with a variance of 7°C in the observations. The correlation coefficient is 0.99, significant at the 95% confidence level. The seasonal range of bottom temperature is small. The RMS difference and correlation coefficient of the near-bottom temperature are 0.8°C and 0.74, respectively (Fig. 6c). The model tends to overestimate the bottom temperature by 1.0°C in winter.

Fig. 6 Monthly sea surface (a) temperature and (b) salinity as well as bottom (c) temperature and (d) salinity at Station 27 from the model and observations.

The observed surface salinity has a seasonal range of 1.2 (Fig. 6b), while the model range is 0.8. The RMS difference in the surface salinity between the model and observations is 0.7, with a correlation coefficient of 0.83, significant at the 95% confidence level. The mean difference (model minus observation) is 0.3. The bottom salinity does not change significantly with the seasons (Fig. 6d), with a mean difference between the model and observations of 0.3.

c Temperature at Mooring Sites

Temperature data from moored instruments were extracted from a database maintained at the Bedford Institute of Oceanography (www.bio.gc.ca/science/data-donnees/base/index-en.php). The mooring sites are located over the Newfoundland and Labrador Shelves, inshore of the 1000 m isobath (Fig. 2, open circles and crosses). Each site has observations ranging from 2 to 900 m and most of the data are within the upper 300 m. It should be noted that observed temperatures over the Labrador Shelf are only available in summer and fall primarily because of sea-ice cover in winter. The model results are interpolated vertically onto the observational depth and time. Then both observational and model results are divided into two groups: one for the Newfoundland Shelf and the other for the Labrador Shelf. Monthly mean temperatures from 1979–2010 for the upper (0–40 m) and bottom layers (below 40 m) are calculated based on 1183 observations over the Newfoundland Shelf and 838 observations over the Labrador Shelf. The observed upper-layer temperature ranges from 0°C to 9°C during the year over the Newfoundland Shelf, but only reaches 6.5°C in summer over the Labrador Shelf (Fig. 7a and Fig. 7c). Seasonal variations in model temperatures are similar to those in the observations but are low in summer. Statistically, the RMS difference between the modelled and observed upper layer temperatures is 1.6°C for both the Newfoundland and Labrador Shelves and the correlation coefficients are 0.92 and 0.61, respectively (both significant at the 95% confidence level), indicating the ability of the model to reasonably reproduce seasonal variability in temperature data. The temperature in the bottom layer is steady, and there are no significant seasonal changes over the Labrador and Newfoundland Shelves (Fig. 7b and Fig. 7d). The RMS differences between the modelled and observed bottom temperatures are 0.8°C and 0.6°C for the Newfoundland and Labrador Shelves, respectively.

Fig. 7 Monthly observed and simulated temperatures at mooring stations over the Newfoundland Shelf for (a) 0–40 m, (b) below 40 m, and the Labrador Shelf for (c) 0–40 m and (d) below 40 m. Moorings locations are indicated in Fig. 2.

d Temperature and Salinity Comparison at the Flemish Cap and Bonavista Transects

The AZMP program collects temperature and salinity data along the Flemish Cap transect at 47°N and the Bonavista transect (Fig. 2) in spring (April and May), summer (July), and fall (November). The data were collected from 1993 to 2003. The model output for the dates closest to the observational dates is extracted. The observational and model results are averaged for the 11-year period by month. Figure 8 shows the temperature and salinity for the upper and bottom layers along the Flemish Cap transect. The upper and bottom layer mean temperature is calculated, using the same method as in Section 3c. The upper layer temperature has significant seasonal variability, low in spring and high in summer and fall. The mean seasonal difference in the temperature of the upper layer is 0.1°C. The temperature in the bottom layer is relatively steady and reaches its highest value of about 3.2°C in fall because of the accumulated heat flux taken during summer. The mean seasonal difference in the temperature of the bottom is 0.4°C. Both upper layer and bottom layer salinities are lower from late spring to summer and higher in fall, corresponding to the sea-ice formation and melting cycle. A mean salinity bias of 0.3 exists between the modelled and observed salinities at the surface and bottom layers.

Fig. 8 Surface layer (solid line with squares) and bottom layer (dashed line with circles) for (a) temperature and (b) salinity for the Flemish Cap transect.

