Inferring peat thickness in Indonesian Peatlands using Airborne Electromagnetic methods

Abstract:

Peatlands are among the most carbon-rich environments of Earth, representing about 25% of the world's soil carbon and exceeding the carbon pool of global vegetation. These estimates are however affected by large uncertainties due to our inability of assessing the peat volume at the regional to the global scale. This is particularly true for peat deposits located in the tropics (like Indonesia) due to their limited accessibility. Furthermore, even though a number of studies have shown how remote sensing can be successfully used to detect the extension of peatlands, its inability for detecting their thickness prevents a precise quantification of their volume and therefore the carbon stored in these environments. In this study, we use Airborne Electromagnetics (AEM) collected over a large peatland in Indonesia, to determine both the topography (through laser altimeters) and the peat bottom surface, i.e. the separation surface between peat and the mineral substrate. Like expected for most peatland ecosystems, peat soils are more resistive than the clay substrate, making peat a suitable AEM target. The challenge, however, lies in the vertical resolution achievable by the method in order to properly capture the depth of the peat-clay interface. Results of the AEM data inversion were validated using peat thickness measurements from coring at 63 locations across the study site. In this study, we show how AEM has the ability to image the bottom morphology of peatlands, providing high accuracy (mean error smaller than 0.4m and standard deviation of 2m) in the estimation of the peat thickness. Results of predicted peat thickness obtained through the AEM methodology were compared to those obtained using a simple linear correlation between peat thickness and soil elevation (DTM) found from the analysis of the data collected at the coring locations. We show that the linear model has the merit of indicating a general simplified model of the morphology of peatlands. However, our results show how the linear model may lead to an oversimplification of the system and produce inaccurate results and large errors in the assessment of peat volume. This work has implications for demonstrating the potential of AEM to rapidly infer peat thickness at regional scales.