Issues of Uncertainty and Scale in Derived Products

Lead Research Organisation: NERC British Geological Survey
Department Name: Environmental Modelling


When deciding how land should be utilised, planners require information about the status of that land and its relative suitability for different uses. For instance, planners might wish to compare the value of land for agricultural production (which might be determined from soil properties), for resource extraction (which might be determined from mineral resource information) and for housing (which will require information about flood and subsistence risk and the suitability of the land for building). The information required to make these assessments is often provided as spatial data products in which the relevant environmental property (e.g. soil pH or flood risk rating) is estimated on the nodes of a spatial-grid which covers the relevant landscape.

These spatial data products are derived from the data and models which are available for the landscape under consideration. A soil pH spatial product might be produced by interpolation of the pH values that were measured in a soil survey of the region. Alternatively, a mathematical model that integrates information about climate, topography and drainage might be used to determine the flood risk.

This project addresses two problems in such use of spatial data products. First, the planners are unlikely to be aware of or account for the uncertainty that is associated with the product. Some uncertainty will almost inevitably arise because it is rarely possible to measure at every relevant location. Second, the spatial product describes the expected value of the environmental property for a specific spatial scale and this might not be the scale at which information is required. For instance, in a mineral resource product, the horizontal scale of the estimates might correspond to the diameter of the boreholes in which the mineral concentrations were measured. However, planners might be interested in the mineral concentrations at a spatial scale equal to the size of a quarry. These issues propagate further if a spatial data product is used as an input to a mathematical or empirical model that leads to a further data product especially if the model has been designed to relate environmental properties at a scale which is inconsistent with the data products.

In this project we will consider strategies for minimising, quantifying and communicating the uncertainty and scale related issues of spatial data products. We will relate these issues to two pertinent data sets regarding the carbon content of UK soils. We will determine how a spatial survey of such data might be cost-effectively designed to yield accurate estimates of the property of interest at different spatial scales. We will develop a statistical algorithm that can use the data that result from such a survey to produce data products at different spatial scales and explore the feasibility of making this algorithm available to end users of the data. Finally, we will consider the propagation of uncertainty when spatial data products are used to derive further products. In particular, we will quantify the uncertainty that results from using a spatial product of the radiometric properties of the soil as an input to a model of soil carbon concentrations. We will quantify this propagated uncertainty and explore the information that must be provided to users of the radiometric product if they are too are to be able to determine the degree of uncertainty that will appear in a resultant product.

This project is primarily a statistical study of the issues of scale and uncertainty in spatial data products. However, in addition to statisticians the project team includes experts in product development, data science and earth science who will provide valuable information and advice regarding about the project findings can be communicated to users of spatial data products.

Planned Impact

The proposed project will lead to information about the uncertainty in spatial data products being made available to decision makers and other users of the products. These products inform land managers about the status of the landscape and are vital in appropriate landscape decision making. The project will also lead to products being at the spatial scale which is of most relevance to a particular decision. These advances could facilitate a shift towards a more probabilistic approach to making landscape decisions that fully accounts for the potential risks and benefits of the various options under consideration. The information presented to decision makers would reflect the precision with which the landscape system is understood and not give undue precedence to 'most likely' yet highly uncertain assessments of the state of the system.

All decision makings could benefit from this greater clarity regarding the knowledge of the state of the relevant landscape system. Amongst users of BGS data products this would range from home buyers being provided with a clearer explanation of the risks associated with a property to government agencies and large corporations having a better understanding of the potential outcomes from a particular land use option. This in turn would lead to more logical and efficient allocation of land that appropriately manages the risks of negative outcomes or the costs of ignoring potential positive outcomes and therefore benefits the wider community.

The case studies in the project will directly inform projects and discussions between BGS and stakeholders such as Defra and Natural England regarding strategies for soil monitoring and peat depth mapping and could inform discussions with the Ministry of Housing Communities and Local Government regarding planning and new homes in south west England and water companies who collect their supply from peatland. The statistical methodologies will be much more widely applicable and utilised in current BGS projects reporting to Defra on parameters for the Water Framework Directive or work required by the Department for Business Energy and Industrial Strategy to assess the vulnerability of groundwater supplies to potential hydraulic fracturing activities. BGS has a strategic collaboration with the Environment Agency and again, the methodologies could be used to report upon groundwater vulnerability or groundwater floods and droughts in a more probabilistic or risk based manner. Water companies have also identified the need for such probabilistic assessments of supply within their water management plans.

The project will demonstrate to data providers, how, through appropriate survey design and analysis of the measurements that result, they can ensure that their data products are of relevance to their customers. The potential for cloud-based bespoke re-computation of the product at a specified spatial scale greatly widens the context in which such products can be successfully used.
Description Maps of environmental properties (e.g. soil type or land use) are used by academics and land managers to assess the suitability of land for different purposes or to model the environmental impact of different land management policies or practices. There is uncertainty associated with these maps because they are generally derived from relatively sparse datasets. The implications of this uncertainty was explored in the context of peat quality mapping. Surveys of peat depth were used to demonstrate how the uncertainty in mapped products varies according to the spatial-scale or resolution at which information is required. New statistical approaches were developed to efficiency quantify this multi-scale uncertainty and to explore how uncertainty in one mapped product propagates into other related products. Novel methods for designing environmental surveys were developed based on these statistical approaches to minimise the uncertainty in the resultant mapped products. This led to an approach to survey peat depth that is considerably more cost-effective than current standard practice.
Exploitation Route The approaches for efficient mapping of peat depth developed in this project were circulated amongst the DEFRA steering committee overseeing the production of a UK peat map and BGS and Natural England have discussed how these approaches should be adopted within this endeavour. More generally, the statistical methodology developed in this project can be used to assess the uncertainty of any derived map of an environmental property, to identify where more data is required to improve this map and efficiency design the required data survey. This will improve the accuracy of all subsequent modelling and land assessment exercises that use these datasets.
Sectors Agriculture, Food and Drink,Digital/Communication/Information Technologies (including Software),Environment

Description The statistical methodology developed in this project have been utilised by Natural England in their planning of a UK peat map.
First Year Of Impact 2020
Sector Environment
Description ENSEMBLE network seedcorn funding
Amount £6,000 (GBP)
Organisation Lancaster University 
Sector Academic/University
Country United Kingdom
Start 09/2019 
End 07/2020
Description Land InSight: a digital twin of the UK's soil carbon and water
Amount £79,800 (GBP)
Funding ID 2021DTUC1Fry 
Organisation Natural Environment Research Council 
Sector Public
Country United Kingdom
Start 12/2021 
End 04/2022