Figure 9 shows the time series of temperature and salinity in the upper and bottom layers along the Bonavista transect. In the upper layer, the temperatures in summer and fall are colder than those at the Flemish Cap transect by at least 2.6°C. The mean difference between the modelled and observed temperature is 0.3°C, with the observed value being lower. In the bottom layer, the model temperature has a seasonal variation consistent with observations but with a mean difference of 1.0°C. Both model values and observations indicate that the surface temperature is colder than the bottom temperature in April. Model surface and bottom layer salinities at the Bonavista transect are higher than observations with mean differences of 0.5 and 0.4, respectively.

Fig. 9 As in Fig. 8, but for the Bonavista transect.

e Validation with Current Meter and Acoustic Doppler Current Profiler (ADCP) Currents

The moored current meter data are also extracted from a database at the Bedford Institute of Oceanography (www.bio.gc.ca/science/data-donnees/base/index-en.php). Observations were collected at several sites on the Flemish Pass area (Fig. 2, blue squares) between 1980 and 1988. About 90% of the data are from 1980, 1983, 1985, and 1986. Typically there are two to five monthly means at each mooring. The mooring depths range from 20 to 950 m, with a median value of 119 m. The analysis of observed data indicates dominant meridional flows of 0.16 m s⁻¹ southward and weak zonal flows of 0.03 m s⁻¹ without a dominant direction, on average. Model results are interpolated to the observational depth and time. Both model results and observations are averaged over the depth and over the 9-year period by month (Fig. 10). The model meridional and zonal currents are consistent with observations. The meridional current has significant seasonal variation: strong in winter and fall but weak in summer. The observed variance is 0.15 m s⁻¹. Statistical evaluations of the model results are summarized in Table 1. The RMS difference is 0.03, 0.03, and 0.04 m s⁻¹ for the zonal component, meridional component, and total speed, respectively. The SDR and VDR are 0.07 and 0.09, respectively, which indicates very good agreement between the model and observations. The also suggests very good agreement for the meridional component and total speed, indicating that the model captures the strong shelf-edge Labrador Current well. The for the zonal component is higher than that for the meridional component. All three primary indices suggest the best agreement in the meridional velocity and total speed.

Fig. 10 Monthly mean depth-averaged currents in the Flemish Pass. Locations are indicated by solid blue squares in Fig. 2.

Table 1. Statistics comparing model currents and monthly mean observations in the Flemish Pass. Locations are shown in Fig. 2. U is the zonal velocity and V is the meridional velocity (m s⁻¹).

	RMS (Observed-Model)	RMS (Observed)	Correlation
Speed	0.04	0.15	0.75	0.17
U	0.03	0.01	0.69	>1
V	0.03	0.15	0.78	0.05
SDR	0.07	—	—	—
VDR	0.09	—	—	—

Current meters were deployed at the edge of the Labrador Shelf (locations marked as A and B in Fig. 2), with observations from December 1991 to June 1992 at site A and from July to October 1991 at site B. A comparison of model monthly mean current profiles with moored measurements in the same month and year is shown in Fig. 11. The model results show overall good agreement with observations for the Labrador Current along the shelf edge. The vertical gradient of the horizontal current is captured well in the simulated results at both sites. Nevertheless, the model overestimates the barotropic component at site A. The comparison indicates that the summer and fall vertical density structure is well represented (Fig. 11, site B). Quantitative statistics are also calculated (Table 2), showing a reasonable comparison in the zonal and meridional velocity components. The RMS difference is 0.08 and 0.09 m s⁻¹ for the zonal and meridional components, respectively, substantially smaller than the respective observed variances of 0.13 and 0.19 m s⁻¹. The SDR and VDR are 0.27 and 0.28, respectively, indicating that the model captures the observed currents well.

Fig. 11 Comparison between selected model monthly mean current profiles and moored measurements. U and V are the zonal and meridional components, respectively. Red squares are observations and the blue lines are model results. See Fig. 2 for the locations of sites A and B.

Table 2. Statistics comparing vertical profiles of monthly mean simulated currents and the monthly mean observations at sites A and B. Locations are shown in Fig. 2. U is the zonal velocity and V is the meridional velocity (m s⁻¹).

