Quantifying the impacts of fire aerosols on global terrestrial ecosystem productivity with the fully-coupled Earth system model CESM

LI Fang

International Center for Climate and Environmental Sciences,Institute of Atmospheric Physics,Chinese Academy of Sciences,Beijing,China

ABSTRACT Fire is a global phenomenon and a major source of aerosols from the terrestrial biosphere to the atmosphere. Most previous studies quantified the effect of fire aerosols on climate and atmospheric circulation,or on the regional and site-scale terrestrial ecosystem productivity.So far,only one work has quantified their global impacts on terrestrial ecosystem productivity based on offline simulations, which, however, did not consider the impacts of aerosol-cloud interactions and aerosol-climate feedbacks.This study quantitatively assesses the influence of fire aerosols on the global annual gross primary productivity(GPP)of terrestrial ecosystems using simulations with the fully coupled global Earth system model CESM1.2. Results show that fire aerosols generally decrease GPP in vegetated areas,with a global total of ?1.6 Pg C yr?1,mainly because fire aerosols cool and dry the land surface and weaken the direct photosynthetically active radiation(PAR).The exception to this is the Amazon region,which is mainly due to a fire-aerosol-induced wetter land surface and increased diffuse PAR. This study emphasizes the importance of the influence of fire aerosols on climate in quantifying global-scale fire aerosols’ impacts on terrestrial ecosystem productivity.

KEYWORDS Fire aerosols;terrestrial ecosystem;gross primary productivity;landatmosphere interaction;Earth system model

1. Introduction

Fire is the primary form of terrestrial ecosystem disturbance on a global scale,and plays a key role in the Earth system (Bowman et al. 2009; Li, Bond-Lamberty, and Levis 2014; Li and Lawrence 2017). It burns ～400 Mha of land vegetated area every year (van der Werf et al.2017),and is regulated by climate and weather,vegetation characteristics,and human activities(Bowman et al.2009; Li, Lawrence, and Bond-Lamberty 2018). Fire affects climate and ecosystems mainly by directly adjusting ecosystem structure and functioning as well as by emitting small particles (aerosols) and trace gases into the air (Randerson et al. 2006; Li, Bond-Lamberty, and Levis 2014;Yue et al.2015;Li and Lawrence 2017;Jiang et al.2016;Li et al.2019).

Many earlier studies have quantified the global effect of fire aerosols. However, most of them focused on the impacts on climate, and reported that fire aerosols generally generated a negative radiative effect,reduced surface air temperature and precipitation, and affected the Hadley circulation (Ward et al. 2012; Tosca,Randerson, and Zender 2013; Jiang et al. 2016;Thornhill et al. 2018; and refs therein). Previous studies on the influence on terrestrial ecosystem productivity are mostly at site or regional scales (Kanniah, Beringer,and Hutley 2012,2013;Cirino et al.2014;Rap et al.2015;Yue et al. 2017). They found that fire aerosols can promote photosynthesis by enhancing diffuse light,reduce photosynthesis due to light attenuation associated with increasing aerosol burden,and affect ecosystem productivity indirectly by adjusting surface climate.

So far,only Yue and Unger(2018)have quantified the influence of fire aerosols on global ecosystem productivity,using offline runs of the chemical transport model GEOS-Chem with prescribed atmospheric forcing. The study found that fire aerosols increased global land gross primary productivity (GPP,carbon input of terrestrial ecosystem,the carbon uptake through photosynthesis) by 1.0 ± 0.2 Pg C yr?1. However, Yue and Unger(2018) considered only the direct effect of fire aerosols,and not the aerosol-cloud interactions and aerosol-climate feedbacks as well as their subsequent impacts on ecosystem productivity. On the other hand, recently,Jiang et al.(2020)found that aerosol-cloud interactions and aerosol-climate feedbacks were the dominant factors in the climatic effect of fire aerosols.

