The BioDT Recreational Potential Model for Scotland

Technical Supplement

Authors
Affiliation

Joe Marsh Rossney

UK Centre for Ecology & Hydrology

Maddalena Tigli

Chris Andrews

Jan Dick

Simon Rolph

Published

June 24, 2025

Modified

June 24, 2025

Abstract

This supplement provides technical details about the Recreational Potential model. This includes a mathematical description of the model, details regarding the source data, and technical aspects of the code.

1 Technical description of the model

While there have been several publications citing the RP model, so far none have included a detailed technical description of the model itself.

1.1 Presence/absence rasters

We begin by defining some notation.

Let \(I : \mathbb{R}^2 \to \{0, 1\}\) be a binary-valued function that maps pairs of coordinates \(x \equiv (x_1, x_2) \in \mathbb{R}^2\) to either \(0\) or \(1\). This is a mathematical representation of a single raster layer in a ‘one-hot’ representation, where a value of \(I(x) = 1\) indicates presence of a feature at the coordinate \(x\), and \(I(x) = 0\) indicates absence.

The RP model currently includes 87 features/items grouped into four components (see Section 3). Each of these has its own presence/absence raster. We will therefore introduce two subscript indices and write each layer as \(I_{ij}(x)\), where

  • \(i \in \{1, 2, 3, 4\}\) labels the component to which \(I_{ij}\) belongs, as specified in Table 1, and
  • \(j \in \{1, 2, \ldots, n_i\}\) labels the item/feature within that component, of which there are \(n_i\).

As a concrete example, if \(j=1\) labels the ‘pond’ item, then \(I_{41}(x)\) takes a value of \(1\) whenever there is a pond at \(x\), and a value of \(0\) elsewhere.

\(i\) Component Short code
1 Landscape SLSRA
2 Natural Features FIPS_N
3 Infrastructure FIPS_I
4 Water Water
Table 1: Integer label and short-code for the four components of the RP model.

1.2 Proximity contributions

For certain features like rivers and lakes it does not make sense to restrict their contributions to the Recreational Potential to the pixels in which they are present. One does not need to be standing in a lake to appreciate it!1

It makes sense to attempt to account for the presence of these features nearby.

Let \(\mathcal{P}_{ij} \subset \mathbb{R}^2\) denote the set of coordinates in which the feature \((ij)\) is present, i.e.2

\[ \mathcal{P}_{ij} := \left\{ x' \in \mathbb{R}^2 \mid I_{ij}(x') = 1 \right\} \, . \tag{1}\]

We first compute a ‘distance raster’ in which each pixel is assigned a value that is the (Euclidean) distance to the nearest pixel in which the feature \((ij)\) is present,

\[ d_{ij}(x) = \min_{x'\in \mathcal{P}_{ij}} \lVert x - x' \rVert_2 \, , \tag{2}\]

where \(\lVert x - x' \rVert_2 = \sqrt{(x_1 - x'_1)^2 + (x_2 - x'_2)^2}\).

These distances must be mapped to the unit interval \([0, 1]\) such that \(0 \mapsto 1\) and \(\infty \mapsto 0\), i.e. so that presence of the feature in this pixel results in the highest value of \(1\), tailing off to \(0\) when the feature is very far away.

There are many different ways to achieve this. Following Paracchini et al. (2014, p. 374 – Eq. 1), the function chosen is a smooth, monotonically decreasing function in the logistic family,

\[ u(d_{ij}(x) \,; \alpha, \kappa) = \frac{\kappa + 1}{\kappa + \exp(\alpha \, d_{ij}(x))} \, . \tag{3}\]

It remains to specify values for the parameters \(\kappa\) and \(\alpha\), the latter of which is most important since it determines the rate at which the value of \(u\) declines as the distance to the nearest feature increases.

The parameters chosen in the original version of this code were3

\[ \kappa = 6 \, , \qquad \alpha = 0.01011 \, , \tag{4}\]

resulting in a function, Figure 1, that halves at \(d\approx200\) (metres) and drops to \(0.1\) at \(d\approx400\).4 At this distance, features with a user-provided score of \(10\) (i.e. the highest) would be equivalent to a local feature given a score of \(1\) (the lowest). It is safe to assume that at distances greater than this the contribution from the feature would be indiscernible to the user.

Code
alpha <- 0.01011
kappa <- 6
logistic_func <- function(d) {
    (kappa + 1) / (kappa + exp(alpha * d))
}

d <- seq(0, 500, by = 1)
u <- logistic_func(d)

plot(d, logistic_func(d), ylim=c(0, 1), type="l", lwd=2,
    xlab=expression("d"), ylab=expression("u(d, alpha, kappa)"))
Figure 1: A plot of the function which maps positive distances to the unit interval.

1.3 Calculation of a single component

As mentioned previously, there are 87 features in total, each of which corresponds to (a) persona score between \(0\) and \(10\),

\[ \rho_{ij} \in \{0, 1, \ldots, 10\} \, , \tag{5}\]

and (b) a raster layer \(f_{ij}(x)\). For the landscape and natural features components (i.e. \(i = 1,2\)) the feature rasters are precisely the binary presence/absence rasters discussed previously,

\[ f_{ij}(x) = I_{ij}(x) \, , \qquad i = 1, 2 \, . \tag{6}\]

For the infrastructure and water components (\(i=3, 4\)) the features rasters contain the proximity contributions described above,

\[ f_{ij}(x) = u(d_{ij}(x) \,; \alpha, \kappa) \, , \qquad i = 3, 4 \, . \tag{7}\]

Each component is calculated as a simple weighted sum of the features contained within it,

\[ F_i(x \,; \rho_i) = \sum_{j=1}^{n_i} \rho_{ij} f_{ij}(x) \, , \tag{8}\]

where we have permitted \(\rho_i\) to denote the set of persona scores \(\{\rho_{i1}, \rho_{i2}, \ldots, \rho_{i n_i} \}\) for the \(i\)-th component.

1.4 Domain-dependent rescalings

Let \(F \vert_{\mathcal{R}}\) denote the restriction of a function \(F\) to a subdomain \(\mathcal{R} \subset \mathbb{R}^2\) which defines the ‘region of interest’ in which we wish to compute Recreational Potential. We now define a family of rescaled functions,

\[ \hat{F}(x \,; \mathcal{R}) := \frac{F(x) - \min F \big\vert_{\mathcal{R}}}{\max F \big\vert_{\mathcal{R}} - \min F \big\vert_{\mathcal{R}}} \, , \tag{9}\]

each of which corresponds to a particular choice of restriction, i.e. a particular region of interest.

Assuming that \(F(x)\) is not a constant function on \(\mathcal{R}\),5 these rescaled functions are ‘normalised’ to the unit interval within the region of interest, i.e.

\[ \min_{x\in\mathcal{R}} \hat{F}(x \,; \mathcal{R}) = 0 \, , \qquad \max_{x\in\mathcal{R}} \hat{F}(x \,; \mathcal{R}) = 1 \, . \tag{10}\]

It is important to be aware that with this approach to rescaling introduces a dependence on the region of interest. That is,

\[ \hat{F}(x \,; \mathcal{R}_1) \neq \hat{F}(x \,; \mathcal{R}_2) \, . \tag{11}\]

Thus, the RP value at any given pixel should be expected to change when a different area is selected, even if the persona and the input data stays the same.

1.5 Calculation of Recreational Potential

To compute the Recreational Potential value, the four rescaled components are summed,

\[ F_{\mathrm{RP}}(x \,; \rho, \mathcal{R}) = \sum_{i=1}^4 \hat{F}_i(x \,; \rho_i , \mathcal{R}) \, , \tag{12}\]

and rescaled \(F_{\text{RP}} \longrightarrow \hat{F}_\text{RP}\) using Equation 9 once more (with the same \(\mathcal{R}\)) so that the result falls in the interval \([0, 1]\).

2 Production of input rasters

The data production, or raster pre-processing stage refers to the production of the rasters that are directly multiplied by the persona scores in the dynamic part of the model. That is, the output of this stage is four multi-layered rasters \((f_1, f_2, f_3, f_4)\) corresponding to the four components, where \(f_i := (f_{i1}, f_{i2}, \ldots, f_{i n_i})\), whose individual layers \(f_{ij}\) are defined by Equation 6 (for \(i = 1, 2\)) and Equation 7 (for \(i=2,3\)).