	RMS (Observed-Model)	RMS (Observed)	Correlation
U	0.08	0.13	0.24	0.36
V	0.09	0.19	0.33	0.24
SDR	0.27	—	—	—
VDR	0.28	—	—	—

Gridded current fields for the Flemish Cap transect in April (spring), July (summer), and November (fall) were generated by Han et al. (2008) using vessel-mounted Acoustic Doppler Current Profiler (ADCP) data from 1992 to 2004. The model currents in April, July, and November are averaged by month to produce monthly mean currents over this period, which are then interpolated to the centre of the ADCP grid cells. There is good qualitative agreement between the present model and the ADCP measurements in terms of the spatial distribution pattern and the current strength for the dominant southward current (Fig. 12). Both the model and observations show that the southward inshore and shelf-edge branches of the Labrador Current is strong in fall and weak in summer. There are also indications of recirculation and significant near-bottom flow which is weak in the model results. The RMS difference in speed is 0.05, 0.05, and 0.07 m s⁻¹ for spring, summer, and fall, respectively. The respective velocity difference ratio, accounting for both the eastward and northward components, is 0.15, 0.30, and 0.29 for spring, summer, and fall, respectively, suggesting good agreement. The model agreement with observations in this study is better than in some previous modelling effects (Han et al., 2008, 2011), as shown in Table 3. Note that unlike the climatological monthly mean currents in Han et al. (2008, 2011), the present monthly mean currents are calculated over the ADCP observational period.

Fig. 12 Comparison between the model currents with vessel-mounted ADCP data along the Flemish Cap transect in spring, summer, and fall. Only the normal component (m s⁻¹; positive southward) is shown.

Table 3. Statistics comparing Flemish Cap transect currents and the observed values in spring, summer, and fall. Transect locations are shown in Fig. 2. Units are m s⁻¹.

	VDR		RMS difference of speed
Month	Han et al. (2011)	This study	Han et al. (2008)	Han et al. (2011)	This study
Spring	0.16	0.15	0.08	0.07	0.05
Summer	0.34	0.31	0.09	0.08	0.05
Fall	0.22	0.30	0.09	0.08	0.07

f Sea-Ice Concentration and Extent

Monthly mean model sea-ice extent from 1979 to 2010 is compared with observations derived from the digital archive regional dataset prepared by the Canadian Ice Service (http://iceweb1.cis.ec.gc.ca/30Atlas10/?lang=en). The observed and modelled sea-ice extent is defined as the total ice-covered area south of 55°N with an ice concentration above 0.1 in each model cell. Figure 13 shows the seasonal evolution of the sea-ice extent from November to May. Both the model and the observations show maximum ice extent in February and March. The model maximum ice extent in February and March is in good agreement with observations. The present model agrees better with the observed ice extent in April than Yao et al.’s (2000) but underestimates ice extent in May. Ikeda et al. (1996) speculated that this underestimation during the spring retreat period may be the result of using a two-category ice model. With the inclusion of multiple ice categories, Yao et al. (2000) showed improved agreement with observations during the spring retreat period. However, Hibler and Walsh (1982) and Walsh, Hibler, and Ross (1985) obtained comparable results between seven-category and two-category ice models during the spring retreat time. The results from the two-category ice model in the present study also show that the simulated ice coverage is in good qualitative and approximate quantitative agreement with observations.

Fig. 13 Monthly sea-ice extent south of 55°N.

