2 Available data for model development
In order to carry out the numerical modelling works for the Wider Loch Linnhe System ( WLLS), the following data have been collated:
- Bathymetry data, required for creating the bathymetry for the numerical model.
- Forcing data, required for specifying the forcing conditions in the numerical flow models.
- Calibration and validation data, required for calibrating and validating the numerical models.
This section of the report describes the data collated for the WLLS model area. Where appropriate, reference is made to the overall project data review report (Halcrow, 2012). Note that the proposed model domains shown in this section are not the final model domains but an approximation.
2.2 Bathymetric Data
2.2.1 Coastline Data
Two coastline data sets have been obtained for use in this study. These are: the Global Self-consistent, Hierarchical, High-resolution Shoreline ( GSHHS) distributed by National Geophysical Data Centre ( NGDC) in the US, and Ordnance Survey Mapping.
The GSHHS coastline comes in different resolutions. For the UK, the best resolution available is the World Vector Shoreline ( WVS) designed to be used at a resolution of 1:250,000. The GSHHS coastlines have been data processed to ensure they are free of internal inconsistencies such as erratic points and crossing segments.
The Ordnance Survey ( OS) Vector Map District contains tidal boundary polylines, which are at Mean High Water Spring level ( MHWS) in Scotland and MHW in England and Wales. The GSHHS data is considered appropriate for use in areas where the model resolution is coarse, the OS vector map district MHWS line should be used in areas of higher resolution.
2.2.2 Global/Regional Gridded Data Sets
Three existing coarse resolution bathymetry data sets have been identified which cover the study area the GEBCO_08, the ETOPO-1 grid and the EMODnet grid. These are described briefly below. Details regarding these datasets are provided in Halcrow (2012).
2.2.3 General Bathymetric Chart of the Oceans ( GEBCO)
The GEBCO_08 data set is a global DTM at 0.5 minute resolution generated from a database of bathymetric soundings with interpolation between soundings guided by satellite-derived gravity data. The dataset is produced by GEBCO (http://www.gebco.net).
Known errors or discontinuities in the data set occur between regions where data is derived from satellite data and detailed bathymetric survey - this is evident in a grid pattern in the Southern North Sea Region, and a discontinuity at 0°E. Marine Scotland has highlighted errors where false banks occur on the shelf around the Shetland Isles (Hughes, 2014).
Figure 2-1 shows the GEBCO_08 bathymetry for the British Shelf and the source of the data. The discontinuity at 0°E and the grid pattern in the North Sea are clearly visible although this does not affect this model.
ETOPO-1 is a global DTM at 1 minute resolution produced by NOAA National Geophysical Data Center. The documentation states that this uses the GEBCO_08 data set for the British Shelf. Due to the lower resolution this dataset has not been considered further.
184.108.40.206 European Marine Observation and Data Network ( EMODnet)
The European Marine Observation and Data Network ( EMODnet) have produced DTMs for the Greater North Sea and Celtic Seas at 0.25 minute resolution (about 250m east-west direction and 450m north-south directions). The grids are based on bathymetric surveys and terrain models developed by external data providers including the UK Hydrographic Office ( UKHO), and the GEBCO_08 Grid 0.5 minute resolution dataset where no other data is available. Data sets are made available through the EMODnet website http://www.emodnet-hydrography.eu/. Further details of EMODnet are provided in Halcrow (2012).
Figure 2-2 shows where UK Hydrographic office data has been incorporated into the EMODnet dataset and the differences between the EMODnet and GEBCO_08 bathymetry. Comparison of the EMODnet and GEBCO_08 data sets shows significant differences where the data from the UKHO and other hydrographic offices has been included. Differences are generally greater in areas where the GEBCO_08 has been interpolated, and the UKHO data has been used in the EMODnet bathymetry, for example around 1.5°W 56.3°N, due east of the Firth of Tay. The large differences west of Norway are due to incorporation of Norwegian hydrographic office data. There are also differences north west of the British Shelf around Iceland, where the EMODnet data is sourced from the GEBCO_08 grid. However these have not been investigated as they are not considered important for the study area.
Due to the inclusion of the majority of the UKHO data, the EMODnet bathymetry is considered appropriate for use as the base bathymetry for model construction in areas where the resolution will be in the order of one kilometre. Higher resolution bathymetry data is however required in areas where the model mesh is finer to represent bed or flow features. Therefore other datasets are required as described below.
2.2.4 Hydrographic Data
Three sources of hydrographic survey data have been identified; the United Kingdom Hydrographic Office ( UKHO), the International Council for Exploration of the Sea ( ICES) and Marine Scotland's data sets.
