Global soil organic carbon removal by water erosion under climate change and land use change during 1850–2005 AD

Abstract

15The onset and expansion of agriculture has accelerated soil erosion by rainfall and runoff substantially, mobilizing vast quantities of soil organic carbon (SOC) globally. Studies show that at timescales of decennia to millennia this mobilized SOC can significantly alter previously estimated carbon emissions from land use change (LUC).However, a full understanding of the impact of erosion on land-atmosphere carbon exchange is still missing. The 20 aim of our study is to better constrain the terrestrial carbon fluxes by developing methods compatible with Earth System Models (ESMs) in order to explicitly represent the links between soil erosion by rainfall and runoff and carbon dynamics. For this we use an emulator that represents the carbon cycle of a land surface model, in combination with the Revised Universal Soil Loss Equation model. We applied this modeling framework at the global scale to evaluate the effects of potential soil erosion (soil removal only) in the presence of other perturbations 25 of the carbon cycle: elevated atmospheric CO 2 , climate variability, and LUC. We found that over the period 1850-2005 AD acceleration of soil erosion leads to a total potential SOC removal flux of 100 Pg C of which 80% occurs on agricultural, pasture and natural grass lands. Including soil erosion in the SOC-dynamics scheme results in a doubling of the cumulative loss of SOC over 1850 -2005 due to the combined effects of climate variability, increasing atmospheric CO 2 and LUC. This additional erosional loss decreases the cumulative global carbon sink on 30 land by 5 Pg for this specific period, with the largest effects found for the tropics, where deforestation and agricultural expansion increased soil erosion rates significantly. We also show that the potential effects of soil erosion on the global SOC stocks cannot be ignored when compared to the effects of climate change or land use change on the carbon cycle. We conclude that it is necessary to include soil erosion in assessments of LUC and evaluations of the terrestrial carbon cycle. 35Biogeosciences Discuss., https://doi

Simulating Erosion‐Induced Soil and Carbon Delivery From Uplands to Rivers in a Global Land Surface Model

“…This is an advantage over complex, mechanistic erosion models (Francesconi et al, 2016; Nearing et al, 1989; Singh et al, 2006) which are only applicable for single small catchments, due to the limitations of data needed for model parameterization and the considerable computation workloads. In the past two decades, a growing number of large‐scale soil erosion‐C cycle models have been developed (Bouchoms et al, 2017; Chappell et al, 2016; Liu et al, 2003; Lugato et al, 2016; Naipal et al, 2016, 2018; Van Oost et al, 2005; Zhang et al, 2014). However, most of these models are run at a time step of 1 year or even a decade (Rosenbloom et al, 2001), and thus cannot represent the impacts of seasonal vegetation and event‐based rainfall dynamics on C erosion and river delivery.…”

“…Although several studies already estimated the gross soil erosion rates in the Rhine catchment or across Europe (Cerdan et al, 2010; Naipal et al, 2016, 2018; Panagos et al, 2015), to our knowledge, only a few studies have estimated the sediment delivery from uplands to the river network (Van Dijk & Kwaad, 1998; Asselman, 1999; Borrelli et al, 2018). In this study, the gridded product of net soil erosion/deposition rate (erosion minus deposition, in Mg ha −1 yr −1 ) from ESDAC (Table 1, Borrelli et al, 2018) was used to calibrate the model parameters a , b , c , and d in Equations , , , , and and to evaluate the simulated results.…”

“…This is an advantage over complex, mechanistic erosion models (Nearing et al, 1989;Singh et al, 2006;Francesconi et al, 2016) which are only applicable for single small catchments, due to the limitations of data needed for model parameterization and the considerable computation workloads. In the past two decades, a growing number of large-scale soil erosion -C cycle models have been developed (Liu et al, 2003;Van Oost et al, 2005;Zhang et al, 2014;Chappell et al, 2016;Lugato et al, 2016;Bouchoms et al, 2017;Naipal et al, 2016Naipal et al, , 2018.…”

“…The absence of water erosion‐related fluxes in LSMs induces systematic bias in the modeling of regional and global C budgets. The sign and magnitude of this bias is poorly constrained and may range from a 1 Pg C yr −1 source of atmospheric CO 2 to a sink of 1 Pg C yr −1 (Van Oost et al, 2012; Regnier et al, 2013; Nadeu et al, 2015; Lauerwald et al, 2017; Naipal et al, 2018). One of the main reasons why water erosion processes are not routinely included in LSMs is the high spatial resolution required to represent the effects of topography on overland flow, erosion and sediment redistribution along hillslopes and floodplains (Naipal et al, 2016) and the long time scales involved that would require too much computational resources for integrating this process in a full‐fledged LSM.…”

“…However, these studies did not represent the temporal evolution of terrestrial ecosystem fluxes as they relied on static vegetation and SOC maps (Table S1 in the supporting information). Naipal et al (2015, 2016, 2018) went one step further and introduced a scheme that couples a sediment C removal module with biomass and soil C dynamics controlled by climate, CO 2 and land use to simulate the large‐scale lateral movement of sediment and C during the last century. However, as the sediment budget module was not embedded into the terrestrial C cycle model (in this case, the ORCHIDEE LSM), this approach cannot account for the feedbacks of soil erosion on the ecosystem C cycle.…”