4 Seasonal cycle of volume and freshwater transports

a Volume Transport

To estimate the seasonal cycle of the Labrador Current transport, the Seal Island transect is divided into two segments with the inshore part from the coast to 250 m offshore and the shelf-edge part from 200–1700 m (Fig. 14), as in Han et al. (2008). The simulated inshore annual mean transport is about 0.86 Sv (Sv = 10⁶m³s⁻¹), consistent with Lazier and Wright's (1993) observational estimate of 0.8 Sv. The model inshore branch transport has a seasonal change of 0.5 Sv, high in winter and low in summer (Fig. 14a). The transport associated with the shelf-edge and upper slope currents has a strong seasonal cycle, with weak transport in late summer. The mean transport is about 7.5 Sv, consistent with Han et al.’s (2008) model estimate of 7.5 Sv but larger than Lazier and Wright's (1993) estimate of 6 Sv based on current meter observations and geostrophic methods. The model mean volume transport from the coast to 3000 m is estimated to be 23.9 Sv. Dengler, Fischer, Schott, and Zantopp (2006), based on the lowered ADCP measurements made during the summers of 1996 to 2005, estimated a total volume transport of 25.7 Sv at the Seal Island transect from the upper slope to the central Labrador Sea. The present model transport from the coast to the central Labrador Sea is calculated to be 27.5 Sv, agreeing well with Dengler et al.’s (2006) observational estimate. Volume transport at the Flemish Cap transect is also examined (Fig. 15), through the Avalon Channel (from the coast to the 100 m isobath on the offshore side of the Avalon Channel, about 100 km from the coast) and through the Flemish Pass (from the 150 m isobath on the Grand Banks side to the 1000 m isobath on the Flemish Cap side). The inshore Labrador Current through the Avalon Channel flows southward. This flow has an annual mean transport of 0.7 Sv and reaches its minimum in summer. The shelf-edge Labrador Current through the Flemish Pass is southward, with a mean transport of 7.2 Sv, larger than both the observational estimate of 5.8 Sv and the model estimate of 5.5 Sv (Han et al., 2008). The mean volume transport from the coast to the 3000 m isobath is calculated to be 20.6 Sv at the Flemish Cap Transect. Comparing the model mean transports from the Seal Island transect to the Flemish Cap transect, it is found that the shelf-edge Labrador Current follows tightly along the isobath before reaching the Flemish Pass. In contrast, nearly 90% of the shelf-edge Labrador Current turns offshore from the Flemish Pass to the tail of the Grand Banks.

Fig. 14 Monthly volume transport (a) across the inshore portion, (b) across the upper slope, and (c) from the coast to the 3000 m isobath across the Seal Island transect. The annual mean transport is also depicted (black dashed line).

Fig. 15 Monthly volume transport (a) across the inshore portion, (b) through the Flemish Pass, and (c) from the coast to the 3000 m isobath across the Flemish Cap transect. The annual mean transport is also depicted (black dashed line).

b Freshwater Transport

The freshwater transport associated with the Labrador Current has been calculated based on hydrographic observations (Mertz et al., 1993; Petrie & Buckley, 1996). The transport of freshwater has a profound effect on vertical stratification and consequently on some physical and biological processes. In all the previous studies on freshwater fluctuations, a reference salinity, , based on the background oceanic salinity, was applied in order to determine the freshwater transport across a transect. The transport can be estimated as follows:where S and v are the model salinity and current normal to the transect, respectively; z is the depth, and x is the along-transect distance; was assumed to be 34.8 for consistency with Mertz et al.’s (1993) calculation.

The freshwater transport across the Seal Island and Flemish Cap transects is estimated for the same segments as the volume transport estimation in Section 4a. Figures 16a and 16b show the monthly freshwater transport for the inshore and offshore branches of the Labrador Current at the Seal Island transect. Both branches have seasonal cycles, with weak transport in July for the inshore branch and in August–September for the offshore branch. The mean freshwater transport is 0.04 Sv for the inshore branch and 0.06 Sv for the offshore branch. The total mean freshwater transport across the Seal Island transect is estimated to be 0.12 Sv (Fig. 16c), which is consistent with the total freshwater transport of 0.13 Sv estimated by Mertz et al. (1993) for the Hamilton Bank line (see ST in Fig. 2) based on salinity data from July 1980. The total freshwater transport is high in February and low in August, with a seasonal range of 0.09 Sv.

Fig. 16 Monthly freshwater transport (a) across the inshore portion, (b) across the upper slope, and (c) from the coast to the 3000 m isobath across the Seal Island transect. The annual mean freshwater transport is also depicted (dashed black line).

Figures 17a and 17b show the monthly freshwater transport for the inshore and offshore branches of the Labrador Current at the Flemish Cap transect. The mean freshwater transports for the inshore and offshore branches are 0.04 and 0.08 Sv, respectively. The total mean freshwater transport along the Flemish Cap transect is 0.14 Sv (Fig. 17c), consistent with Mertz et al.’s (1993) estimate of Sv based on the averaged salinity from September and March. These authors also estimated the freshwater transport along the offshore part to be Sv. The freshwater transport has a seasonal cycle of 0.03 Sv associated with the inshore branch, largest in January and December and smallest in August (Fig. 17a). The shelf-edge branch has a seasonal range of 0.03 Sv, largest in June and smallest in October (Fig. 17b). The total freshwater transport is large in January and December and small in late summer, with a seasonal range of 0.04 Sv (Fig. 17c). Although there are differences between the simulated freshwater transport and the observation-based estimates for the offshore portion at both transects, it should be noted that the observational estimate is based on spring and summer data. Another possible reason for the difference is the model overestimation of salinity. A 0.3 overestimation in the simulated salinity could lead to a maximum 0.035 Sv underestimation of the total freshwater transport. After adding this, the total mean freshwater transport is 0.155 and 0.175 Sv for the Seal Island and Flemish Cap transects, respectively.