The UKHO have a memorandum of understanding with Marine Scotland making their high resolution bathymetric survey available. Most of these data have already been incorporated into the EMODnet bathymetry, however further data has since become available. The location of the UKHO data in the WLLS model domain is shown in Figure 2-3a where it has been indicated on top of the EMODnet data. A closer view of Loch Linnhe is provided in Figure 2-3b.
The ICES surface dataset holds over 100 years of ship based observations, including soundings. There are over 2 million data points in the ICES data set within the study area, providing a good coverage over most areas. The ICES website ( http://ocean.ices.dk/) states that data are quality controlled by contributing organisation and visually inspected by experienced staff to further improve the quality of these data. However it is expected that due to the age of some of the sounding data and the differences in measurement methods, data logging and processing that there may be significant differences or scatter between the soundings. Marine Scotland used the ICES dataset to identify and correct anomalies in the GEBCO_08 data set off the coast of Shetland. See Halcrow, 2012, for more detail regarding hydrographic data and the differences observed between datasets.
2.2.5 NOOS 1.0
NOOS 1.0: A gridded dataset for the UK continental shelf at 1 arc-minute resolution was produced under the aegis of NOOS (an operational oceanography organisation for the NW European Shelf (see Halcrow, 2012 for more information). The NOOS bathymetry incorporates local datasets made available by oceanographic institutions in countries around the North Sea, however no detailed source attribution information is available for the bathymetry, and it was last revised in 2004. Bathymetric surveys collected by the UKHO post 2004 are therefore not incorporated in to the bathymetry, and it is uncertain to what extent earlier UKHO and other national hydrographic office datasets were incorporated.
After consideration of this data and comparison against other datasets (Halcrow, 2012) it was concluded that the NOOS bathymetry should not be used west of 0°E and has therefore not been used for the WLLS model.
2.2.6 Other data sources
Other identified data sources include digital Admiralty charts ( C-MAP) and SeaZone. However, these datasets were not used for this study due to licensing restrictions as discussed fully in Halcrow (2012). A licence enabling Halcrow to digitise the required Admiralty Charts was obtained from the Hydrographic Office and the digitising undertaken. This allows the data to be used into the future for this project without paying a licence fee every year. The digitised Admiralty Charts are used to fill the gaps in the digital bathymetry data available for the ECLH model.
2.2.7 Summary of bathymetry data availability for the Wider Loch Linnhe Area
Figures 2-3a shows data availability for the wider Loch Linnhe system; there is detailed bathymetric survey for the Sound of Mull, and Firth of Lorn, but not for the Firth of Jura. Figure 2-3b shows data availability within Loch Linnhe itself; there are some ICES ship tracks but no detailed bathymetric survey from the UKHO. Loch Linnhe and Loch Sunart are covered by Admiralty Charts. Figure 2-3 in Halcrow (2012), shows that there are considerable differences between the EMODnet bathymetry and the GEBCO_08 bathymetry where UKHO data has been incorporated into the EMODnet dataset in this region. In the outer Loch Linnhe system there is no additional bathymetry available from the UKHO in the EMODnet data, therefore based on differences found elsewhere there would be some uncertainty with the depths in this area. Both datasets are too coarse resolution to model the loch and surrounding lochs. Inaccuracies in the EMODnet bathymetry where it is derived from GEBCO_08 data mean that it will be necessary to correct and modify the coarse resolution data at the boundaries of the model domain.
SAMS has undertaken a number of bathymetric surveys in the years 2009-2011, including Loch Linnhe and Loch Etive and the Civil Hydrography Programme ( CHP) have planned surveys of Loch Linnhe, the sound of Jura, the Passage of Tiree, the outer approaches to the Firth of Lorne. These data sets apart from the Passage of Tiree were not available at the time of the mesh generation and were not used. Admiralty Chart data was used for Loch Linnhe and surrounding areas with limited bathymetry data.
2.3 Forcing Data
Forcing data is required for a yearlong climatological model run of the WLLS flow model, for six month long runs May to October 2011 and May to October 1991, and for calibration using observed data for approximate 1 month periods. The following forcing data is required;
- meteorological - including wind speed/stress, atmospheric pressure, surface heat flux, precipitation and evaporation
- hydrological - river flux
- oceanic open boundaries - including temperature, salinity and velocity
In addition, surface winds and offshore wave boundary data are required for the wave model.
