Abstract: In the recently finished EU-funded project Carbo-Extreme, we developed a simple probabilistic method for quantifying vulnerabilities and risks to ecosystems (http://iopscience.iop.org/1748-9326/8/1/015032). The method defines risk as expected loss due to environmental hazards, and shows how such risk can be calculated as the product of ecosystem vulnerability and hazard probability. The method was used with six different vegetation models to estimate current and future drought risks for crops, grasslands and forests across Europe (http://www.biogeosciences.net/11/6357/2014/bg-11-6357-2014.html).In the still ongoing UK-funded project GREENHOUSE, the focus is on spatial upscaling of local measurements and model predictions of greenhouse gas emissions to wider regions. As part of this work, we are comparing different model upscaling methods – ranging from naive input aggregation to geostatistics – and quantify the uncertainties associated with the upscaling. This work builds on an earlier inventory of model upscaling methods that was produced in a collaboration of CEH-Edinburgh and the University of Bonn (https://www.stat.aau.at/Tagungen/statgis/2009/StatGIS2009Van%20Oijen1.pdf). Here we show a comparison of the methods using model predictions for the border region of England and Scotland. No Label

Abstract: In the framework of AgMIP (Agricultural Model Intercomparison Project; www.agmip.org), several articles have been published in which about 10 leading, agro-economic models analysed the impact of climate change on agricultural yields, area, consumption and food prices (Lotze-Campen et al. 2014, Nelson et. al 2014a,b Schmitz et al. 2014). A part of these articles are available freely through the publisher (e.g. http://www.pnas.org/content/111/9/3274). PIK has not only contributed through model simulations with the spatially explicit, agro-economic model MAgPIE, but also by coordinating this activity. Starting with AgMIP phase II in 2015, AgMIP has now for the first time conducted the model-analysis for different “Shared Socio-economic Pathways” (short SSPs). A first study has been published in the renowned journal “Environmental Research Letters” (Wiebe et al. 2015). These are important contributions to task 2.3 which aimed at simulating the impact of global climate changes on agricultural systems.Another study which is under revision in the journal PNAS, investigates the impact of climate change on agricultural welfare. The results of this paper are based on simulations with 20 different General Circulation Models (GCMs). This provides the opportunity to understand the uncertainty inherent in the different climate models better and improves the credibility of results.All mentioned articles and results are based on harmonized yield changes, which are a result of multi-model simulations, conducted in the framework of ISI-MIP (Inter-Sectoral Impact Model Intercomparison Project) and coordinated at PIK. These model results are publicly available (www.isi-mip.org) and part of an open source strategy of the institute. The modelling group around the agro-economic model MAgPIE (Model of Agriculture and its Impact on the Environment) currently discusses an open source strategy for publishing the model code. As a first step, a detailed description of the model will be available shortly (http://redmine.pik-potsdam.de/projects/magpie/wiki).PIK and the modelling group around MAgPIE have also contributed to the geoportal GLUES (Global Assessment of Land Use Dynamics, Greenhouse Gas Emissions and Ecosystem Services) where project partners can publish and share global and regional data sets as well as model results on scenarios of land use, climate change and economic development. MAgPIE results on landuse change, emissions and deforestation for different socio-economic scenarios have been made available there (http://catalog-glues.ufz.de/terraCatalog/Start.do;jsessionid=80F6A3D2C446674B898881D0589887E4). No Label

Abstract: In regionalization studies the spatial resolution of driving data is often restricted by data availability or limited computational capacity. Method and level of spatial driver aggregation in upscaling studies are sources of uncertainty and might bias aggregated model results. The suitability of upscaled model results using aggregated driving data depends on both the sensitivity of the model to these model drivers and the scale of interest to which the model output will be aggregated. An important component of soil plant atmosphere systems is the soil organic matter content influencing GHG emissions and the soil fertility of croplands.The implications of driver aggregation schemes on different system properties of croplands have been examined in a scaling exercise within the joint research project MACSUR. In this study, meteorological driving data and data on soil properties on several aggregation levels have been used to calculate the organic carbon change of cropland soils of North Rhine-Westphalia with an ensemble of biogeochemical models.The results of this scaling exercise show that the aggregation of meteorological data has little impact on modeled soil organic carbon changes. However, model uncertainty increases slightly with decreasing scale of interest from NUTS 2 level to smaller grid cell size. Conversely, the aggregation of soil properties resulted in high uncertainty ranges constraining the predictable scale of interest for all models. The study gives an indication on adequate spatial aggregation schemes in dependence on the scope of regionalization studies addressing soil organic carbon changes. No Label