Where is the carbon going? - A multi-model comparison

<p>Net terrestrial carbon uptake is primarily driven by increases in net primary productivity (NPP) and/or the residence time of carbon in vegetation and soil. As such, it is of critical importance to accurately quantify spatio-temporal variation in both terms and determine their drivers. Both NPP and residence times are modulated by changing environmental conditions, including climate change and variability, atmospheric CO<sub>2</sub>, and Land Use and Land Cover Changes (LULCC). For the historical period, 1901-2019, outputs from a suite of Dynamic Global Vegetation Models (DGVMs) from the TRENDY consortium, driven with observed changes in climate, CO<sub>2</sub>, and LULCC are analysed. Changes in global and regional carbon fluxes, stocks, and residence times are quantified, as well as an attribution to the underlying drivers. We find that over the historical period the majority of models simulate an increase in NPP, predominantly driven by enhanced atmospheric CO<sub>2</sub> concentrations. This generally leads to increased carbon storage in both vegetation and soils, however there is no agreement across models on the partitioning between vegetation and soils. This increased storage also acts to reduce soil carbon residence times due to a relative increase in carbon allocated in the faster decomposing soil pools. LULCC over this period has acted to reduce carbon inputs to the system and reduce vegetation carbon residence times due to conversion of forests to shorter vegetation. We find there is a large variation in simulated global and regional fluxes, stocks, and residence times in resonse to changes in climate, implying there are considerable uncertainties in current DGVMs. We therefore use long-term global observations of productivity and biomass change to constrain model estimates and provide insight into a process attribution for biospheric change as well as highlighting areas for future model improvement.</p>

Advancing representation of anthropogenic fire in dynamic global vegetation models

<div> <p>It is now commonly-understood that improved understanding of global fire regimes demands better representation of anthropogenic fire in dynamic global vegetation models (DGVMs). However, currently there is no clear agreement on how human activity should be incorporated into fire-enabled DGVMs and existing models exhibit large differences in the sensitivities of socio-economic variables. Furthermore, existing approaches are limited to empirical statistical relations between fire regime variables and globally available socio-economic indicators such as population density or GDP. Although there has been some limited representation in global models of the contrasting ways in which different classes of actors use or manage fires, we argue that fruitful progress in advancing representation of anthropogenic fire in DGVMs will come by building on agent-based modelling approaches. Here, we report on our progress developing a global agent-based representation of anthropogenic fire and its coupling with the JULES-INFERNO fire-enabled DGVM.  </p> </div><div> <p>Our modelling of anthropogenic fire adopts an approach that classifies ‘agent functional types’ (AFTs) to represent human fire activity based on land use/cover and Stephen Pyne’s fire development stages. For example, the ‘swidden’ AFT represents shifting cultivation farmers managing cropland and secondary vegetation in a pre-industrial development setting. This approach is based on the assumption that anthropogenic fire use and management is primarily a function of land use but influenced by socio-economic context, leading different AFTs to produce qualitatively distinct fire regimes. The literature empirically supports this assumption, however data on human fire interactions are fragmented across many academic fields (including anthropology, geography, land economics). Therefore, we developed a Database of Anthropogenic Fire Impacts (DAFI) containing 1798 case studies of fire use/management from 519 publications, covering more than 100 countries and all major biomes (except Arctic/Antarctic). We discuss DAFI development, patterns in the resulting data, and possible applications. Specifically, DAFI is used with ancillary data (e.g. biophysical, socio-economic indicators), classification and regression methods to test and refine our initial AFT classification, characterise AFT fire variables, and distribute AFTs spatially. Our model will then simulate AFT distributions for alternative scenarios of change (e.g. specified by the Shared Socioeconomic Pathways). </p> </div><div> <p>Coupling distinct models can be achieved in a variety of ways, but broadly we can distinguish between ‘loose’ coupling in which information flow is uni-directional, and ‘tight’ coupling in which information flows are integrated with feedbacks and dynamic updating. Our intention is to tightly couple our AFT model with JULES-INFERNO, such that fire use and suppression behaviours from the former influence simulated fire ignitions and burned area in the latter. Reciprocally, total burned area simulated by JULES-INFERNO will feedback to influence spatial distribution of AFTs in the next time step, modifying anthropogenic fire patterns for the next step of DGVM simulation. We discuss the potential for this tight model coupling to capture socio-ecological feedbacks in fire regimes, as well as  possible pitfalls and steps needed to test and verify model outputs. These are early steps in an important journey to improve representation of anthropogenic fire in DGVMs.</p> </div>

Improving the representation of high-latitude vegetation distribution in dynamic global vegetation models