The present study provides the first quantitative assessment of fire aerosols on global ecosystem productivity GPP that takes into account the influence of cloudaerosol-climate interactions on terrestrial ecosystem productivity. The impacts of fire aerosols are quantified by comparing present-day simulations with and without fire aerosols based on the atmosphere-land-ocean-seaice coupling of the Community Earth System Model(CESM). It includes a state-of-the-art aerosol model,MAM4, which can model the direct, semi-direct and indirect aerosol effects, and surface-albedo effects.Also, the ocean model rather than fixed-SST forcing method is used to account for both the fast and slow climate system responses to fire aerosols.

Figure 1.Spatial distribution of annual GPP(units:g C m?2 yr?1)from(a)FLUXNET observations and(b)the CESM-simulated FIRE run,averaged for the period 2003-11.The global total(units:Pg C yr?1)and spatial correlation(Cor)between observations and simulations are also given.

2. Model and experiment design

CESM is a widely used global coupled model simulating the Earth’s atmosphere,ocean,land,and sea ice developed by the National Center for Atmospheric Research (Hurrell et al. 2013). The present study adopts the version of CESM1.2 (http://www.cesm.ucar.edu/models/cesm1.2/).The model is comprised of the Community Atmosphere Model,version 5.3(atmosphere component),in which the aerosol model MAM3 is replaced by the new version of MAM (MAM4) to improve the aerosol simulations in the mid-high latitudes (Liu et al. 2016); the Community Land Model,version 4(CLM4,land component);the Slab Ocean Model(ocean component);and Community Ice Code,version 4 (sea-ice component). Photosynthesis in CLM4 is based on the model of Farquhar, von Caemmerer, and Berry (1980), as modified by Collatz et al. (1991) for C3 plants and the model of Collatz, Ribas-Carbo, and Berry(1992)for C4 plants.

Two experiments are run to quantify and understand the impacts of fire aerosols:FIRE and NOFIRE.The setup is the same in both experiments apart from the use of different fire aerosol emissions as input data. FIRE uses the 2003-11 (i.e., nine years) daily black carbon (BC),particulate organic matter (POM), and sulfur dioxide emissions based on the fire surface emissions of the Global Fire Emissions Database, version 3.1 (GFED3.1)(van der Werf et al.2010;Mu et al.2011)and the vertical distribution of emissions scheme in the AeroCom protocol (Dentener et al. 2006). In NOFIRE, fire aerosol emissions are set to zero. Other input data are the default inputs provided with CESM1.2.

The two experiments are run for 99 years with the fire emissions described above, with the simulations of the last 27 years analyzed. Both experiments are run using a finite volume 0.9°(latitude)×1.25°(longitude)grid for the atmosphere,ocean,and land components combined with a displaced-pole grid of around 1° for the sea-ice component,and a temporal resolution of 30 min.

Clark,Ward,and Mahowald(2015)and Grandey,Lee,and Wang(2016)found that sub-monthly and interannual variabilities in fire aerosol emissions are important for estimation of radiative and climate effects of fire aerosols. Therefore, in the present study, the interannually varying daily GFED3.1 fire aerosol emissions are used rather than the default monthly fire aerosol emissions without interannual variability in CESM. In addition, the model platform includes the aerosol indirect effect and climate feedbacks (e.g., fire aerosols increase the Arctic sea ice and snow cover, which lead to long-term changes in surface air temperature and precipitation over local and remote regions). Both have been found to dominate the climatic impacts of fire aerosols in many previous studies (e.g., Jiang et al. 2016, 2020), but were not considered in the only global quantitative study to date about global GPP changes due to fire aerosols -that of Yue and Unger(2018).

3. Results

This section begins with an evaluation of the CESM simulations, and then the fire aerosols’ spatial distribution and their impacts on aerosol optical depth (AOD)are presented. Based on the results, the effects of fire aerosols on the global annual GPP of terrestrial ecosystems are quantified, and the related reasons investigated.

