Methods for Modeling Soil Organic Carbon Distribution at a Landscape Scale: a Case Study on Dartmoor, Southwest England

Blanket peat and organic ‘moorland’ soils cover large parts of upland Britain and are the dominant component of the soil organic carbon (SOC) resource in the United Kingdom (Milne and Brown, 1997). To manage soil SOC effectively, a landscape scale understanding of peatland (SOC) distribution is needed to provide policy makers and land managers with information on the location and vulnerability of soil carbon within individual land holdings. However, current understanding of the quantity and distribution of SOC throughout British peatlands is limited. The national, low resolution SOC studies available (e.g. Milne and Brown, 1997), do not provide the accuracy (Garnett et al, 2000), nor the resolution required. This study responds to the need for improved estimates of SOC distribution by providing a high resolution, landscape scale, carbon inventory for an area of peatland in the southwest of England. It uses an easily replicable methodology based on geographical information systems (GIS) and field-based sampling, which is applicable to similar peatlands in the UK and perhaps elsewhere. Spatial interpolation, using multinomial logistic regression in ArcGIS 9.1 was used to estimate the mass and distribution of SOC within Dartmoor National Park. Using a step-wise approach, relationships between the topographic, hydrological and ecological controls on SOC distribution were modelled. The sampling strategy enabled statistical relationships to be drawn between peat depth, bulk density and carbon content data obtained in the field, and previously available digitised, soil, vegetation and Digital Elevation Model (DEM) datasets. Multinomial logistic regression also allows reduction in the uncertainty caused by the conceptualised discrete datasets, used within the model. This multivariate approach works well for modelling the SOC distribution on Dartmoor’s peatland and shallow organic soils and has considerable potential to work in other similar peatland and moorland environments. The data required for the model are commonly available and easily obtainable, giving this methodology potential for broadening the understanding of less studied areas of peatland in the UK. Ultimately the model could be used to: increase the accuracy of the national soil carbon inventory, provide better baseline data for scientists studying SOC change, and provide a valuable decision making tool for land managers