Estimations of peatland carbon stocks often use generalised values for peat thickness and carbon content. Ground penetrating radar (GPR), a rapid technique for field data collection, has been increasingly demonstrated as an appropriate method of mapping peat thickness. Light Detection and Ranging (LiDAR) data as a method for understanding peatland surface elevation are also becoming more widely available. Reliable mapping and quantification of site-specific carbon stocks (e.g. upland raised bogs) is therefore, becoming increasingly feasible, providing a valuable contribution to regional, national and potentially global carbon stock assessments. This is particularly important because raised bogs, such as those found in South Wales are considerable carbon stores. They are, however, susceptible to climate warming owing to their southerly location within the UK. Accurate estimates of peatland carbon stocks has broader importance because world-wide peatland carbon stores are significant and threatened by climate change, posing a substantial challenge not only due to climate feedbacks if this stored carbon is released into the atmosphere, but also the impact on the other ecosystem services that they provide. Here, we assess the value of an integrated GPR, LiDAR and Geographic Information System (GIS) approach to improve estimation of regional carbon stocks. We apply the approach to three ombrotrophic raised bogs in South Wales, UK, selected for their conservation value and their topographically-confined raised bog form. GPR and LiDAR are found to be well suited, respectively, to mapping peat thickness at bog scale and surface elevation, thus allowing surface and basal topographies to be evaluated using GIS. In turn, this allows peat volumes to be estimated. For the first time, we record values between 55,200 m3 and 163,000 m3 for the sites considered here. The greater confidence in these peat volume estimates results from the ability to calibrate the GPR velocity using a depth-to-target calibration with peat cores extracted at locations encompassing the deepest bog area. Peat thickness is mapped at the bog scale with near centimetre precision, improving the robustness of subsequent volume calculations and our understanding of the contribution of these small but numerous sites to regional carbon stocks. Our evaluation shows that GPR corresponds well with conventional manual probing but is minimally invasive and therefore less disturbing of sensitive peatland sites, while also offering improved coverage and spatial resolution with less time and cost. In combination with measured bulk density and organic carbon contents, these peat volumes allow carbon stocks to be estimated with greater confidence compared to conventional approaches, having values between 2181 ± 122 tonnes carbon and 6305 ± 351 tonnes carbon at our three sites.

A method to estimate peat depth and extent is vital for accurate estimation of carbon stocks and to facilitate appropriate peatland management. Current methods for direct measurement (e.g. ground penetrating radar, probing) are labour intensive making them unfeasible for capturing spatial information at landscape extents. Attempts to model peat depths using remotely sensed data such as elevation and slope have shown promise but assume a functional relationship between current conditions and gradually accrued peat depth. Herein we combine LiDAR-derived metrics known to influence peat accumulation (elevation, slope, topographic wetness index (TWI)) with passive gamma-ray spectrometric survey data, shown to correlate with peat occurrence to develop a novel peat depth model for Dartmoor.
Total air absorbed dose rates of Thorium, Uranium and Potassium were calculated, referred to as radiometric dose. Relationships between peat depth, radiometric dose, elevation, slope and TWI were trained using 1334 peat depth measurements, a further 445 measurements were used for testing. All variables showed significant relationships with peat depth. Linear stepwise regression of natural log-transformed variables indicated that a radiometric dose and slope model had an r2 = 0.72/0.73 and RMSE 0.31/0.31 m for training/testing respectively. This model estimated an area of 158 ±101 km2 of peaty soil >0.4 m deep across the study area. Much of this area (60 km2) is overlain by grassland and therefore may have been missed if vegetation cover was used to map peat extent. Using published bulk density and carbon content values we estimated 13.1 Mt C (8.1-21.9 Mt C) are stored in the peaty soils within the study area. This is an increase on previous estimates due to greater modelled peat depth. The combined use of airborne gamma-ray spectrometric survey and LiDAR data provide a novel, practical and repeatable means to estimate peat depth with no a priori knowledge, at an appropriate resolution (10 m) and extent (406 km2) to facilitate management of entire peatland complexes.