The process of producing these rasters is currently split into two distinct parts:

  1. A manual part in which categorical rasters are constructed from raw downloaded datasets using QGIS
  2. A scripted part in which these categorical rasters are converted into the form required by the model, using R.

The scripted part is executed by running inst/scripts/data_production/main.R (or the containerised version) with the manually-produced categorical rasters already present. This script mostly just calls functions defined in R/data_production.R.

2.1 Manipulation in QGIS

2.1.1 Landscape component

The Landscape component raster consists of 40 layers, each representing one of its features (see Section 3 for full list). These features can be grouped into different sub-categories:

  • CP (1 feature) : The Country Parks (CP) feature was sourced from NatureScot (2020).

  • HNV (1 feature): The High Nature Value farming feature was sourced from European Environment Agency (2023).

  • LCM (28 features): The Land Cover Map (LCM) featured were sourced from NatureScot and Space Intelligence (2023).

  • NNR (1 feature): The Natural Nature Reserves (NNR) feature was sourced from NatureScot (2022a).

  • NP (1 feature): The National Parks (NP) feature was sourced from Stirling Council (2024)

  • NR (1 feature): The Nature Reserves (NR) feature was sourced from NatureScot (2021).

  • RP (1 feature): The Regional Parks (RP) feature was sourced from NatureScot (2022b).

  • RSPB (1 feature): The Protected sites for birds feature was sourced from Royal Society for the Protection of Birds (2024).

  • SAC (1 feature): The Special Area for Conservation (SAC) feature was sourced from NatureScot (2023a).

  • SPA (1 feature): The Special Protection Areas (SPA) feature was sourced form NatureScot (2023b).

  • SSSI (1 feature): The Sites of Specific Scientific Interest (SSSI) feature was sourced from NatureScot (2024).

  • SWT (1 feature): The Scottish Wildlife Trust (SWT) feature was sourced from Scottish Wildlife Trust (2024).

  • WLA (1 feature): The Wild Land Area (WLA) feature was sourced form NatureScot (2014).

The datasets used for this component were clipped to a Scotland Boundary shapefile (Office for National Statistics, 2018), and they were rasterized to a 20m x 20m resolution.

2.1.2 Natural Features component

The 24 features in this component are grouped into three main categories:

  • Landform features (16 features): The landform data was extracted from the Scotland 250K Soils Map. (The James Hutton Institute, 2022). The following methodologies were applied:

    1. Dissolve: the dataset was “dissolved” in QGIS using the “Landform“ attribute, reducing it to 334 unique polygon features.

    2. Reclassification: the 334 landform types were manually grouped into 16 broader categories based on descriptive keywords (e.g. “hills”, “foothills” etc…). See Section 5.

    3. Rasterization: the dataset was converted into a raster at a 20 m x 20 m resolution.

  • Slope features (6 features): Slope data were derived from the Copernicus GLO-30 Digital Elevation Model (DEM), a global 30 m resolution dataset (European Space Agency, 2021). The following methodology was applied:

    1. Slope calculation: the DEM was converted into a matching CRS (EPSG: 27700) using Warp (Reproject) using QGIS, with a slope raster calculated using GDAL Slope with its default settings.

    2. Up-sampling: the raster was resampled to a 20 m x 20 m resolution.

  • Soil features (2 features): The landform data was derived from the Scotland 250K Soils Map (The James Hutton Institute, 2022) The following methodology was applied:

    1. Initial dissolve: The dataset was “dissolved” by field “genSoilTyp” in QGIS, reducing it to 12 unique attribute values.

    2. Grouping: The values were manually aggregated into two soil classes: organic (peat) and mineral based on the original classifications (see Section 5).

    3. Rasterization: the dataset was converted into a raster at a 20 m x 20 m resolution.

2.1.3 Infrastructure Features component

The 10 features are grouped into three main categories:

  • Roads Tracks (5 features): Road Tracks data was sourced from the OS OpenRoads dataset (Ordnance Survey, 2023b). The following methodologies were applied:

    1. Dissolve: The OpenRoads dataset (Ordnance Survey, 2023b) was “dissolved” in QGIS by the field “Road type” and “Tracks”.

    2. Adding access roads: the attribute field “Access roads” from the OSM_Roads dataset (OpenStreetMap, n.d.) was merged into the dissolved dataset.

    3. Rasterization: the combined dataset was rasterized at a 20 m x 20 m resolution.

  • Cycle Network (4 features): The National Cycle Network data was sourced from the Sustrans National Cycle Network dataset (Sustrans GIS, 2024). The following methodology was applied:

    1. Dissolve: The dataset was “dissolved” in QGIS using the two attribute fields “Desc_” and “Surface”.

    2. Grouping: The “Desc_” features were then grouped in (1) on road, and (2) off road; whilst the “Surface” features grouped in (3) paved and (4) unpaved. This produced four distinct feature classes

    3. Rasterization: The dataset was converted into raster format at a 20 m x 20 m resolution.

  • Local Path Network (1 feature): The “Local Path Network” data was sourced from the Scotland’s Core Paths dataset provided by the Improvement Service (2024). The dataset was converted into raster format at a 20 m x 20 m resolution.

2.1.4 Water component

The 13 features can be divided into two categories:

  • Rivers (6 features): The river data was sourced from the Os Open Rivers (Ordnance Survey, 2023a) and the OS Boundaries dataset. The following methodology was applied:

    1. merging: To avoid naming inconsistencies, the “Region” field from the OS Boundaries dataset was joined to the Os Open Rivers dataset (Ordnance Survey, 2023a).

    2. initial dissolve: the merged dataset was “dissolved” in QGIS using the “name” and “Region” fields. This combined river segments sharing the same name and region into single features, reducing the dataset into 74 997 features.

    3. Reclassification: To simplify the river data at a national scale, a reclassification was carried out (see [Appendix I: Reclassification info] for examples). This is a subjective and based on:  the “form” attribute (indicating water type), or keywords searches in the “name” field, and researcher knowledge of major rivers of national and regional importance.

    4. Rasterization: the dataset was converted into raster format at a 20 m x 20 m resolution.

  • Lakes (7 features): The lakes data was sourced from the Spatial inventory of UK waterbodies dataset (Taylor, 2021). The following methodology was applied:

    1. Reclassification: The dataset contained 25404 features for Scotland. To make it into a relevant and scorable dataset it was reduced to six features based on the given size (in hectares) of a waterbody using the attribute “Poly_Area_Ha” (see Section 5).

    2. Rasterization: the datasets were converted into raster format at a 20 m x 20 m resolution.

2.2 Quantising the slope raster

With one exception, every raster layer produced by the manual process is an integer-valued categorical raster. The single exception is a ‘slope’ raster containing the average absolute gradient \(s(x)\) within each pixel.

We use terra::classify to quantise this continuous range into six categories according to their steepness,

\[ C_{\text{slope}}(x) = \begin{cases} 1 & &0 \leq s(x) < 1.72 \\ 2 & &1.72 \leq s(x) < 2.86 \\ 3 & &2.86 \leq s(x) < 5.71 \\ 4 & &5.71 \leq s(x) < 11.31 \\ 5 & &11.31 \leq s(x) < 16.7 \\ 6 & &16.7 \leq s(x) \, . \end{cases} \tag{13}\]

These classification intervals differ from those used in Dick et al. (2022) but are consistent with the earliest known R implementation of the RP model.

2.3 Reprojection

The categorical rasters produced in the previous stage do not necessarily share the exact same coordinate grid. They can be coerced to share the same grid using terra::project.

We projected all rasters onto a reference raster with the following properties:

  • Coordinate reference system (CRS): EPSG:27700 (British National Grid)
  • Resolution: 20 metres x 20 metres
  • Extent: \((x_\text{min}, x_\text{max}, y_\text{min}, y_\text{max}) = (-10000, 660000, 460000, 1220000)\), which is a rectangular box around Scotland.

We used the ‘nearest’ interpolation mode (method="near") since the data is categorical. The reprojected rasters therefore contain the same set of integer values as the original rasters.

2.4 Conversion to one-hot representation

The next stage converts the categorical rasters (\(C_i(x) \in \{0, 1, \ldots, n_i\}\) into a set of presence/absence rasters for each feature. These binary-valued rasters are referred to as the one-hot representation of the data.

