Effect of climate data on simulated carbon and nitrogen balances for Europe

Abstract

In this study, we systematically assess the spatial variability in carbon and nitrogen balance simulations related to the choice of global circulation models (GCMs), representative concentration pathways (RCPs), spatial resolutions, and the downscaling methods used as calculated with LPJ‐GUESS. We employed a complete factorial design and performed 24 simulations for Europe with different climate input data sets and different combinations of these four factors. Our results reveal that the variability in simulated output in Europe is moderate with 35.6%–93.5% of the total variability being common among all combinations of factors. The spatial resolution is the most important factor among the examined factors, explaining 1.5%–10.7% of the total variability followed by GCMs (0.3%–7.6%), RCPs (0%–6.3%), and downscaling methods (0.1%–4.6%). The higher‐order interactions effect that captures nonlinear relations between the factors and random effects is pronounced and accounts for 1.6%–45.8% to the total variability. The most distinct hot spots of variability include the mountain ranges in North Scandinavia and the Alps, and the Iberian Peninsula. Based on our findings, we advise to conduct the application of models such as LPJ‐GUESS at a reasonably high spatial resolution which is supported by the model structure. There is no notable gain in simulations of ecosystem carbon and nitrogen stocks and fluxes from using regionally downscaled climate in preference to bias‐corrected, bilinearly interpolated CMIP5 projections.

Predicted Land Carbon Dynamics Are Strongly Dependent on the Numerical Coupling of Nitrogen Mobilizing and Immobilizing Processes: A Demonstration with the E3SM Land Model

“…These uncertain responses were reported in several model intercomparison projects, such as phase 5 of the Coupled Model Intercomparison Project (CMIP5; Hoffman et al 2014;Shao et al 2013;Taylor et al 2012), the North American Carbon Program Multiscale Synthesis and Terrestrial Model Intercomparison Project (MsTMIP; Huntzinger et al 2012Huntzinger et al , 2017, and the Trends and Drivers of the Regional-Scale Sources and Sinks of Carbon Dioxide project (TRENDY; http://dgvm.ceh.ac.uk/node/9). These large modeling differences were attributed to several factors, including uncertain input data, uncertain model structures, and uncertain model parameterizations (e.g., Blanke et al 2016;Clein et al 2007;Tang and Zhuang 2008;Luo et al 2015Luo et al , 2017Wieder et al 2015a,b). Further, ESMs have been criticized for ignoring nutrient controls on the terrestrial carbon cycle (e.g., Wieder et al 2015b).…”

Reply to Dr.Kerkweg

“…The predictive power of existing land biogeochemical models is diminished by uncertainties from structural design, numerical implementation, model parameterization, initial conditions, and forcing data (Tang and Zhuang, 2008;Tang et al, 2010;Luo et al, 2015;Wieder et al, 2015a;Blanke et al, 2016;Tang and Riley, 2016). Among these, developing better model structures and mathematical formulations have been identified as priorities.…”

Large uncertainty in ecosystem carbon dynamics resulting from ambiguous numerical coupling of carbon and nitrogen biogeochemistry: A demonstration with the ACME land model

“…Intercomparison Project Phase 5 (CMIP5) showed very large differences among those models' predictions (e.g., Arora et al, 2013;Friedlingstein et al, 2014;Shao et al, 2013;Koven et al 2015a). Such differences are often attributed to the four types of uncertainties, including structural (Tang et al, 2010;Wieder et al, 2015a), numerical (Yeh and Tripathi, 1989), parameterization (Tang and Zhuang, 2008;Luo et al, 2015), and forcing data (Clein et al, 2007;Blanke et al, 2016), which 5 are, respectively, loosely related to the four stages of BGC model design: (I) conceptualizing the relevant mechanisms and translating them into governing equations; (II) numerical encoding of the governing equations; (III) process module calibration and parameterization; and (IV) model analyses and applications. There have been numerous examples of how one could quantify and reduce these uncertainties (e.g., Tang andZhuang, 2008, 2009;Williams et al, 2009;Lichstein et al, 2014;Wei et al, 2014;Shi et al, 2015).…”