An Assessment of the Performance of the PLUS+ Tool in Supporting the Evaluation of Water Framework Directive Compliance in Scottish Standing Waters

Phosphorus is one of the main causes of waterbodies in Scotland being at less than good ecological status (GES) in terms of the water framework directive (WFD). In Scotland, there are more than 8000 standing waters, defined as lakes and reservoirs that have a surface area of more than 1 hectare. Only about 330 of these are monitored routinely to assess compliance with the WFD. The export coefficient tool PLUS+ (phosphorus land use and slope) has been developed to estimate total phosphorus (TP) concentrations in the unmonitored sites; modelled values are then compared to WFD target concentrations for high, good, moderate, poor, and bad status to assess compliance. These type-specific or site-specific targets are set by the regulatory authority and form part of a suite of physical, chemical, and ecological targets that are used to assess GES, all of which must be met. During development, the PLUS+ tool was applied to 323 monitored catchments and 7471 unmonitored catchments. The efficacy of the tool was assessed against TP concentrations observed in 2014 and found to perform well in the rural catchments. 51% of standing waters had the same modelled and observed WFD class (i.e., High, Good, Moderate, Poor, Bad), and a further 40% of standing waters had a modelled WFD class that was within one class of observed water quality. The tool performed less well in catchments with larger inputs of TP from urban sources (e.g., sewage). The greatest deviations between measured and modelled classes were explained by the shortage of information on wastewater treatment works, fish farms, migratory birds, levels of uncertainty in TP measurements, and the amount of in-lake re-cycling of P. The limitations of the tool are assessed using data from six well documented case study sites and recommendations for improving the model performance are proposed.


Figure 1. PLUS+ TP and P sources.
The default model output is based on the 2011 census data for which a population density of 0.3 people per hectare has been used to define a threshold between urban and rural population. All waste water for urban populations is assumed to reach the surface water body via waste water treatment works (WWTW) and a per capita P load rate (loss coefficient) has been defined by default as 0.9125 kg P per capita. Waste water from rural populations is assumed to enter the surface water body via septic tanks and a per capita P load rate (loss coefficient) has been defined as a default value of 0.25 kg P per capita.

Calculation of surface water total P concentration
The overall total P load to a waterbody is calculated from summing the estimates of point sources, land cover total P load (see section 2.3.1 and 2.3.2) and P from upstream standing water bodies. This total load is used to calculate a total P concentration using OECD (Eutrophication of Waters, where [TP] is the standing water body total P concentration in micrograms per litre (µg/l), J is the sum of inputs of P in milligrams per year (mg/yr), R is the runoff (Equation 3), TW is the water residence time in years and a and b are model parameters.
For standing water bodies with a mean depth of >= 3 m, (a, b) = (1.55, 0.82). For shallow standing water bodies with a mean depth < 3 m, (a, b) = (1.02, 0.88). The values a and b for each water body are read from the LoadPrecursor table (see Section 6). Measured or modelled mean depths are available for all standing water bodies are also in the LoadPrecursor data set.
The water residence time, TW is calculated as follows: where A is the area of the standing water body in square metres (m 2 ), z is the mean depth in meters and R is the runoff in cubic metres per year (m 3 /yr). Runoff is available for all standing water bodies in the database. It is calculated in the following manner: where j is the local runoff in m/yr (this value was taken from the model NIRAMS [Dunn 2004]), A is the local catchment area in m 2 and sum of RUp is the sum of upstream runoff which can be calculated from upstream inflows.
The sum of inputs of phosphorus, J, is calculated in the following manner: where the first term is the sum of the different phosphorus loads for the different proportions of land cover classes, s is the sewage load (WWTW and septic tanks), x is the sum of any additional point source inputs and y is an alteration that a user may wish to make when running a scenario. The last term is the sum of upstream inputs. The equation is solved in a cascading manner, starting with 0 Strahler stream order standing waters and proceeding down a chain of connected catchments (each order summing the relevant and topologically determined upstream orders), making it possible to simulate TP concentrations for a series of connected standing water bodies.
The OECD equations used in PLUS+ are only valid for standing waters where annual TP load and annual turnover control the observed TP (OECD, 1982).

PLUS+ User Interface
The tool is designed to be used in a production environment using Esri ArcGIS. The tool has flexible inputs of data sets through pull-down menus, management of scenarios through tabbed windows and it provides the user with full control of the model parameters. This is described in detail in the user guide [1].