2.3.2 Meteorological forcing
220.127.116.11 UK Met Office Model Data
Two data streams from the Met Office forecast models have been archived at NOC (Liverpool) for operational modelling:
- for operational tide-surge modelling on the continental
shelf, using the 2d tide-surge model (
- These data comprise of surface wind and atmospheric pressure only, at 1-h intervals, from Mid-May 1991 to present. From 1991 to 1995 the data is at 50 km resolution, post 1995 the data is at 12 km resolution.
- for Irish Sea Observatory operational modelling system,
running the 3d baroclinic hydrodynamic model,
on (i) the Atlantic Margin Model (
AMM, ~12km) and
(ii) the nested Irish Sea model (
The data comprise the following, from 2004 to 2007 with some
gaps, and continuously from 2007 to 2011, all at 12 km
- Global model output for the Atlantic at 6-hour intervals - 10m wind (E and N components); sea level pressure; low, medium and high level cloud coverage; specific humidity at 1.5m, air temperature at 1.5m; total accumulated precipitation;
- Mesoscale model output at 3-hour intervals - same variables
18.104.22.168 ECMWF Data
Additional meteorological forcing data was taken from the ERA ( ECMWF Re-Analysis) - Interim dataset http://www.ecmwf.int/en/forecasts/datasets. This data includes:
- Downwards longwave radiation and downwards shortwave radiation on a 0.75° grid, accumulated over 12 hours from midnight and noon. Available from 1989 to present.
- Evaporation and precipitation on a 0.75° grid, accumulated over 12 hours from midnight and noon. Available from 1989 to present, used in 1991 runs.
- Air temperature, sea level pressure, dew point temperature on a 0.125° grid at 6 hourly. Available from 1989 to present. Note that dew point temperature was used along with air temperature to calculate the relative humidity in the 1991 runs.
- U and V components of wind on a 0.125° grid at 6 hourly interval. Available from 1989 to present, used to fill in the gaps in the Met Office wind data in the 1991 runs.
22.214.171.124 Climatological Forcing
Climatological forcing was derived from the ERA40 and ERA-Interim datasets, which were used to force the POLCOMS AMM (~12km) model for the 45 year hindcast (1960-2004). See Wakelin et al. (2012) and Holt et al., (2012). A licence to use these data has been provided by the European Centre for Medium range Weather Forecasting ( ECMWF) for this study. A one-year climatological forcing for the temperature and salinity ( i.e. heat flux and precipitation) has been derived.
2.3.3 Meteorological observations
Loch Linnhe is approximately 6km wide and surrounded by mountains. The highest resolution Met office model output is at 4 km and local scale variations in the wind in this area is unlikely to be captured in the model. Therefore, the use of local measurements will also be made.
Local wind measurements around the Loch Linnhe system were made by Marine Scotland in 2011 and 2012 as part of an intensive measurement campaign in the Loch. Data is also available from 1991.
- Davis weather station data on Underwater Centre Pier in Fort William between July - November 2011
- Davis weather station data on Duart Point, Loch A'Choire and Cuil Bay between April - November 2011
In addition there is meteorological data for one station in 1991, as part of a previous intensive measurement campaign.
There are Met Office weather stations at Dunstaffnage, Aonach Mor, Colonsay and Tiree (although Aonach Mor is at high altitude). The Scottish Association of Marine Science ( SAMS) also maintain a weather station in Dunstaffnage http://dalriada.sams.ac.uk/sams_weather/.
The number of sites in 2011 are sufficient to carry out a correlation analysis for the wind conditions between the available data locations (obtained in 2011), to provide relationships of wind speed and direction for a range of directional sectors between these sites and data from the Met Office Mesoscale model.
Meteorological data at four weather stations around Loch Sunart were provided by Marine Scotland. However there was no overlap with the 2011 data, therefore this data could not be included in the main wind correlation analysis, however some consideration of the topographic influence on wind within Loch Sunart have been made.
Met Office model data from May 1991 is archived at NOCL at 25 km resolution, but no temperature and salinity forcing data are available for this time. For 2011 full forcing data from the Met Office models is archived at NOCL at 12km resolution.
2.3.4 Hydrological Data (Fresh Water Inflows)
In order to simulate the effect that river flow has upon salinity in coastal waters, river flux data are required. The Centre for Ecology and Hydrology ( CEH) Grid-to-Grid ( G2G) model was used to supply freshwater inflows to the various coastal models for this study.
The output that CEH provided from the G2G model were:
1. River discharge data (time series data) at all coastal locations in Scottish waters with the G2G model. The data cover 1 March 2007 to 30 September 2010 at 15 minute intervals.
2. River discharge data (time series data) at all coastal locations around Shetland and Northern Ireland with the G2G model. The data cover 1 March 2007 to 30 September 2010.