Consider the following categorical raster,

\[ C_i = \begin{bmatrix} \ddots \\ & 1 & 0 & 0 \\ & 2 & 1 & 0 \\ & 2 & 1 & 3 & \\ & & & & \ddots \end{bmatrix} \, . \tag{14}\]

The one-hot encoding of this raster is written

\[ C_i \overset{\text{one hot}}{\longrightarrow} ( I_{i0}, I_{i1}, I_{i2}, I_{i3}, \ldots) \, , \tag{15}\]

where the binary-valued presence/absence rasters are, in this case,

\[ \begin{align} I_{i0} &= \begin{bmatrix} \ddots \\ & 0 & 1 & 1 \\ & 0 & 0 & 1 \\ & 0 & 0 & 0 & \\ & & & & \ddots \end{bmatrix} \, , \quad I_{i1} = \begin{bmatrix} \ddots \\ & 1 & 0 & 0 \\ & 0 & 1 & 0 \\ & 0 & 1 & 0 & \\ & & & & \ddots \end{bmatrix} \, , \\ I_{i2} &= \begin{bmatrix} \ddots \\ & 0 & 0 & 0 \\ & 1 & 0 & 0 \\ & 1 & 0 & 0 & \\ & & & & \ddots \end{bmatrix} \, , \quad I_{i3} = \begin{bmatrix} \ddots \\ & 0 & 0 & 0 \\ & 0 & 0 & 0 \\ & 0 & 0 & 1 & \\ & & & & \ddots \end{bmatrix} \, . \end{align} \tag{16}\]

The implementation of this step involves the use of terra::lapp to parallelise the calculation over the spatial domain. In practice we use NA instead of 0 to denote absence.

one_hot_pixel <- function(x) {
    out <- matrix(0, nrow = length(x), ncol = length(feature_mapping))
    for (i in seq_along(feature_mapping)) {
        out[, i] <- ifelse(x == as.numeric(feature_mapping[i]), 1, NA)
    }
    return(out)
}

layer <- terra::lapp(
    layer,
    fun = one_hot_pixel,
    filename = outfile,
    overwrite = TRUE,
    wopt = list(
        names = names(feature_mapping),
        datatype = "INT1U"
    )
)
1
Take a single pixel value (x) of a single-layered raster and return the one-hot representation at that point, i.e. a vector of n_i values of which up to one is 1 (present) and the remainder are NA (absent).
2
Apply this function in a vectorised fashion to every pixel of the input faster.
3
Write the outputs to outfile as they are computed, to reduce memory requirements.

2.5 Stacking into component rasters

This stage is nothing more than combining many single-layered rasters files into a single multi-layered raster for each component. Roughly speaking, we do

# Construct a list of rasters from each .tif file in `infiles`
rasters <- lapply(infiles, terra::rast)

# Construct a single multi-layered raster from this list
stacked <- terra::rast(rasters)

# Re-apply layer names
layer_names <- unlist(lapply(rasters, names))
names(stacked) <- layer_names

# Write the combined raster to disk
terra::writeRaster(stacked, outfile)

Note that terra::rast(rasters) does not preserve layer names when rasters is a list of rasters, hence the need for the two lines prior to writeRaster.

A warning about layer names

It is very unfortunate that the names we have used for the layers follow the same format (filename_i for integers i) as the names that are automatically generated by terra::rast. Because of this, if these two lines that re-apply layer names are omitted, it is not obvious that anything is wrong; no error will be thrown and one can go onto compute RP values, since the names are as expected.

However, the results will be wrong, since terra::rast will apply the right names in the wrong order.

Future developers are strongly recommended to fix this problem by using different layer names. This issue and suggested solutions are discussed in depth in the Developers’ Report.

2.6 Proximity contributions

This stage applies Equation 2 and Equation 3 to each layer of the Infrastructure and Water components.

In fact Equation 2 is the most challenging computation in the whole processing pipeline, due to the fact that its memory requirements scale poorly with the size of the input raster.

Every single other processing step is entirely local; all of the information required to generate an output value for a pixel at coordinate \(x\) is contained in the input pixels at coordinate \(x\). This makes it trivial to process the data in chunks, and memory costs scale linearly with the area of the chunks.

In contrast, calculating distance to the nearest feature is strongly non-local; the output value at \(x\) is a function of the full set of input values. Naively, the entire raster must be loaded to compute the output at a single location.

To perform this calculation we used terra::distance. It is not necessarily true that the underlying C++ implementation of GDALComputeProximity, which is called by terra::distance, actually loads the entire raster into memory immediately. However, for Scotland at 20 metre resolution, we found that the memory requirements for this distance calculation (several 10s of GB) surpass the typical memory available on a standard personal computer by an order of magnitude.

To get around this issue, the original Scotland-wide raster was divided into 20 overlapping spatial ‘windows’. Note that the size of the overlap needed to be at least as large as the effective cutoff distance of 500m. After the distance step was complete within each window, the windows were stitched together, discarding the overlaps.

The distance raster was also truncated at 500m, setting any values above this to NA which simply passes through the function applying Equation 3 and ensures that we don’t waste disk space storing high-precision floats that are functionally equivalent to zero.

2.6.1 Visual demonstration

Here we provide a visual demonstration of how a ‘proximity’ raster is obtained from a raster containing just the presence/absence data. The example provided is for the FIPS_I_RoadsTracks_2 feature layer, which corresponds to ‘A’ roads.

The presence/absence raster traces out the path of the road. The distance from each cell to the nearest present cell is then calculated using the terra::distance() function . The resulting raster, labeled distance in the example map, contains the distance in meters from the center of each cell to the nearest feature. Application of Equation 3 results in the ‘proximity’ layer.

The three layers can be toggled using the control panel in the top-right corner.

3 List of items/features

Here we list the full set of 87 items/features that make up the layers of the input rasters, and must be scored by the user to construct a persona.

Component Short code Num. layers Description
Landscape SLSRA 40 Land cover type, landscape designations and conservation, and farmland of high nature value.
Natural Features FIPS_N 24 Landform types, soil types and slope.
Infrastructure FIPS_I 10 Roads, tracks, footpaths and cycle networks.
Water Water 13 Lakes and rivers.

3.1 Landscape component

Name Description
SLSRA_CP_2 Country Park
SLSRA_HNV_2 Designated High Nature Value (HNV) farmland
SLSRA_LCM_1 Alpine and subalpine grassland
SLSRA_LCM_2 Arable land and market gardens
SLSRA_LCM_3 Arctic, alpine and subalpine scrub
SLSRA_LCM_4 Bare field or exposed soil
SLSRA_LCM_5 Base-rich fens and calcareous spring mires
SLSRA_LCM_6 Broadleaved deciduous woodland
SLSRA_LCM_7 Built-up area
SLSRA_LCM_8 Coastal dunes and sandy shore
SLSRA_LCM_9 Coastal shingle
SLSRA_LCM_10 Dry grassland
SLSRA_LCM_11 Freshwater
SLSRA_LCM_12 Inland cliffs, rock pavements and outcrops
SLSRA_LCM_13 Lines of trees, small planted woodlands, early-stage woodland and coppice
SLSRA_LCM_14 Littoral sediment or saltmarsh
SLSRA_LCM_15 Mesic grassland
SLSRA_LCM_16 Mixed deciduous and coniferous woodland
SLSRA_LCM_17 Non-native coniferous plantation
SLSRA_LCM_18 Raised and blanket bog
SLSRA_LCM_19 Riverine and fen scrubs
SLSRA_LCM_20 Rock cliffs, ledges and shores
SLSRA_LCM_21 Scots pine woodland
SLSRA_LCM_22 Screes
SLSRA_LCM_23 Seasonally wet and wet grassland
SLSRA_LCM_24 Temperate montane scrub
SLSRA_LCM_25 Temperate shrub heathland
SLSRA_LCM_26 Valley mires, poor fens and transition mires
SLSRA_LCM_27 Windthrown woodland
SLSRA_LCM_28 Woodland fringes and clearings and tall forb stands
SLSRA_NNR_2 National Nature Reserve (NNR)
SLSRA_NP_2 National Park
SLSRA_NR_2 Nature Reserve
SLSRA_RP_2 Regional Park
SLSRA_RSPB_2 Royal Society for the Protection of Birds (RSPB) Reserve
SLSRA_SAC_2 Special Area of Conservation (SAC)
SLSRA_SPA_2 Special Protection Area (SPA)
SLSRA_SSSI_2 Site of Special Scientific Interest (SSSI)
SLSRA_SWT_2 Scottish Wildlife Trust Reserve
SLSRA_WLA_2 Wild Land Areas
Table 2: Description of items/features in the Landscape component.

