Publication - Strategy/plan

Scottish Shelf Model. Part 5: Wider Loch Linnhe System Sub-Domain

Published: 11 Mar 2016
Part of:
Marine and fisheries

Part 5 of the hydrodynamic model developed for Scottish waters.

Scottish Shelf Model. Part 5: Wider Loch Linnhe System Sub-Domain
5 Summary and Conclusions

5 Summary and Conclusions

5.1 Introduction

This report documents the work carried out in developing the Wider Loch Linnhe System ( WLLS) model. This work includes: data collated for the numerical modelling, setup, calibration and validation of the flow model, and longer term (six month) simulations for 2011 and 1991 and the year-long climatology simulation required for this study.

The FVCOM model was chosen because of its capabilities as well as it being freely available, which then fulfils the aim for this and other models developed under the same project to become community models.

5.2 Hydrodynamic model calibration and validation

The WLLS hydrodynamic model was setup using bathymetry taken from a number of sources, from the freely available but coarser EMODnet/ NOOS data, to the UKHO and Marine Scotland higher resolution datasets. Where data from these sources was not readily available, Admiralty Charts were digitised (with permission from the Hydrographic office) to fill in any gaps. All bathymetry was reduced to mean sea level as the common datum.

The model mesh was created with the SMS mesh generator using a spherical coordinate system (latitude and longitude). The model was run with 10 vertical sigma layers of variable thickness with a vertical datum of Mean Sea Level ( MSL). The variable depths had two layers at the surface which are 1m thick, and two layers at the bottom each being 2.5m thick. The remaining 6 layers were equally spaced.

An analysis of the data available for forcing the hydrodynamic ( HD) model showed that periods in 2011 were the most appropriate as calibration and validation periods, as all of the necessary forcing data required by the model are available. Datasets for calibration and validation of the model in the form of timeseries of water levels and current speeds were also available within Loch Linnhe and Loch Sunart. Additionally temperature and salinity profiles were available for comparisons with the model.

Boundary conditions for water levels, depth-averaged currents, temperature and salinity were taken from the Atlantic Margin Model ( AMM) developed by NOC-L. These were applied using a nested boundary approach. Water levels and currents were provided at hourly intervals, whereas the temperature and salinity were provided at daily intervals for each of the 40 layers in the AMM. Much of the meteorological forcing was provided by NOC-L and derived from the Met Office model ( i.e. wind components, air temperature, air pressure and precipitation. The remaining parameters came from the ECMWF ( i.e. radiation, evaporation, relative humidity). The heating input was calculated internally by FVCOM rather than provided externally. This was found to provide the best results for sea surface temperature. River flow data was provided by CEH from their Grid to Grid model. At three of the large lochs where river flow was specified at the mouth in the CEH dataset, the flows were redistributed to all the rivers feeding the Lochs. River salinity was set at 0 psu, and temperature at 7 degrees Celsius, which was felt appropriate when considering the observed nearshore water temperatures. This was later increased to 10 degrees Celsius after consideration of the climatology results and based on limited published data on annual mean temperature in Scottish rivers. Therefore all 2011 simulations have been run with a river temperature of 10 degrees Celsius and the climatology simulations have been run with a river temperature of 7 degrees Celsius.

Comparisons between the model results and measurements of water level and current speeds showed reasonable agreement, the location of measured data within eddies made it difficult to exactly replicate the timing and direction of peak flows. Comparisons of the 10 layer baroclinic model showed that salinity comparisons with data were generally within 1 psu and the temperature comparisons were within 0.5 Celsius in line with our target.

5.3 2011 simulations

A six month simulation of May to October 2011 was required. The inputs used in this model run are from the same sources as the calibration and validation runs. As with those runs the comparison with surface currents, temperature and salinity at the data buoy were reasonable considering the location of the data buoy within an eddy. Temperatures were overestimated during the summer months but fell back in line with measured temperatures for October. This implies that the model heats and cools at a higher rate than seen in measured data. The salinity CTD comparisons show an increase in salinity during May (1 psu) and October (2 psu) compared to the calibration and validation runs. This illustrates the importance of the initial conditions on the results. Alternatively, it may also suggest that the input of fresh water during this period is underestimated by the G2G model data.

5.4 1991 simulation

A six month simulation of May to October 1991 was also required. The data sources for 1991 vary from those used in 2011. The water level, current, temperature and salinity data came from the POLCOMS 12km model provided by NOC-L. No met forcing was available from the AMM model, so all parameters apart from wind were taken from the ECMWF interim datasets. Wind data was available from the met office model at hourly intervals on a 50km grid. River flow data from Marine Scotland for 3 rivers in Upper Loch Linnhe and diffuse inputs for the other catchments were used.

Overall the 1991 model runs replicate the water levels, currents and salinity in the WLLS well. The current comparisons in particular were an improvement on the data buoy comparisons for 2011. This supports the suggestion that the location of the data buoy has an influence on the current comparisons. The temperature comparisons were not as good, with the model heating up and cooling down at a higher rate than seen in the measured data. The stratification of the water column was picked up well by the model even though in some cases the absolute values were 1-2 degrees Celsius and 1-2 psu.

5.5 Climatology simulations

Another requirement of this study was to produce a one year climatic run based upon climatological forcing to represent a typical annual cycle. The model was therefore run for the period January to December. Mean boundary forcing for water levels (mean yearly tides), currents, temperature and salinity were taken from the Scottish Waters Shelf Model climatology results. An efficient method was developed to interpolate the forcing data onto the nested boundary nodes and elements. River climatology was also provided by CEH and used for this study following analysis by NOC-L, river salinity was set at 0psu and temperature at 7 degrees Celsius. Meteorological forcing was derived by NOC-L from ECMWF ( ERA-Interim) averaged data to provide monthly mean wind-stress, pressures, heating and evaporation minus precipitation from the period 1981-2010.

Average monthly temperature and salinity simulated by the model were compared against sea surface temperature and salinity climatological datasets and residual currents for the months of February and August; the results showed lower surface temperature in February but slightly higher temperatures in August. The salinity close to land was slightly lower than the comparison data in both months.

Mean spring and neap tidal ranges and currents were also calculated using M2 and S2 water level and current constituents and then compared against ABPmer model of the area. Comparisons are generally good, with the main difference found between the Outer Hebrides and Tiree, and at the mouth of Loch Linnhe. The ABPmer model does not resolve the channel between Skye and Mull and the mainland and the full extent of Loch Linnhe. The WLLS model does however resolve this channel and therefore this is likely to be the reason for the differences observed and the benefit of the finer resolution WLLS model.