Carbon sequestration in soils of SW-Germany as affected by agricultural management-Calibration of the EPIC model for regional simulations

Global emissions trading allows for agricultural measures to be accounted for the carbon sequestration in soils. The Environmental Policy Integrated Climate (EPIC) model was tested for central European site conditions by means of agricultural extensification scenarios. Results of soil and management analyses of different management systems (cultivation with mouldboard plough, reduced tillage, and grassland/fallow establishment) on 13 representative sites in the German State Baden-Murttemberg were used to calibrate the EPIC model. Calibration results were compared to those of the Intergovernmental Panel on Climate Change (IPCC) prognosis tool. The first calibration step included adjustments in (a) N depositions, (b) N-2-fixation by bacteria during fallow, and (c) nutrient content of organic fertilisers according to regional values. The mixing efficiency of implements used for reduced tillage and four crop parameters were adapted to site conditions as a second step of the iterative calibration process, which should optimise the agreement between measured and simulated humus changes. Thus, general rules were obtained for the calibration of EPIC for different criteria and regions. EPIC simulated an average increase of +0.341 Mghumus-Cha(-1) a(-1) for on average 11.3 years of reduced tillage compared to land cultivated with mouldboard plough during the same time scale. Field measurements revealed an average increase of +0.343 MgC ha(-1) a(-1) and the IPCC prognosis tool +0.345 MgCha(-1) a(-1). EPIC simulated an average increase of +1.253 MgC ha(-1) a(-1) for on average 10.6 years of grassland/fallow establishment compared to an average increase of +1.342 Mg humus-Cha(-1) a(-1) measured by field measurements and +1.254MgCha(-1) a(-1) according to the IPCC prognosis tool. The comparison of simulated and measured humus C stocks was r(2) > 0.825 for all treatments. However, on some sites deviations between simulated and measured results were considerable. The result for the simulation of yields was similar. in 49% of the cases the simulated yields differed from the surveyed ones by more than 20%. Some explanations could be found by qualitative cause analyses. Yet, for quantitative analyses the available information from farmers was not sufficient. Altogether EPIC is able to represent the expected changes by reduced tillage or grassland/fallow establishment acceptably under central European site conditions of south-western Germany. (C) 2008 Elsevier B.V All rights reserved.