3.2 Natural Features component

Name Description
FIPS_N_Landform_1 Foothills
FIPS_N_Landform_2 Mountains
FIPS_N_Landform_3 Terraces
FIPS_N_Landform_4 Flood plain
FIPS_N_Landform_5 Beaches or Dunes
FIPS_N_Landform_6 Rocks or Scree
FIPS_N_Landform_7 Depressions
FIPS_N_Landform_8 Hills
FIPS_N_Landform_9 Lowlands
FIPS_N_Landform_10 Rock Walls
FIPS_N_Landform_11 Uplands
FIPS_N_Landform_12 Valley sides
FIPS_N_Landform_13 Valley bottom
FIPS_N_Landform_14 Built-up areas
FIPS_N_Landform_15 Saltings
FIPS_N_Landform_16 Hummocks, mounds or moraines
FIPS_N_Slope_1 No slope
FIPS_N_Slope_2 Gentle slope
FIPS_N_Slope_3 Medium slope
FIPS_N_Slope_4 Steep slope
FIPS_N_Slope_5 Very steep slope
FIPS_N_Slope_6 Extremely steep slope
FIPS_N_Soil_1 Peat or Organic
FIPS_N_Soil_2 Mineral
Table 3: Description of items/features in the Natural Features component.

3.3 Infrastructure Features component

Name Description
FIPS_I_LocalPathNetwork_2 Path
FIPS_I_RoadsTracks_4 Minor or local road
FIPS_I_RoadsTracks_2 A Road
FIPS_I_RoadsTracks_1 Motorway
FIPS_I_RoadsTracks_5 Access roads or Track
FIPS_I_RoadsTracks_3 B Road
FIPS_I_NationalCycleNetwork_1 On Road: Paved Surface
FIPS_I_NationalCycleNetwork_2 Traffic Free: Unpaved Surface
FIPS_I_NationalCycleNetwork_3 Traffic Free: Paved Surface
FIPS_I_NationalCycleNetwork_4 On Road: Unpaved Surface
Table 4: Description of items/features in the Infrastructure Features component.

3.4 Water component

Name Description
Water_Lakes_1 Pond
Water_Lakes_2 Lochan
Water_Lakes_3 Small Lochs
Water_Lakes_4 Medium Lochs
Water_Lakes_5 Large Lochs
Water_Lakes_6 Major Lochs
Water_Rivers_1 Minor river or tributary
Water_Rivers_2 Unnamed minor stream or tributary
Water_Rivers_3 Major river or tributary
Water_Rivers_4 Named minor stream or tributary
Water_Rivers_5 Lake
Water_Rivers_6 Tidal river or estuary
Water_Rivers_7 Canal
Table 5: Description of items/features in the Water component.

4 List of data sources

Data sources used for the creation of the underlaying raster datasets.
Name Component Description Source Attribute statement Conditions of use Data Owner/IPR Version
Scotland Boundary NA Shapefile of the Scotland boundary - NUTS level1 Office for National Statistics, 2018 Contains public sector information licensed under the Open Government Licence v3.0. Contains OS data © Crown copyright and database right 2024 Freely available for any use (public domain or CC0) UK Govt. Jan-18
Core paths network FIPS_I shapefile containing the core path network Improvement Service, 2024 Contains public sector information licensed under the Open Government Licence v3.0. NA Improvement Service 15/02/2024
Roads FIPS_I Shapefile contains road types Ordnance Survey, 2023b Contains Ordnance Survey data licensed under the Open Government Licence v3.0. Freely available for any use (public domain or CC0) Ordnance Survey (OS) Oct-23
Tracks FIPS_I Shapefile containing road types within OpenStreetMap (OSM) filtered for tracks. OpenStreetMap, n.d. Map data copyrighted OpenStreetMap contributors and available from https://www.openstreetmap.org Freely available for any use (public domain or CC0) Ordnance Survey (OS) NA
National Cycle Network FIPS_I shapefile containing the National Cycle Network Sustrans GIS, 2024 Contains public sector information licensed under the Open Government Licence v3.0. NA NA 18-Feb-24
Landform classes from Scotland Soil Map FIPS_N A shapefile containing polygons of the soil types and landorm at a 250k:1 scale The James Hutton Institute, 2022 National Soil Map of Scotland v1.4 copyright and database right The James Hutton Institute 2024. Used with the permission of The James Hutton Institute. All rights reserved. Any public sector information contained in these data is licensed under the Open Government Licence v.2.0 Freely available with obligation to quote the source James Hutton Institute 1.4
Slope FIPS_N Created from Copernicus GLO-30 Digital Elevation Model European Space Agency, 2021 European Space Agency, Sinergise (2021). Copernicus Global Digital Elevation Model. Distributed by OpenTopography. https://doi.org/10.5069/G9028PQB. Accessed: 2024-02-15 Freely available with obligation to quote the source ESA 01/01/2011 - 07/01/2015
Soilclasses from Scotland Soil Map FIPS_N A shapefile containing polygons of the soil types and landorm at a 250k:1 scale The James Hutton Institute, 2022 National Soil Map of Scotland v1.4 copyright and database right The James Hutton Institute 2024. Used with the permission of The James Hutton Institute. All rights reserved. Any public sector information contained in these data is licensed under the Open Government Licence v.2.0 Freely available with obligation to quote the source James Hutton Institute 1.4
Spatial inventory of UK waterbodies Water Lakes Taylor, 2021 Contains data supplied by UK Centre for Ecology & Hydrology licensed under the Open Government Licence v3.0. Freely available with obligation to quote the source UKCEH v3.6
OS Open Rivers Water Rivers Ordnance Survey, 2023a Contains Ordnance Survey data licensed under the Open Government Licence v3.0. Freely available with obligation to quote the source Ordnance Survey (OS) 2.3 - (04/2023)
Land Cover Map SLSRA EUNIS L2 Raster containing pixels of land cover habitat type NatureScot and Space Intelligence, 2023 Maps and data created by Space Intelligence with input and support from NatureScot , © SNH Freely available with obligation to quote the source UKCEH 2022
NNR SLSRA Shapefile containing areas designated as NNR NatureScot, 2022a Contains public sector information licensed under the Open Government Licence v3.0. Freely available for any use (public domain or CC0) Nature Scot 24/07/2023
National Parks SLSRA Shapefile containing Scotlands National Parks Stirling Council, 2024 © Scottish Government, contains Ordnance Survey data © Crown copyright and database right 2024. Freely available for any use (public domain or CC0) Scottish Govt. 20/02/2023
NR SLSRA Shapefile containing areas designated as Nature Reserves NatureScot, 2021 Contains public sector information licensed under the Open Government Licence v3.0. Freely available for any use (public domain or CC0) Nature Scot 24/07/2023
Regional Park SLSRA Shapefile containing areas designated as Regional Park NatureScot, 2022b Contains public sector information licensed under the Open Government Licence v3.0. Freely available for any use (public domain or CC0) Nature Scot 24/07/2023
RSPB Reserves SLSRA Shapefile containing RSPB reserves Royal Society for the Protection of Birds, 2024 Data reproduced with the permission of RSPB.   © Crown Copyright. Ordnance Survey licence number 100021787 (2024) Freely available with obligation to quote the source RSPB NA
SAC SLSRA Shapefile containing areas designated as Special Areas of Convervation (SAC) NatureScot, 2023a Contains public sector information licensed under the Open Government Licence v3.0. Freely available for any use (public domain or CC0) Nature Scot 29-Aug-23
SPA SLSRA Shapefile containing areas designated as SPA NatureScot, 2023b Contains public sector information licensed under the Open Government Licence v3.0. Freely available for any use (public domain or CC0) Nature Scot 24/07/2023
SSSI SLSRA Shapefile containing areas designated as SSSI NatureScot, 2024 Contains public sector information licensed under the Open Government Licence v3.0. Freely available for any use (public domain or CC0) Nature Scot 15/02/2024
SWT Reserves SLSRA Shapefile containing Scottish Wildlife Trust (SWT) reserves Scottish Wildlife Trust, 2024 Scottish Wildlife Trust Freely available with obligation to quote the source Scottish Wildlife Trusts NA
Wildland SLSRA Shapefile containing areas designated as Wildland NatureScot, 2014 Contains public sector information licensed under the Open Government Licence v3.0. Freely available for any use (public domain or CC0) Nature Scot 2014 (24/7/2024)
CountryParks SLSRA Shapefile containing areas designated as CountryParks NatureScot, 2020 Contains public sector information licensed under the Open Government Licence v3.0. Freely available for any use (public domain or CC0) Nature Scot 24/07/2023
High Nature Value (HNV) farmland SLSRA Shapefile containing HNV farmland land parcels European Environment Agency, 2023 N/A available via EEA data portal under open access (ODC-by) Freely available for any use (public domain or CC0) EEA NA
Table 6: List of data sources.