3. River discharge climatological data (long term daily/seasonal discharge data) at all coastal locations for Scotland (including Shetland) and Northern Ireland with the G2G model. Daily averaged data was provided, the averaging period covered 1962-2011.
In addition, Marine Scotland also provided the following river data sets for 1991.
1. River flux data for 3 rivers (River Lochy, River Nevis and Loch A'Choire)
2. Diffuse river inputs for Loch Linnhe and all side lochs
3. River temperature data (calculated using an empirical relationship relating the air temperature at Oban to river temperatures).
For the WLLS Model, the boundary data was derived from NOC-L's Atlantic Margin Model ( AMM) with a 12km resolution. Water levels along with temperature and salinity timeseries are applied at the model boundaries for specific periods coincident with times that calibration data is available. For the 1991 runs, boundary came from the POLCOMS model on a 12km grid. The WLLS climatology runs were forced using results from the Scottish Shelf model climatology run.
2.4 Calibration Data
Model calibration was undertaken against observation datasets for periods of up to 1 month. Calibration is required for water level, currents, temperature and salinity. In addition, the result of the 1 year climatology runs are compared against accepted general flow characteristics including current speed and direction (seasonal variability) and seasonal temperature and salinity cycles. Sections 2.4.2 to 2.4.4 present data found in freely available sources, however section 2.4.5 presents data from specific surveys targeting Loch Linnhe.
2.4.2 Water Level
Figure 2-4 shows all the locations of water level observations that are available in the WLLS region. These come from three main sources: tide gauge data from the BODC National Oceanographic Database ( NODB); bottom pressure data from the NODB, and analysed tidal data from NOC.
In addition, we have access to tidal data from TotalTide - a digital version of the UK Admiralty tide tables, from the UK Hydrographic Office. The locations of these datasets are shown in Figure 2-5. Because these data are based on harmonic analyses, water level estimates for any past or future date are obtainable, or via the use of constituents from the Admiralty tide tables. All available water level data available post year 2000 are shown in Figure 2-6.
Datasets on currents have been found from a number of sources; all locations are shown in Figure 2-7. These come from the BODC National Oceanographic Database ( NODB) and the TotalTide software, from UK Hydrographic Office. As Figure 2-8 shows, there are only a few datasets from the BODC National Oceanographic Database since year 2000. In some cases, vertical current profiles are available; these are shown in Figure 2-9.
The methodology used by TotalTide for calculating currents is not known. In addition, these data have been estimated for the use of shipping; therefore, a greater weighting may be placed on surface currents than currents near the sea bed.
The Atlas of UK Marine Renewable Energy Resources ( www.renewables-atlas.info) contains information on peak tidal current speeds over a mean spring and a mean neap tide. The dataset was derived from the POL HRCS Model, with peak spring and neap current speeds calculated from the major 2 or 4 tidal harmonics. Although this dataset is limited, it is freely available on a 0.0167° x 0.025° (latitude x longitude) grid throughout the region shown in Figure 2-10.
2.4.4 Temperature and Salinity
Temperature and salinity validation was carried out using selected hydrographic stations which were identified from the British Oceanographic Data Centre data holdings for UK. There are a very large number of datasets from CTD (Conductivity, Temperature and Depth) and bottle casts, both from the BODC National Oceanographic Database and the ICES database. Additionally, some of the CEFAS WaveNet buoys record sea surface temperature.
Figure 2-11 shows the locations of the temperature observations and Figure 2-12 shows the locations of the salinity observations. As Figure 2-13 shows, the temperature and salinity observations have occurred throughout the last two decades, with many observations throughout the model domain having occurred over the last two years. Figure 2-14 shows which of these observations include profiles over the entire water depth. Most temperature and salinity observations occurred at the same location and time.
In addition, the Ocean Data analYsis System for SEA ( ODYSSEA) dataset is a re-analysis of satellite observations of sea surface temperature. Daily mean average sea surface temperatures since 01/10/2007 have been obtained, on a 0.1° x 0.1° grid.
The results from the climatic run were compared with climatological atlas information for temperature and salinity, from the World Ocean Atlas ( WOA) and International Council for Exploration of the Seas ( ICES) climatological datasets.
2.4.5 Summary of data availability for the WLLS model including site specific survey data from 1991and 2011
Very few water level observations have been found near Loch Linnhe. Figure 2-6 shows water level observations since 2000; the majority of these data come from TotalTide. Tide levels at Tobermory, Mull exist since 1990. The "restricted" data point shown on Figure 2-6 between Mull and Oban is pressure recorder data collected between June and August 2010 by Marine Scotland.