Fig. 17 Monthly freshwater transport (a) across the inshore portion, (b) through the Flemish Pass, and (c) from the coast to the 3000 m isobath at the Flemish Cap transect (47°N). The annual mean freshwater transport is also depicted (dashed black line).

Freshwater transport depends on both volume transport and salinity. The seasonal cycle of the inshore freshwater transport at both transects corresponds well to that of the inshore volume transport (Fig. 16a and Fig. 17a). Strong inshore volume transport leads to much more significant inshore freshwater transport. The offshore freshwater transport at the Seal Island transect is intensified in late winter as a result of the strong winter volume transport. In contrast, the seasonal cycle of freshwater transport through the Flemish Pass (Fig. 17c) depends mainly on that of salinity. Overall, the seasonal cycle of the freshwater transport at both transects is dominated by that of the volume transport.

5 Discussion and summary

A three-dimensional, prognostic ice–ocean model has been developed for the Newfoundland and Labrador Shelves. The prognostic model is based on NEMO-OPA9 and NEMO-LIM2, and is forced by daily winds, heat flux, and SODA monthly sea level and inflows at the four open boundaries. The ice open boundary condition is from Hu and Myers (2014). The simulated results have been evaluated using temperature, salinity, current, and sea-ice data.

The present model is able to reproduce the seasonal change in ocean temperature and salinity not only within the mixed layer but also near the sea bottom. At Station 27, the surface and bottom temperatures and salinity are well reproduced. However, a mean difference of 0.3 exists for salinity as a result of an overestimation of saltier inflow at the northern open boundary. The model captures well the observed seasonal variability and spatial difference in temperature and salinity at the Flemish Cap and Bonavista transects. Detailed comparison of the prognostic model circulation with vessel-mounted ADCP data indicates good agreement. Monthly model current profiles are also validated with selected current meter data during the 1980–1988 and 1991–1992 periods. The prognostic model results indicate significant seasonal and spatial variations in the regional circulation. The shelf region is dominated by the equatorward Labrador Current along the shelf edge and along the Labrador and Newfoundland coasts. Annual mean transports for the inshore and shelf-edge Labrador Current are consistent with previous model results and observations. The simulated ice extent agrees well with observations. The examination of volume and freshwater transport suggests significant seasonal difference over the Labrador and Newfoundland Shelves. The shelf-edge Labrador Current has a mean transport of 7 Sv with a seasonal range of 2 Sv. The inshore Labrador Current has a mean transport of 0.7 Sv with a seasonal range of 0.6 Sv. The total mean freshwater transports at the Seal Island and Flemish Cap transects are estimated to be 0.12 Sv and 0.14 Sv, respectively, consistent with observational estimates. The seasonal cycle of the freshwater transport is dominated by the volume transport over the Labrador and Newfoundland Shelves.

The present ice–ocean coupled model shows overall good skill in reproducing seasonal hydrography, circulation, and ice variability. However, further improvements could be made in several aspects. For example, the model tends to underestimate the mixed-layer depth in summer, especially over shallow waters. Improved turbulence closure could lead to a more accurate simulation of the vertical temperature and salinity structure over shallow water. The irregular coastline is not well resolved in the current study, resulting in uncertainty in sea-ice extent.

Acknowledgements

We thank Paul Myers for providing sea-ice conditions on the northern boundary of our model and Ingrid Peterson for providing the sea-ice extent data. The NARR reanalysis data are provided by the NOAA/OAR/ESRL PSD, Boulder, Colorado, USA, from their website at http://www.esrl.noaa.gov/psd/. We thank the two anonymous reviewers for their constructive and thorough reviews.

Disclosure statement

No potential conflict of interest was reported by the authors.

Additional information

Funding

This work was funded by the Aquatic Climate Change Adaptation Services Program (ACCASP), Fisheries and Oceans Canada.