5 Reclassification tables

5.1 Landform

Landform Reclassification
Undulating foothills and valley sides with gentle and strong slopes: non-rocky Foothills
Valley sides with steep and very steep slopes: non- to moderately rocky Valley sides
Lowland with dry ridges and wet hollows with gentle and strong slopes: slightly rocky Lowlands
Undulating lowlands and uplands with gentle and strong slopes: non-rocky Undulating ground
Hills with strong slopes Hills
Ridged hills with gentle to strong slopes: very rocky Hills
Mountains with gentle and strong slopes: non- to very rocky Mountains
Undulating hills with gentle and strong slopes: moderately rocky Undulating ground
Lowlands with complex gentle and strong slopes: moderately rocky Lowlands
Rugged dissected lowlands and hills with gentle and strong slopes: very rocky Hills
Hill sides with valley sides and hills with steep and very steep slopes: moderately and very rocky Hills
Foothills with strong and steep slopes Foothills
Hills and valley sides and undulating lowlands with gentle to steep slopes Hills
Hill sides with steep and very steep slopes: moderately and very rocky Hills
Lowlands with gentle slopes: non-rocky Lowlands
Hill and valley sides with gentle and strong slopes Valley sides
Hills and valley sides with steep and very steep slopes: non-rocky Hills
Undulating uplands with complex gentle slopes: slightly to moderately rocky Uplands
Hills and valley sides with gentle to steep slopes: slightly to moderately rocky Hills
Hills and undulating lowlands with gentle and strong slopes: moderately rocky Hills
Hills with simple strong and steep slopes Hills
Hills and valley sides with gentle and strong slopes: non-rocky Hills
Mountains with gentle to very steep slopes: slightly to very rocky Mountains
Hills with gentle and strong slopes: slightly rocky Hills
Hill sides and undulating lowlands with strong and steep slopes: moderately rocky Hills
Hill sides with gentle and strong slopes: slightly and moderately rocky Hills
Hills with complex strong slopes: moderately rocky Hills
Hills with complex gentle to strong slopes: slightly to moderately rocky Hills
Drumlins and undulating lowlands with gentle and strong slopes Undulating ground
Lowlands with gentle and strong slopes: non- and slightly rocky Lowlands
Unknown Built-up Areas
Undulating lowlands and hills with gentle to steep complex slopes: slightly to moderately rocky Undulating ground
Hills and valley sides with strong slopes: non-rocky Hills
Mountains with gentle and strong slopes: slightly and moderately rocky Mountains
Hummocky moraines on valley sides Hummocks / Moraines
Hill sides with steep and very steep slopes: moderately to very rocky Hills
Hummocky moraines: often bouldery Hummocks / Moraines
Terraces with gentle slopes: non-rocky Raised beach / Terrace / Coastal flats
Undulating lowlands and hills with gentle and steep slopes Undulating ground
Undulating lowlands and uplands with gentle and strong slopes Undulating ground
Hill sides with steep slopes: non-rocky Hills
Beaches and dune slacks with minor dunes with gentle and strong slopes Beaches / Dunes
Hummocky valley and slope moraines: often bouldery Hummocks / Moraines
Hill sides with steep and very steep slopes: slightly to very rocky Hills
Undulating foothills with gentle slopes: slightly rocky Foothills
Peat Valley bottom / Peats
Hill sides with steep and very steep slopes: non-rocky Hills
Hummocky valley moraines: non-rocky Hummocks / Moraines
Valley and hill sides with gentle and strong slopes Valley sides
Valley sides and floors with gentle to steep slopes: often with landslips Valley sides
Ridged low hills with gentle to very steep slopes: very rocky Hills
Hill and valley sides with simple and complex strong slopes Valley sides
Depressions and foothills with gentle slopes Depressions
Hills and valley sides with generally concave with strong and steep slopes Hills
Mounds with terraces and undulating lowlands with gentle to steep slopes Undulating ground
Hill summits with gentle and strong slopes: slightly rocky with bouldery Hills
Hill sides with gentle and strong slopes: non-rocky Hills
Mountains with gentle and strong slopes: non- to moderately rocky Mountains
Hills with strong and steep slopes: moderately rocky with boulders Hills
Undulating lowlands with gentle slopes: slightly rocky Undulating ground
Hummocky valley and slope moraines Hummocks / Moraines
Strongly terraced hills with steep slopes: slightly rocky Hills
Hills with steep slopes: slightly to moderately rocky Hills
Raised beach terraces and mounds with minor undulations and very gentle slopes Raised beach / Terrace / Coastal flats
Hills with gentle to steep slopes: slightly rocky Hills
Hills with strong and steep slopes: non- and slightly rocky Hills
Terraced hills with gentle and strong slopes: moderately rocky Hills
Hills and valley sides with strong to very steep slopes: moderately rocky Hills
Hill sides with steep slopes: moderately rocky Hills
Hill and valley sides with steep slopes: slightly to moderately rocky Valley sides
Undulating foothills with gentle slopes: non-rocky Foothills
Terraces of fluvioglacial and riverine origin with gentle slopes Raised beach / Terrace / Coastal flats
Undulating lowlands with gentle and strong slopes: non-rocky Undulating ground
Hills with strong and steep slopes Hills
Valley sides with gentle and strong slopes Valley sides
Uplands and northern lowlands with gentle and strong slopes Uplands
Weakly terraced hill sides with steep and very steep slopes: very rocky Hills
Valley floors with terraces and mounds with with gentle and strong slopes Valley bottom / Peats
Undulating uplands with strong slopes Uplands
Hill sides with gentle to steep slopes: moderately and very rocky Hills
Hill sides with gentle to steep slopes: moderately rocky Hills
Hummocky valley moraines: often bouldery Hummocks / Moraines
Broad valley bottoms and depressions with rock knolls Valley bottom / Peats
Undulating lowlands and hills with gentle and strong slopes: slightly rocky Undulating ground
Undulating uplands with strong and gentle slopes Uplands
Hills and valley sides with strong and steep slopes: moderately rocky Hills
Mounds and ridges with gentle to steep slopes Undulating ground
Hills with complex strong slopes: slightly to moderately rocky Hills
Valley sides and undulating lowlands with gentle and strong slopes: slightly rocky Valley sides
Undulating lowlands and foothills with gentle and strong slopes with hummocky moraines: non- and slightly rocky Foothills
Depressions and foothills with gentle slopes: slightly rocky Depressions
Hills with gentle to steep slopes: very rocky with boulders Hills
Hills with gentle to steep slopes: moderately rocky Hills
Lower valley sides and floors with gentle to steep slopes: often with landslips Valley sides
Hills and valley sides with steep and strong slopes Hills
Mountain summits with strong to very steep slopes: moderately to very rocky Mountains
Hills with strong to very steep slopes: very rocky Hills
Hills and undulating uplands with strong and steep slopes Hills
Rock walls and scree: very rocky Rock Walls
Lowlands and hillsides with strong and steep slopes: slightly rocky Lowlands
Valley floors and lowland with gentle slopes Valley bottom / Peats
Dissected lowlands and hills with gentle and strong slopes: moderately rocky Lowlands
Mountain summits with gentle to steep slopes: moderately and very rocky with scree Mountains
Gentle complex hill slopes and occasional flats: moderately rocky Hills
Undulating lowlands and hills with gentle and strong slopes: non-rocky Undulating ground
Undulating lowlands with mounds and terraces with gentle slopes Undulating ground
Hills and undulating uplands with gentle and strong slopes Hills
Hills and valley sides with gentle to steep slopes Hills
Hillocky lowlands with gentle and strong slopes: slightly rocky Hills
Hill sides with steep complex slopes: moderately rocky Hills
Mounds on valley sides with strong slopes Undulating ground
Rugged hills with gentle to strong slopes: very rocky Hills
Undulating lowlands and hills with gentle and strong slopes: non- and slightly rocky Undulating ground
Gently undulating to hilly with gentle and strong slopes: non-rocky Undulating ground
Foothills with strong slopes Foothills
Hill sides with very steep slopes: moderately rocky with with cliffs Hills
Hill and valley sides with gentle to steep slopes: non-rocky Valley sides
Undulating foothills with gentle slopes Foothills
Foothill with gentle slopes Foothills
Hills and valley sides with gentle to strong slopes: non-rocky Hills
Hummocky valley and slope moraines: slightly rocky with often bouldery Hummocks / Moraines
Lowlands with gentle and strong complex slopes: slightly to moderately rocky Lowlands
Foothills with gentle and strong slopes Foothills
Undulating lowlands with gentle and strong slopes and hummocky moraines: non- to slightly rocky Undulating ground
Beaches and dunes with gentle and strong slopes Beaches / Dunes
Mountain summits with gentle and strong slopes: non- to slightly rocky Mountains
Rugged hills with gentle and strong slopes: very rocky Hills
Undulating lowlands and foothills with strong and gentle slopes Undulating ground
Lochs Fresh Water
Undulating hills with strong slopes: non-rocky Hills
Hill and valley sides with steep and very steep slopes: moderately and very rocky Valley sides
Hill and valley sides with strong and steep slopes Valley sides
Undulating lowlands and hills with strong and steep slopes: non-rocky Undulating ground
Depressions and undulating uplands with gentle slopes Depressions
Depressions and footslopes Depressions
Rugged hill sides with strong and steep slopes: very rocky Hills
Mountain or hill summits with gentle and strong slopes: non- to very rocky with bouldery Mountains
Raised beach terraces and undulating dunes (machair) with gentle and steep slopes: non-rocky Raised beach / Terrace / Coastal flats
Mountain summits with gentle to very steep slopes: very rocky with scree Mountains
Hills with simple convex strong and steep slopes Hills
Undulating uplands and depressions with gentle and strong slopes Uplands
Weakly terraced hill sides with gentle to steep slopes: moderately rocky Hills
Undulating lowlands and foothills with gentle slopes: non-rocky Undulating ground
Hills and undulating uplands with gentle to steep slopes Hills
Depressions and hill sides with gentle slopes Hills
Undulating lowlands with gentle and strong slopes and stepped hillsides with steep slopes: non- to moderately rocky Undulating ground
Dunes and slacks with wide slope range Beaches / Dunes
Undulating foothills with gentle and strong slopes Foothills
Undulating lowlands and hill sides with gentle and strong slopes: non-rocky Undulating ground
Lowland with gently sloping terraces Lowlands
Undulating lowlands with gentle and strong slopes: slightly to moderately rocky Undulating ground
Foothills and undulating lowlands with gentle and strong slopes Foothills
Hill crests and ridges with gentle slopes Hills
Hill sides with gentle and strong slopes: moderately rocky Hills
Mountain summits with gentle and strong slopes Mountains
Undulating lowlands and foothills with gentle and strong slopes: non- and slightly rocky Undulating ground
Basins and valleys Valley bottom / Peats
Undulating lowland and foothills with gentle and strong slopes: non- and slightly rocky Undulating ground
Valley floors with mounds and terraces Valley bottom / Peats
Hummocky valley moraines Hummocks / Moraines
Mounds and undulating lowlands with gentle slopes Undulating ground
Hills and valley sides with strong and steep slopes: non-rocky Hills
Undulating uplands with complex and short gentle and strong slopes Uplands
Valleys and depressions amongst hills and uplands with gentle slopes Depressions
Depressions and undulating uplands with complex gentle slopes: moderately rocky Depressions
Undulating uplands with strong simple slopes Uplands
Valley sides with concave and regular with gentle and strong slopes: non-rocky Valley sides
Undulating lowlands and foothills with gentle to steep slopes: moderately rocky Undulating ground