Figure 2. The main PLUS+ user interface
PLUS+ uses WFD water quality breakpoints and the SEPA red-amber-green traffic light system to alert the user to how close water bodies' TP concentration is to a water quality boundary (e.g. from moderate to poor) and their capacity for the uptake of additional P before status is downgraded to a lower category. PLUS+ also calculates how much P would be required to be removed from a water body for the status to be improved. In doing these calculations PLUS+ takes account of in-flows and out-flows to the catchment being analysed.

Per capita sources of P
The UK 2011 census data was used as the primary source of information for determining the location and amount of P from waste water treatment plants and sceptic tanks.
Data zone boundaries (the key geography for the dissemination of small area statistics in Scotland) and Ordnance Survey MasterMap residential address locations (the most comprehensive address location data) were used to calculate a mean count of persons per residential property for each data zone. A raster population density map was then created from the residential property locations and the calculated mean number of persons per property from the appropriate data zone. PLUS+ differentiates between urban and rural per capita P exports and uses a threshold of 0.3 persons per hectare as a population density boundary. Following trials with a range of mapping unit sizes, a grid of 500 m (25 Ha) was chosen to aggregate the population to determine whether the locations were urban or rural. While this was found to result in many small hamlets being described as urban it was the minimum resolution required to provide an acceptable degree of precision for the depiction of larger urban units. This revision to the population data has resulted in an increase to the modelled P load of the 8030 sub-catchments of 96,000 Kg compared to the previous version of PLUS+. The largest increase is in the sub-catchment from south-west of Edinburgh to Leith docks which has seen its urban population increase from 181,585 to 198,819 with a resulting increase in P load of 15,702 kg. Of the 8030 subcatchments 1,043 (13 %, mean increase 99.3 kg) have seen their P load increase and 993 (12 %, mean reduction 7.2 kg) have seen their P load decrease as a result of the updating of the population model.

WFD Breakpoints and Thresholds
WFD breakpoints are used to determine the water body status -High, Good, Moderate, Poor, Bad.
The WFD breakpoint data is supplied by SEPA.

Figure 3 Distribution of TP break point values
The Scottish Environment protection Agency (SEPA) uses a traffic light system of red/amber/green to highlight catchments with a P concentration which is close to a threshold for a less favourable condition (status). The PLUS+ display of the status of each sub-catchment includes a graphic showing the P concentration of the selected catchment with its current calculated status and showing the distance to the higher and lower thresholds (Figure 4). If a concentration is within 3 % of the interval between two adjacent class boundaries above the lower bound then it is classed as red. If the concentration is between 3 % and 20 % of the interval above the lower bound then it is classed as orange, otherwise it is classed as green. ScenarioData.mdb largely duplicates the SourceData above but with names suffixed with "_S" to denote scenario tables. SlopeClass_LandCover and FlowRouting are not included as they will not be modified. Additionally ScenarioData.mdb includes a Note that the two SEPA monitoring tables are not duplicated within the scenario structure so that whether the baseline data or scenario data are being loaded the SEPA tables must be selected first. Additionally, the measured values are not incorporated in the PLUS+ modelling calculations, they are included in the reports and output windows for information and comparison with the modelled values. Point sources of P added to the model by a user through PLUS+ will be stored in the scenario table PointSource_S. Any number of point sources may be added by the user (only subject to operating system and database restrictions). This table can also be edited manually to add point sources to a scenario. The scenario database does not include a "FlowRouting" table as the topological relationships of standing water bodies will not be modified by any of the scenario creations.
The Visual Basic source code and the data tables are available in a GitHub repository at https://github.com/DavidHutton/plus [2]. The computer code has comments to guide comprehension and it includes detailed descriptions of the data tables required. These are also described in the user guide [1]. Please note that this software will only run under the Visual Basic for Applications environment of the Esri ArcMap software and that a valid Esri licence is required.

Export coefficients
The minimum, maximum and median land cover export coefficients are detailed in Table 4. In sensitivity testing each of the minimum, median and maximum settings were used in model runs (i.e. a run was made using the minimum settings and then using the median settings etc.), in all other analyses the median values were used.   Table 5. This table shows that 67 % of the land described as "heather moor" in the original land cover (first column) was updated to "coarse grassland" in the revised land cover (second column). While the old and revised land covers did not match (the value in column "Match" = "No") there was no change in the associated P export as the Original to Revised export coefficients are both 0.06 kg/yr/ha (final column).
The impact of change in land cover is discussed in the sensitivity analysis section of the report.