Intensive measurement campaigns were undertaken in 1991, 2011 and 2012 and will be useful for the Loch Linnhe case study area. A summary of the data available in 2011 and 2012 is provided in Figure 2-15. The data collected consists of:
1991 Measurement campaign:
- CTD deployments from summer cruises in 1987-1991
- 1991, monthly repeated stations throughout the loch and monthly undulating tows in Loch Linnhe
- Current meters, 13 single-point current meter data mostly in the upper basin (1987-91) but in 1991 also in middle and outer basins. One ADCP instrument data from 1991
- Water level recorder data from four instruments in 1987, 1989, 1990 and 1991
- One thermistor chain data from 1990
- Thermistor chain data from 1991 from various locations in the loch
2011 measurement campaign:
From a dedicated field campaign carried out in 2011 (May and October) as part of the MSS sealice dispersal project:
- Repeat of main 1991 sampling locations
- Cross-sections in Upper and Lower Loch Linnhe
- At most plankton sampling sites
- CTD data from plankton sampling sites in December 2010
- Tracks of drifters, deployed during July and October 2011 field campaigns. and possibly in 2012
- Single-point current meter data at nine locations around the loch close to shore within 2 week periods in May and October 2011
- One profiling current meter for May and October 2011 close to Loch A'Choire
- Multi-parameter buoy surface current data North of the Corran Narrows from April - December 2011
- ADCP data at model boundaries, as described above
- Water level data from two pressure sensors deployed from April - November 2011 close to Sound of Mull and Fort William
All datasets have been made available.
Figure 2-8 shows the availability of other current data in this area since 2000 which was not part of the intensive measurement campaigns. In addition to the TotalTide data, the point shown between Mull and Jura represents ADCP (circled in orange on Figure 2-8) current measurements collected December 2008 and January 2009. These data appears to have been collected near the bed only.
Table 2-1 summarises the available data for two periods. Given that there exists meteorological forcing in 2011 as well as CTD and current measurements, this would suggest that a period or periods within 2011 would be the most suitable for model calibration. Calibration for temperature and salinity for the 1991 measurement campaign would be harder as the meteorological data held by NOC-L only has wind and pressure for this period (May 1991- present).
Therefore to conclude it is felt that there is sufficient data available for calibration and validation of the local Loch Linnhe model.
2.5 Conclusions and Recommendations
A review has been undertaken to identify data that are relevant to the setting up, forcing and calibration of the WLLS model. It has been found that there are datasets available providing coverage over a wide spatial and temporal field.
The EMODnet data is considered appropriate for use as the base bathymetry for model construction. This data formed our base coarser resolution data but was supplemented with higher resolution data.
Further UKHO data and other higher resolution datasets from ICES and Marine Scotland have been used to replace the coarser resolution data in areas that they overlap, with appropriate checks for consistency. However even with these data there are areas which have been identified in the data review report (Halcrow, 2012) as not having sufficient bathymetry data at a fine enough resolution. In this case data from digitised Admiralty Charts have been used.
2.5.2 Forcing data
For this case study tidal forcing, temperature and salinity data have been obtained from the NOC-L AMM model to provide boundary conditions to the WLLS model.
Meteorological forcing for the WLLS model will be derived from the Met Office model data that NOC-L holds. For Loch Linnhe, more localised wind data will also be used to develop an improved picture of the local variability in wind strength and direction by analysing observed wind data available in the locality and correlating with the coarser wind data thus providing higher resolution variability within the WLLS model domain.
The Met Office data provides wind data from 1991 to present day, however other parameters such as sea level pressure, low, medium and high level cloud coverage, specific humidity at 1.5m, air temperature at 1.5m, total accumulated precipitation and sensible heat flux are only available from 2007 to 2011. Data is available from ECMWF ERA-Interim datasets for the 1991 period.
Fluvial inputs are taken from G2G river flow data obtained from CEH for the WLLS area. CEH also carried out G2G runs to provide river data for 2011. Flux and temperature data for 3 rivers with in Loch Linnhe have been provided by Marine Scotland for the 1991 runs.
2.5.3 Calibration Data
In general there is sufficient data with which to undertake calibration for water level, currents, temperature and salinity by using the 2011 survey data. A summary of the dates where suitable calibration data is available is provided in Table 2‑1.
In summary we conclude that there are sufficient data for the calibration of the WLLS model using the data in 2011 and 1991; Figure 2-15 summarises the data available in 2011.
Table 2‑1 Available data
Loch Linnhe Wider system