Undulating lowlands and hills with strong slopes Undulating ground
Immature soils Rocks / Scree
Undulating foothills with gentle and strong slopes: slightly rocky Foothills
Drumlins and complex gentle slopes: non- and moderately rocky Hummocks / Moraines
Mounds and terraces with gentle slopes Undulating ground
Undulating hill and upland with gentle and strong slopes Hills
Hills and lowlands with gentle to steep slopes: slightly and moderately rocky Hills
Rugged hills and valley sides with strong and steep slopes: very rocky with scree Hills
Raised beach terraces and mounds with gentle slopes Raised beach / Terrace / Coastal flats
Undulating lowlands and hills with gentle and strong slopes: slightly rocky to rocky Undulating ground
Undulating lowlands and foothills with strong slopes Undulating ground
Undulating uplands and hills with gentle and strong slopes: non- and slightly rocky Undulating ground
Hills with gentle to steep slopes: moderately to very rocky Hills
Ridged lowland and hills with gentle and strong slopes: slightly rocky Hills
Hills with simple convex strong slopes Hills
Valleys and depressions with concave gentle slopes Valley bottom / Peats
Hills with steep slopes: moderately and very rocky Hills
Valley sides with steep and very steep slopes: moderately and very rocky Valley sides
Undulating lowlands and valley sides with gentle and strong slopes Undulating ground
Hills with complex strong and steep slopes: non-rocky Hills
Hills with gentle to steep slopes: non- and slightly rocky Hills
Rugged hills and valley sides with steep and very steep slopes: very rocky with scree Hills
Mountains with gentle to very steep slopes: non- to very rocky Mountains
Undulating uplands with gentle slopes Uplands
Mounds and discontinuous terraces with gentle slopes Undulating ground
Hills with strong and steep slopes: slightly rocky Hills
Hills with simple convex steep and strong slopes Hills
Undulating lowlands and foothills with gentle and strong slopes: non-rocky Undulating ground
Undulating lowlands and hills with gentle to very steep slopes: non-rocky Undulating ground
Hills with complex strong and steep slopes: moderately to very rocky Hills
Lowlands with gentle slopes Lowlands
Mounds with ridges and terraces with gentle to steep slopes Undulating ground
Undulating uplands with gentle and strong slopes Uplands
Mountains with gentle and strong slopes: non-rocky Mountains
Undulating lowlands and uplands with gentle slopes: non-rocky Undulating ground
Hill and valley sides with gentle and strong slopes: non-rocky Valley sides
Mountains with gentle to very steep slopes: moderately rocky Mountains
Undulating lowlands and hills with gentle and strong slopes Undulating ground
Undulating lowlands and terraces with gentle slopes Undulating ground
Hill sides with steep slopes: slightly rocky Hills
Valley sides with gentle and strong slopes: slightly and moderately rocky Valley sides
Depressions and hill sides with gentle and strong slopes Depressions
Rugged hills with strong and steep slopes: very rocky with some scree Hills
Terraces and mounds with gentle slopes Hummocks / Moraines
Undulating foothills and uplands with gentle and strong slopes Foothills
Valley sides with strong to very steep slopes: moderately rocky Valley sides
Broad ridge with gentle and strong slopes Hills
Undulating lowlands with gentle and strong slopes: slightly rocky Undulating ground
Rugged hills with strong and steep slopes: very rocky with scree Hills
Undulating lowlands with gentle and steep slopes Undulating ground
Foothills and lowlands with gentle and strong slopes Foothills
Foothills with gentle slopes Foothills
Valley floors and lowlands with rock knolls and moundy moraines: complex gentle slopes Valley bottom / Peats
Hillsides with steep and very steep slopes: slightly to very rocky Hills
Hills and uplands with gentle slopes Hills
Hummocky valley moraines: slightly to very bouldery Hummocks / Moraines
Undulating lowlands and hills with gentle and strong slopes: moderately rocky Undulating ground
Valley sides with gentle and strong slopes: non-rocky Valley sides
Hill sides and undulating land with gentle and strong slopes: non-rocky Hills
Mountains with gentle to steep slopes: moderately and very rocky Mountains
Hillocky and weakly terraced land with gentle to strong slopes: moderately rocky Hills
Uplands with gentle slopes Uplands
Undulating lowland and lower hill slopes with gentle to strong slopes: non-rocky Undulating ground
Hills and valley sides with very steep slopes: some crags with scree and boulders: very rocky Hills
Hills and valley sides with gentle and strong slopes: non-rocky with some scree Hills
Hills with gentle and strong slopes Hills
Ridged lowland with gentle to steep slopes: very rocky Lowlands
Saltings Beaches / Dunes
Gently sloping lowlands: slightly and moderately rocky Lowlands
Drumlins and depressions with gentle and strong slopes: slightly rocky Hummocks / Moraines
Terraces with gentle slopes: slightly rocky Raised beach / Terrace / Coastal flats
Mountain summits with gentle to steep slopes: non- to very rocky with scree Mountains
Hills and valley sides with gentle to steep slopes: non-rocky Hills
Lowlands and hill sides with gentle to very steep slopes: moderately rocky Lowlands
Undulating hill sides with gentle and strong slopes: moderately rocky Hills
Raised beach terraces and undulating dunes (machair) with gentle slopes: moderately rocky Raised beach / Terrace / Coastal flats
Hills and valleys with gentle to steep slopes Hills
Hills and valley sides with frequently terraced with with gentle and strong slopes: slightly rocky Hills
Undulating lowlands with gentle slopes Undulating ground
Hills and undulating lowlands with gentle and strong slopes Hills
Undulating lowlands and hills with gentle to steep slopes: slightly rocky Undulating ground
Ridged low hills with strong slopes: moderately rocky Hills
Mountain summits with gentle and strong slopes: non- and slightly rocky Mountains
Hills and valley sides with steep and very steep slopes: very rocky Hills
Gently undulating lowlands: slightly rocky Lowlands
Hill and mountain summits with gentle and strong slopes: slightly and moderately rocky Hills
Undulating lowlands and foothills with gentle and strong slopes Undulating ground
Depressions and gentle slopes Depressions
Hill sides with strong to very steep slopes: moderately to very rocky Hills
Valley sides and undulating lowlands with gentle and strong slopes: moderately to very rocky Valley sides
Hillocky lowland with gentle and strong slopes: moderately rocky Hills
Mountain summits with strong to very steep slopes: very rocky Mountains
Undulating uplands with gentle and strong slopes: non-rocky Uplands
Hills and valley sides with very steep slopes Hills
Undulating lowlands and stepped hillsides with strong and very steep slopes: moderately rocky Undulating ground
Mounds and terraces with gentle and strong slopes Undulating ground
Hill and valley sides with strong and steep slopes: moderately rocky Valley sides
Hill sides with strong and steep slopes Hills
Valley sides with simple and complex gentle and strong slopes Valley sides
Undulating lowlands with gentle slopes: moderately rocky Undulating ground
Hill sides with steep and very steep regular slopes: moderately rocky Hills
Foothills and undulating uplands with gentle slopes Foothills
Ridged hills with gentle to strong slopes: moderately rocky Hills
Raised beach platform: moderately rocky Raised beach / Terrace / Coastal flats
Hill and valley sides with strong to very steep slopes: moderately rocky and bouldery Valley sides
Foothills and depressions with gentle slopes Foothills
Elongated depressions Depressions
Mounds and undulating lowlands with gentle and strong slopes Undulating ground
Hills and valley sides with gentle and strong slopes: slightly rocky Hills
Undulating lowlands and foothills with gentle slopes Undulating ground
Hills with strong and steep complex slopes: moderately to very rocky Hills
Rugged hills with strong and steep slopes: very rocky Hills
Hill and mountain summits with gentle to steep slopes: slightly rocky Hills
Hill sides with often with mounds with terraces and depressions: gentle slopes Hummocks / Moraines
Hummocky moraines and depressions Hummocks / Moraines
Drumlins and undulating lowlands with simple and complex gentle and strong slopes: slightly rocky Undulating ground
Undulating lowlands with gentle and strong slopes Undulating ground
Foothills and lowlands with simple and complex gentle and strong slopes Foothills
Rugged hills with steep and very steep slopes: very rocky with scree Hills
Hill and valley sides with strong to very steep slopes: moderately rocky Valley sides
Undulating lowlands with complex strong slopes: moderately rocky Undulating ground
Undulating lowlands and stepped hill sides with strong and very steep slopes: moderately rocky Undulating ground
Hills and depressions with gentle and strong slopes Hills
Hills with gentle and strong slopes: moderately rocky Hills
Stabilized dunes and undulating raised beaches Beaches / Dunes
Flood plains with river terraces and former lake beds Flood plain
Undulating foothills with gentle to steep slopes: moderately rocky Foothills
Hill summits with gentle slopes: slightly rocky with stony Hills
Mounds and hollows with gentle slopes Undulating ground
Hills and valley sides with strong and steep slopes Hills
Mountian summits with gentle and strong slopes: non-rocky Mountains
Valley floors with terraces and mounds with gentle and strong slopes Valley bottom / Peats
Drumlins with intervening simple and complex gentle slopes Hummocks / Moraines
Foothills with complex gentle and strong slopes Foothills
Undulating hills and uplands with gentle and strong slopes Hills
Hill sides with strong and very steep slopes: very rocky with bouldery Hills
Very steep rock walls and stabilised scree: very rocky Rock Walls
Mountain summits with gentle and strong slopes: non-rocky Mountains
Mountains with gentle to very steep slopes: moderately and very rocky: bouldery Mountains
Ridged lowland with gentle and strong slopes: moderately rocky Lowlands
Hill sides with gentle and strong slopes Hills
Hill and valley sides with strong to very steep slopes: slightly and moderately rocky Valley sides
Undulating lowlands with gentle slopes: non-rocky Undulating ground
Hills and lowlands with gentle to steep slopes: non-rocky Hills
Hills with gentle to strong slopes Hills
Raised beach terraces with gentle slopes Raised beach / Terrace / Coastal flats
Hummocky valley moraines with often bouldery Hummocks / Moraines
Hummocky moraines Hummocks / Moraines
Undulating lowlands and hills with gentle to steep slopes: slightly to very rocky Undulating ground
Dune slacks and undulating raised beaches Beaches / Dunes
Weakly terraced hill sides with steep and very steep slopes: moderately rocky Hills
Valley sides with gentle slopes: non-rocky Valley sides
Hills and hill sides with gentle and strong slopes: moderately rocky Hills
Hills with gentle to strong slopes: slightly rocky Hills
Hills and valley sides with strong to very steep slopes: moderately rocky with scree Hills
Moundy terraces with gentle and strong slopes Undulating ground
Undulating lowlands with gentle to strong slopes: non-rocky Undulating ground
Hill sides and hills with steep and very steep slopes: moderately and very rocky Hills
Hills and undulating lowlands with gentle and strong slopes: non-rocky Hills
Hill and valley sides with steep slopes: non-rocky Valley sides
Hills and valley sides with strong and steep slopes: slightly and moderately rocky Hills
Undulating lowlands and valley sides with gentle and strong slopes: non-rocky Undulating ground
Terraced hills with gentle and strong slopes: slightly to moderately rocky Hills
Undulating lowlands with gentle and strong slopes: moderately rocky Undulating ground
Foothills and depressions with with gentle slopes Foothills
Hills with steep and strong slopes Hills
Table 7: Reclassification of the Natural Features component (landform layers) baed on data extracted from The James Hutton Institute (2022).

