Section 4 Data Sources
The UK is relatively rich in data sets that provide information on different categories of land-cover and land management over time. However, these data sets are collected for different purposes (e.g. agricultural statistics, forest inventory or urban development statistics), may have different classification definitions and spatial resolution and rarely cover the entire time span (1990-2020) or region (all countries of the UK) of interest. In the Bayesian methodology, all observations have an associated uncertainty, and it is straightforward to include multiple data sources of differing quality (i.e. observations with different uncertainties). In the previous work, eight different data sources with these varied characteristics, with three different data structures, were combined in our posterior estimate of land use and land-use change. It follows from Bayes Theorem that the impact of new data on the posterior predictions of land use will depend on their associated uncertainty, and how much they differ from other observations and prior expectations. We prioritise publicly-available data sets, because they provide transparency, ease-of-access and a higher likelihood of continuing revision and availability.
Given our starting point of having a working method for Scotland, we need to extend this to the rest of the UK. The first task here is assessing the available data and the addition of further data sources into the DA procedure. This requires updating existing data sets, and adding sources which give additional data for England, Wales and Northern Ireland. The main data sets of relevance are shown in Table 4.1.
The first stage is to review these data sets for their potential, in terms of information content, consistency, and ease of use, and prioritise work to include them on this basis. The next stage is to add these new data sets, which require a variable degree of data processing - translating other land-use classifications into the LULUCF classification, transforming raw data into one or more of the data structures we require as necessary, placing the data in a common spatial framework (the OSGB grid), and calculating time series and/or rasterising as required. Each of the data sets come with a native classification for land use or cover which require integration. To do this, the information in these data sets was assigned to the appropriate LULUCF categories using a systematic approach. The available descriptions and metadata for the dataset classes were compared to the UK LULUCF definitions and the rationale for assigning each data class will be recorded. This process was carried out with discussion across the project team, so that ambiguous cross-classifications were considered as a group and agreed. The most awkward cases are where classes are added / removed or definitions change over the time span of the dataset, which introduces discontinuities in the data, and methods for handling this are documented.
| Dataset name | Spatial coverage | Time period | LULUCF category |
|---|---|---|---|
| Land Cover Map | UK | 1990, 2015, 2017-2019 (annual) | Forest Land, Cropland, Grassland, Wetlands, Settlement, Other Land |
| Countryside Survey | UK | 1990, 1998, 2007 | Cropland, Grassland, Settlement, Other Land |
| Northern Ireland Countryside Survey | NI | 2000, 2007 | Cropland, Grassland, Forest Land, Settlement, Other Land |
| National Forest Estate Sub-compartments | GB | Historic-2019 | Forest Land |
| National Forest Inventory | GB | Historic-2019 | Forest Land, Forest land converted to other land use categories |
| Forestry Statistics | UK | Historic-2019 | Forest Land |
| UK June Agricultural Census | England, Scotland, Wales, Northern Ireland | Historic-2019 | Cropland, Grassland |
| EDINA | GB | ||
| Monitoring Landscape Change | England, Wales | 1947, 1969, 1980 | Forest Land, Cropland, Grassland, Wetlands, Settlement, Other Land |
| National Countryside Monitoring Scheme | Scotland | 1947, 1969, 1980 | Forest Land, Cropland, Grassland, Wetlands, Settlement, Other Land |
| CORINE Land Cover | UK | 1990, 2000, 2006, 2012,2018 | Forest Land, Cropland, Grassland, Wetlands, Settlement, Other Land |
| UKCEH Land Cover plus crops | GB | Annual 2015-2019 | Cropland, Grassland |
| MHCLG land-use change statistics | England | Annual 2013-2018 | Settlement |
| Integrated Administration and Control System (IACS) | England, Scotland, Wales | 2014 - onwards | Cropland, Grassland |
| Holdings-level agricultural data | England, Scotland, Wales, Northern Ireland | 2000 - onwards | Cropland, Grassland |
| Northern Ireland Forest Service Sub-compartment Boundaries | NI | 2020 | Forest Land |
| Northern Ireland Woodland Base Map | NI | 2020 | Forest Land |
| Northern Ireland Agricultural Census - 5km Grid | NI | 1997-2016 | Cropland, Grassland, Rough Grazing, Farm woodland |
The specific processing and issues with each data source are described in the following sections.