Vegetation is an important component in global ecosystems, affecting the physical, hydrological and biogeochemical properties of the land surface. Accordingly, the way vegetation is parameterized strongly influences predictions of future climate by Earth system models. To capture future spatial and temporal changes in vegetation cover and its feedbacks to the climate system, dynamic global vegetation models (DGVMs) are included as important components of land surface models. Variation in the predicted vegetation cover from DGVMs therefore has large impacts on modelled radiative and non-radiative properties, especially over high-latitude regions. DGVMs are mostly evaluated by remotely sensed products and less often by other vegetation products or by in situ field observations. In this study, we evaluate the performance of three methods for spatial representation of present-day vegetation cover with respect to prediction of plant functional type (PFT) profiles – one based upon distribution models (DMs), one that uses a remote sensing (RS) dataset and a DGVM (CLM4.5BGCDV; Community Land Model 4.5 Bio-Geo-Chemical cycles and Dynamical Vegetation). While DGVMs predict PFT profiles based on physiological and ecological processes, a DM relies on statistical correlations between a set of predictors and the modelled target, and the RS dataset is based on classification of spectral reflectance patterns of satellite images. PFT profiles obtained from an independently collected field-based vegetation dataset from Norway were used for the evaluation. We found that RS-based PFT profiles matched the reference dataset best, closely followed by DM, whereas predictions from DGVMs often deviated strongly from the reference. DGVM predictions overestimated the area covered by boreal needleleaf evergreen trees and bare ground at the expense of boreal broadleaf deciduous trees and shrubs. Based on environmental predictors identified by DM as important, three new environmental variables (e.g. minimum temperature in May, snow water equivalent in October and precipitation seasonality) were selected as the threshold for the establishment of these high-latitude PFTs. We performed a series of sensitivity experiments to investigate if these thresholds improve the performance of the DGVM method. Based on our results, we suggest implementation of one of these novel PFT-specific thresholds (i.e. precipitation seasonality) in the DGVM method. The results highlight the potential of using PFT-specific thresholds obtained by DM in development of DGVMs in broader regions. Also, we emphasize the potential of establishing DMs as a reliable method for providing PFT distributions for evaluation of DGVMs alongside RS.

Historical CO₂ emissions from land use and land cover change and their uncertainty

Emissions from land use and land cover change are a key component of the global carbon cycle. However, models are required to disentangle these emissions from the land carbon sink, as only the sum of both can be physically observed. Their assessment within the yearly community-wide effort known as the “Global Carbon Budget” remains a major difficulty, because it combines two lines of evidence that are inherently inconsistent: bookkeeping models and dynamic global vegetation models. Here, we propose a unifying approach that relies on a bookkeeping model, which embeds processes and parameters calibrated on dynamic global vegetation models, and the use of an empirical constraint. We estimate that the global CO2 emissions from land use and land cover change were 1.36±0.42 PgC yr−1 (1σ range) on average over the 2009–2018 period and reached a cumulative total of 206±57 PgC over the 1750–2018 period. We also estimate that land cover change induced a global loss of additional sink capacity – that is, a foregone carbon removal, not part of the emissions – of 0.68±0.57 PgC yr−1 and 32±23 PgC over the same periods, respectively. Additionally, we provide a breakdown of our results' uncertainty, including aspects such as the land use and land cover change data sets used as input and the model's biogeochemical parameters. We find that the biogeochemical uncertainty dominates our global and regional estimates with the exception of tropical regions in which the input data dominates. Our analysis further identifies key sources of uncertainty and suggests ways to strengthen the robustness of future Global Carbon Budget estimates.

Improving the representation of high-latitude vegetation in Dynamic Global Vegetation Models

Vegetation is an important component in global ecosystems, affecting the physical, hydrological and biogeochemical properties of the land surface. Accordingly, the way vegetation is parameterised strongly influences predictions of future climate by Earth system models. To capture future spatial and temporal changes in vegetation cover and its feedbacks to the climate system, dynamic global vegetation models (DGVM) are included as important components of land surface models. Variation in the predicted vegetation cover from DGVMs therefore has large impacts on modelled radiative and non-radiative properties, especially over high-latitude regions. DGVMs are mostly evaluated by remotely sensed products, but rarely by other vegetation products or by in-situ field observations. In this study, we evaluate the performance of three methods for spatial representation of vegetation cover with respect to prediction of plant functional type (PFT) profiles – one based upon distribution models (DM), one that uses a remote sensing (RS) dataset and a DGVM (CLM4.5BGCDV). PFT profiles obtained from an independently collected vegetation data set from Norway were used for the evaluation. We found that RS-based PFT profiles matched the reference dataset best, closely followed by DM, whereas predictions from DGVM often deviated strongly from the reference. DGVM predictions overestimated the area covered by boreal needleleaf evergreen trees and bare ground at the expense of boreal broadleaf deciduous trees and shrubs. Based on environmental predictors identified by DM as important, we suggest implementation of three novel PFT-specific thresholds for establishment in the DGVM. We performed a series of sensitivity experiments to demonstrate that these thresholds improve the performance of the DGVM. The results highlight the potential of using PFT-specific thresholds obtained by DM in development and benchmarking of DGVMs for broader regions. Also, we emphasize the potential of establishing DM as a reliable method for providing PFT distributions for evaluation of DGVMs alongside RS.