Peat soil thickness and carbon storage in the Belgian High Fens: insights from multi-sensor UAV remote sensing

Peatlands are known to store a large amount of carbon, but global warming and associated changes in hydrology have the potential to accelerate peatland carbon emissions. An in-depth understanding of carbon dynamics within these peatlands is therefore important. However, peatlands are complex ecosystems, and acquiring accurate and reliable estimates of how much carbon is stored underneath the Earth’s surface is inherently challenging even at small scales. Here, Unmanned Aerial Vehicles (UAVs) equipped with RGB, multispectral, thermal infrared, and LiDAR sensors were combined with Ground Penetrating Radar (GPR) technology and traditional field surveys, to provide a comprehensive 4D monitoring of a peatland landscape in the Belgian High Fens. Data was collected along a hillslope-floodplain transition. We aimed to establish links between the above- and below-ground factors that control soil carbon status, identify the key drivers of carbon storage as well as explore the potential of UAV remote sensing for spatial mapping of peat depth and carbon stock. Our results indicated that peat thickness widely varied (0.2 to 2.1 m) at small scales and is negatively correlated with elevation (r= -0.39, p<0.001). We found that soil organic carbon (SOC) stock is spatially organized, as abundant carbon was observed at the summit and shoulder of the hill, with an average storage of 670.93 ± 108.86 t/ha and 601.47 ± 133.40 t/ha, respectively. Moreover, the carbon storage exhibited heterogeneity under different vegetation types, with trees having the highest mean SOC stocks at 722.21 ± 37.92 t/ha. Through multiple linear regression, we identified 6 environmental variables that can explain 71.44% of SOC stock variance. Clay content is the most critical factor, accounting for nearly 40% of the variance, followed by topography. Contributions from land surface temperature and vegetation remain below 10%. In addition, UAV data provided accurate estimations of both peat depth and SOC stock, with RMSE and R² values of 0.13 m and 0.88 for the peat depth test dataset, and 114.42 t/ha and 0.84 for the SOC stock. Our study bridged the gap between surface observations and the hidden carbon reservoir below, this not only allows us to improve our ability to assess the spatial distribution of C stocks but also contributes to our understanding of the drivers of C turnover in these highly heterogeneous landscapes, providing insights for environmental science and climate projections.