5.2 Soil

Reclassification genSoilTyp
Organic Peaty podzols, Peaty gleys, Peat,
Mineral Mineral podzols, Mineral gleys, Brown soils, Montane soils, Immature soils, Alluvial soils, Calcareous soils
Table 8: Reclassification of the Natural Features component (soil layers) based on The James Hutton Institute (2022).

5.3 Rivers

Class Example search terms or process
Major river or tributary Specific e.g. “Tay”, “Spey” etc…
Tidal river or estuary taken from dataset attribute “form”
Minor river or tributary e.g. “River”, “Abhainn”, “Uisge”
Named minor stream or tributary e.g. “Allt”, “Alltan”, “Burn”, “Feadan”, “Sruth”, “Caochan”
Unnamed minor stream or tributary Waterways with no names
Canal taken from dataset attribute “form”
Lake taken from dataset attribute “form”
Table 9: Reclassification of the Water component (river layers) based on Ordnance Survey (2023a).

5.4 Lakes

Class size (ha)
Pond \(<0.2\)
Lochan \(0.2-2\)
Small loch \(2-10\)
Mediun loch \(10-100\)
Large loch \(100-500\)
Major loch \(>500\)
Table 10: Reclassification of the Water component (lakes layers) based on data from Taylor (2021).

Access to code & data

Code for the Recreational Potential model is available as an R Package hosted on GitHub at https://github.com/BioDT/uc-ces-recreation. These reports relate specifically to the version 1.0 release, which is also available on Zenodo at https://doi.org/10.5281/zenodo.15705544.

The model requires some input data that is too large to be bundled in with the package, but which can be downloaded from Dropbox after package installation using biodt.recreation::download_data(). This data is also included in the Zenodo listing.

The source for these documents and this website can be found at https://github.com/BioDT/ces-recreation-reports.

