Geoderma (Jul 2024)
An adaptive mapping framework for the management of peat soils: A new Irish peat soils map
Abstract
Accurate mapping for effective management of peat soils is necessary to help reduce GHG emissions and improve environmental quality. However, mapping remains a major challenge as definitions of peat soils vary substantially between jurisdictions and organizations, while field data are sparse and difficult to produce, and remote sensing of limited use for converted peatlands. Using an Adaptive Mapping Framework, this paper compiles disparate data sources to update the Derived Irish Peat Map to better facilitate its application for environmental issues. This includes incorporation of areas considered to be underlain by shallow peat soils (≥ 10 cm and ≥ 8.6 % Organic Matter content), and augmentation of the overall test dataset with an additional ∼ 20,000 points.The workflow for map generation employed 20 Decision Tree Output Maps (DTOMs), aggregated into 33 Map Combinations (MCs). These facilitated the addition of new areas and removal of areas where the presence of peat could not be confidently identified. The MC selected for the update had the highest accuracy metrics (≥ 74 %), consisting of DTOMs with a user accuracy ≥ 66 % and assessed over a minimum number of test points ≥ 50. The resulting map reveals peat to underlie 1.66 M ha of Ireland (∼ 23.3 % of the country), identified with values of 83 % for overall accuracy and F1 score for peat areas. This extent is 13.2 % larger than that delineated in previous versions and at least 23.2 % larger than areas presented in other previous maps. The methodology also allows transparency from which data sources can be distinguished to define different key peat thickness ranges (≥ 10 cm, ≥ 30–40 cm), better facilitating assessment of environmental impacts arising from land use change.This approach has potential relevance for peat mapping globally, notably in areas containing disparate datasets (e.g., land cover, soil map, etc.), or employing different production methods. The accuracy metrics generated also suggest that the approach can be used as a basis for implementing or updating national and international regulations concerning carbon-rich soils in comparable settings to those encountered in Ireland.
