Large-scale regionalization of water table depth in peatlands optimized for greenhouse gas emission upscaling

摘要

Fluxes of the three main greenhouse gases (GHG) CO₂, CH₄ andN₂O from peat and other soils with high organic carbon contents are strongly controlled by watertable depth. Information about the spatial distribution of water level isthus a crucial input parameter when upscaling GHG emissions to large scales.Here, we investigate the potential of statistical modeling for theregionalization of water levels in organic soils when data covers only asmall fraction of the peatlands of the final map. Our study area is Germany.Phreatic water level data from 53 peatlands in Germany were compiled in anew data set comprising 1094 dip wells and 7155 years of data. For each dipwell, numerous possible predictor variables were determined using nationallyavailable data sources, which included information about land cover, ditchnetwork, protected areas, topography, peatland characteristics and climaticboundary conditions. We applied boosted regression trees to identifydependencies between predictor variables and dip-well-specific long-termannual mean water level (WL) as well as a transformed form (WL_t).The latter was obtained by assuming a hypothetical GHG transfer function andis linearly related to GHG emissions. Our results demonstrate that modelcalibration on WL_t is superior. It increases the explained variance ofthe water level in the sensitive range for GHG emissions and avoids modelbias in subsequent GHG upscaling. The final model explained 45% ofWL_t variance and was built on nine predictor variables that are basedon information about land cover, peatland characteristics, drainage network,topography and climatic boundary conditions. Their individual effects onWL_t and the observed parameter interactions provide insight intonatural and anthropogenic boundary conditions that control water levels inorganic soils. Our study also demonstrates that a large fraction of theobserved WL_t variance cannot be explained by nationally availablepredictor variables and that predictors with stronger WL_t indication,relying, for example, on detailed water management maps and remote sensing products,are needed to substantially improve model predictive performance.