Funding

Funding for OpenNESS came from the European Community’s Seventh Framework Programme (FP7/2007-2013) under grant agreement no 308428, OpenNESS Project (Operationalisation of Natural Capital and Ecosystem Services: From Concepts to Real-world Applications.)

Funding for BioDT came from the European Union’s Horizon Europe Research and Innovation Programme under grant agreement No 101057437 (BioDT project, https://doi.org/10.3030/101057437). Views and opinions expressed are those of the author(s) only and do not necessarily reflect those of the European Union or the European Commission. Neither the European Union nor the European Commission can be held responsible for them.

Funding for SPEAK came from the Natural Environment Research Council – Growing Shoots Partnership and application co-creation bursary. NE/Y005805/1 _Growing Shoots.

CRediT statement

JMR: software; visualisation; writing. MT: software; visualisation; writing. CA: conceptualisation; data curation; funding acquisition; software. JD: conceptualisation; funding acquisition; project administration. SR: funding acquisition; software.

Acknowledgements

We wish to acknowledge significant contributions to a previous version of the Recreational Potential model by Will Bolton. JMR would also like to thank Tomáš Martinovič for assisting with understanding the requirements of BioDT.

We are very grateful to all the participants of the SPEAK project which this output is based on including: Alastair Leaver, Alice MacSporran, Brian Cassidy, Chris Pollock, David Giles, Jean Cowie, Joanna Gilliatt, Justyna Olszewska, Laura Taylor, Maddi Bunker, Rebecca MacLennan, S Mayes, Shaila Rao, Steve MacKinnon and Tom Gebbie, Active Stirling. We are equally grateful to an additional 18 participants who preferred to remain anonymous.

Correspondence

  • {jand,chan}@ceh.ac.uk for enquiries relating to the BioDT or SPEAK projects, or ongoing and future work.

  • chan@ceh.ac.uk for enquiries relating to the data sources and QGIS processing.

  • {joemar,madtig}@ceh.ac.uk for enquiries relating to the code and anything else appearing in the technical supplement and/or developer’s report.

  • simrol@ceh.ac.uk for enquiries relating to the biodiversity component of the BioDT project.

References

Dick, J., Andrews, C., Orenstein, D. E., Teff-Seker, Y., & Zulian, G. (2022). A mixed-methods approach to analyse recreational values and implications for management of protected areas: A case study of cairngorms national park, UK. Ecosystem Services, 56, 101460. https://doi.org/10.1016/j.ecoser.2022.101460
European Environment Agency. (2023). High Nature Value (HNV) farmland. 2012 accounting version, Nov. 2017. https://www.eea.europa.eu/en/datahub/datahubitem-view/1bd26e8f-8ea0-45e0-b6bf-9ed2baff5d28?activeAccordion=1070000%2C61.
European Space Agency. (2021). Copernicus Global Digital Elevation Model. https://portal.opentopography.org/raster?opentopoID=OTSDEM.032021.4326.3.
Improvement Service. (2024). Core Paths - Scotland [Web Feature Service]. https://data.spatialhub.scot/dataset/core_paths-is.
NatureScot. (2014). Wild land areas 2014. https://opendata.nature.scot/datasets/snh::wild-land-areas-2014/about; NatureScot.
NatureScot. (2020). Country parks. https://opendata.nature.scot/datasets/snh::country-parks/about; NatureScot.
NatureScot. (2021). Nature reserves. https://opendata.nature.scot/datasets/snh::nature-reserves/about; NatureScot.
NatureScot. (2022a). National nature reserves. https://opendata.nature.scot/datasets/snh::national-nature-reserves/about; NatureScot.
NatureScot. (2022b). Regional parks. https://opendata.nature.scot/datasets/snh::regional-parks/about; NatureScot.
NatureScot. (2023a). Special areas of conservation (SAC). https://opendata.nature.scot/datasets/snh::special-areas-of-conservation/about; NatureScot.
NatureScot. (2023b). Special protection areas (SPA). https://opendata.nature.scot/datasets/snh::special-protection-areas/about; NatureScot.
NatureScot. (2024). Sites of special scientific interest (SSSI). https://opendata.nature.scot/datasets/snh::sites-of-special-scientific-interest/about; NatureScot.
NatureScot and Space Intelligence. (2023). Scotland habitat and land cover map – 2022. https://spatialdata.gov.scot/geonetwork/srv/api/records/8462f345-6e9c-45de-b1d2-665a55b9d74a; NatureScot.
Office for National Statistics. (2018). NUTS, level 1 (January 2018) Boundaries UK BFC. https://geoportal.statistics.gov.uk/datasets/e9cfbf2e478a481bb367eb2f2891d504_0/about.
OpenStreetMap. (n.d.). Map of Scotland. https://www.openstreetmap.org.
Ordnance Survey. (2023a). OS Open Rivers. Version 2.3. https://osdatahub.os.uk/downloads/open/OpenRivers.
Ordnance Survey. (2023b). OS Open Roads. https://osdatahub.os.uk/downloads/open/OpenRoads.
Paracchini, M. L., Zulian, G., Kopperoinen, L., Maes, J., Schägner, J. P., Termansen, M., Zandersen, M., Perez-Soba, M., Scholefield, P. A., & Bidoglio, G. (2014). Mapping cultural ecosystem services: A framework to assess the potential for outdoor recreation across the EU. Ecological Indicators, 45, 371–385. https://doi.org/10.1016/j.ecolind.2014.04.018
Royal Society for the Protection of Birds. (2024). RSPB reserves. https://opendata-rspb.opendata.arcgis.com/datasets/6076715cb76d4c388fa38b87db7d9d24/explore; RSPB Open Data Portal.
Scottish Wildlife Trust. (2024). Scottish wildlife trust reserve boundaries (public view). https://scottish-wildlife-trust-swt.opendata.arcgis.com/datasets/SWT::scottish-wildlife-trust-reserve-boundaries-public-view/about; Scottish Wildlife Trust Open Data Portal.
Stirling Council. (2024). National parks scotland. https://opendata.scot/datasets/stirling+council-open+data+-+national+parks+scotland/; Stirling Council Open Data.
Sustrans GIS. (2024). National Cycle Network (Public). https://data-sustrans-uk.opendata.arcgis.com/datasets/Sustrans-UK::national-cycle-network-public-1/about.
Taylor, P. J. (2021). Spatial inventory of UK waterbodies. https://doi.org/10.5285/b6b92ce3-dcd7-4f0b-8e43-e937ddf1d4eb; NERC EDS Environmental Information Data Centre.
The James Hutton Institute. (2022). National Soil Map of Scotland. Version 1.4. https://spatialdata.gov.scot/geonetwork/srv/eng/catalog.search#/metadata/B7E65842-C041-4950-BF0C-3AF06C2DBAE7.

Footnotes

  1. Indeed this is true for almost every type of feature, but these proximity contributions were only computed for roads/paths and water features.↩︎

  2. The right-hand-side of Equation 1 can be read as “The set of all 2d coordinates \(x'\) for which the condition \(I_{ij}(x') = 1\) is satisfied.”↩︎

  3. The provenance of these values, which date back to the earliest version of the code, has been lost. The precision with which \(\alpha\) is given is a particular curiosity.↩︎

  4. Note that this does not exhibit the classic ‘S’ shape of logistic curves since the domain is restricted to the positive real line (distances are non-negative!).↩︎

  5. If \(F(x)\) is constant in the region of interest then the right-hand-side of Equation 9 diverges and \(\hat{F}\) is undefined.↩︎

Reuse

Citation

BibTeX citation:
@report{ukceh2025,
  author = {Marsh Rossney, Joe and Tigli, Maddalena and Andrews,
    Christopher and Dick, Jan and Rolph, Simon},
  publisher = {Zenodo},
  title = {Reports on the {BioDT} {Recreational} {Potential} {Model} for
    {Scotland}},
  version = {1.0},
  date = {2025-06},
  url = {https://doi.org/10.5281/zenodo.15715070},
  doi = {10.5281/zenodo.15715070},
  langid = {en},
  abstract = {These reports describe the BioDT Recreational Potential
    Model for Scotland (1.0), at the end of the BioDT project in June
    2025.}
}
For attribution, please cite this work as:
Marsh Rossney, J., Tigli, M., Andrews, C., Dick, J., & Rolph, S. (2025). Reports on the BioDT Recreational Potential Model for Scotland (Version 1.0). Zenodo. https://doi.org/10.5281/zenodo.15715070