Quantitative assessment of fire and vegetation properties in simulations with fire-enabled vegetation models from the Fire Model Intercomparison Project

Abstract

Abstract. Global fire-vegetation models are widely used to assess impacts of environmental change on fire regimes and the carbon cycle and to infer relationships between climate, land use and fire. However, differences in model structure and parameterizations, in both the vegetation and fire components of these models, could influence overall model performance, and to date there has been limited evaluation of how well different models represent various aspects of fire regimes. The Fire Model Intercomparison Project (FireMIP) is coordinating the evaluation of state-of-the-art global fire models, in order to improve projections of fire characteristics and fire impacts on ecosystems and human societies in the context of global environmental change. Here we perform a systematic evaluation of historical simulations made by nine FireMIP models to quantify their ability to reproduce a range of fire and vegetation benchmarks. The FireMIP models simulate a wide range in global annual total burnt area (39–536 Mha) and global annual fire carbon emission (0.91–4.75 Pg C yr−1) for modern conditions (2002–2012), but most of the range in burnt area is within observational uncertainty (345–468 Mha). Benchmarking scores indicate that seven out of nine FireMIP models are able to represent the spatial pattern in burnt area. The models also reproduce the seasonality in burnt area reasonably well but struggle to simulate fire season length and are largely unable to represent interannual variations in burnt area. However, models that represent cropland fires see improved simulation of fire seasonality in the Northern Hemisphere. The three FireMIP models which explicitly simulate individual fires are able to reproduce the spatial pattern in number of fires, but fire sizes are too small in key regions, and this results in an underestimation of burnt area. The correct representation of spatial and seasonal patterns in vegetation appears to correlate with a better representation of burnt area. The two older fire models included in the FireMIP ensemble (LPJ–GUESS–GlobFIRM, MC2) clearly perform less well globally than other models, but it is difficult to distinguish between the remaining ensemble members; some of these models are better at representing certain aspects of the fire regime; none clearly outperforms all other models across the full range of variables assessed.

Improving prediction and assessment of global fires using multilayer neural networks

“…Third, a small predicted increase in fires in alreadyarid interior Australia is also surprising, but is consistent with the consensus of other models. However, there is poor agreement in model predictions for historical periods in interior as well as southeastern Australia 31 , and our result here could also be an artefact of data limitations as we argue below. Fourth, one region where our model disagrees with the consensus of other models is northern Australia, where we predict an increase in winter-spring fires with temperature whereas other models predict and agree on a decrease 61 .…”

“…First, the contrast in fire sensitivity to increasing temperature between northern and southern Africa may seem surprising due to similarities in weather and vegetation. However, other models agree that fires in northern Africa may decline by end of the century, whereas model agreement is low for southern Africa both in historical 31,62 as well as projected climates. As we have argued, this could be the effect of differences in the anthropogenic niche of fires in the two regions, resulting in a seemingly anomalous occurrence of fire at low temperatures and high precipitation in southern Africa, compared to the overall trend within SHAF as well as the global temperature-precipitation niche of fires.…”

“…However, despite their complexity and mechanistic appeal, DGVMs have only a modest accuracy in predicting the spatial patterns of burned area, with global spatial correlations between predicted and observed burned area in the range of 0.16-0.69 at a resolution of 1 • -2.5 •29, 30 . Most models are also unable to predict the long-term decline in global burned area over the last two decades 19 , or the interannual variability in burned area, with many DGVMs performing worse than random null models 31 . One reason for the lack of accuracy of global models could be a poor characterization of the human-vegetation niche, which, unlike fire-climate interactions, qualitatively differs between regions.…”

Modeling long-term fire impact on ecosystem characteristics and surface energy using a process-based vegetation–fire model SSiB4/TRIFFID-Fire v1.0

“…The Simplified Simple Biosphere Model (SSiB, Xue et al, 1991;Zhan et al, 2003) is a biophysical model which simulates fluxes of surface radiation, momentum, sensible and latent heat, runoff, soil moisture, surface temperature, and vegetation gross and net primary productivity (GPP and NPP) based on energy and water balance. The SSiB was coupled with a dynamic vegetation model, the Top-down Representation of Interactive Foliage and Flora Including Dynamics Model (TRIFFID), to calculate leaf area index (LAI), canopy height, and PFT fractional coverage according to the carbon balance (Cox, 2001;Zhang et al, 2015;Harper et al, 2016;Liu et al, 2019). We have improved the PFT competition strategy and plant physiology processes to make the SSiB4/TRIFFID suitable for seasonal, interannual, and decadal studies (Zhang et al, 2015;Liu et al, 2019).…”

Technical note: Low meteorological influence found in 2019 Amazonia fires

“…We, therefore, choose to regrid all datasets to a resolution of 2.5 • , as interpolating to a finer resolution would provide no new information about the meteorological drivers tested. MCD64A1, soil moisture and equilibrium fuel moisture content were processed using the "rgdal" (Bivand et al, 2016) and "raster" (Hijmans and van Etten, 2014) packages in R (R Core Team, 2015). For MODIS vegetation continuous field (VCF) fractional covers (Dimiceli et al, 2015), tiles were merged and resampled to the model grid using the "gdal" package (GDAL/OGR contributors, 2018).…”