3.1. Evaluation

We first evaluate the GPP simulations from FIRE using FLUXNET-based product (Jung et al.2017).As shown in Figure 1,CESM can replicate the observed spatial pattern of GPP, with a spatial correlation of 0.94 (p< 0.001)between the CESM simulations and observations.It correctly models high GPP values in the tropical forests,moderate values in the tropical savannas, temperate and boreal forests, and croplands, and low values in the shrublands and grasslands. However, CESM generally overestimates GPP in vegetated areas,with a global total of 161 Pg C yr?1for 2003-11,which is 40% higher than the observations and has been reported as a known issue in CLM4 by Lawrence et al.(2011).

Figure 2.Comparison of modeled monthly AOD in the CESM-simulated FIRE run with 2003-11 observations from AERONET sites in(ac) southern Africa, (d-f) South America, and (g) the Arctic. The vertical bars for the AERONET observations are estimated by the standard deviation of multi-year AOD values in that month.See Figure 3(d)for site locations.

Next, the simulated AOD is compared with observations from AERONET (http://aeronet.gsfc.nasa.gov)(Figure 2). The AERONET AOD data at sites (see Figure 3(d) for their locations) largely affected by fire emissions are averaged for the years from 2003 to 2011 to match the period of fire emissions input data used in the present study.As shown in Figure 2,CESM captures the distinct seasonal cycle of the observed AOD at all the sites,i.e.,high values in the fire season,which is the dry season in the tropics and the warm season at mid and high latitudes where fire aerosols in the Arctic mainly come from.It also successfully reproduces the observed AOD for sites over southern Africa (Figure 2(a-c)) and South America(Figure 2(d-f))for the periods outside the fire season. However, it underestimates the AOD in the Arctic and fire-season AOD in southern Africa and South America sites.

Figure 3.Spatial distribution of the annual mean vertically integrated concentration(i.e.,column burden)of fire(a)BC,(b)POM,and(c)sulfate aerosols in the CESM-simulated FIRE run,and(d)changes in AOD induced by fire aerosols(FIRE minus NOFIRE).The locations of the AERONET sites(black dots)in Figure 2 are shown in(d).

3.2. Fire aerosol burdens and impacts on AOD

Figure 3(a-c)shows the spatial patterns of annual-mean fire aerosol burdens. The simulated fire BC, POM, and sulfate aerosols in the atmosphere are mainly distributed in the tropics, with the second high-value region to north of 45° for fire BC and POM. The regions with the maximum fire aerosol burdens over land are close to the post-fire regions (Li, Lawrence, and Bond-Lamberty 2018). In addition, a proportion of fire aerosols exhibit a sufficiently long lifetime to be transported over long ranges from terrestrial post-fire regions to remote open ocean areas and the Arctic.

The fire aerosols can change the AOD.They produce a global area-weighted AOD increase of 7×10?3,corresponding to a relative change of 5.7%.This is lower than the estimate of Tosca, Randerson, and Zender (2013)(10%), which scaled the fire emissions to inflate the modeled AOD. The spatial pattern of the AOD change(Figure 3(d)) is similar to that of the fire POM burdens(Figure 3(b)).

3.3. Impacts of fire aerosols on ecosystem productivity

Fire aerosols reduce GPP in all land vegetated regions except for the Amazon and some regions in southern and eastern North America and in the African savannas(Figure 4(a)).The reduction is most clearly seen in boreal forests and African forests. The total GPP of terrestrial ecosystems is decreased by 1.6 Pg C yr?1, significant at the 0.05 level.In addition,the present results regarding fire aerosols increasing the GPP in the Amazon are consistent with observations (Oliveira et al. 2007; Doughty,Flanner,and Goulden 2010)and an earlier model-based study(Rap et al.2015).

Figure 4.Impacts of fire aerosols(FIRE minus NOFIRE)on global(a)GPP(g C m?2 yr?1),(b)surface air temperature(Tas;units:°C),(c)precipitation (mm d?1), (d) root-zone soil moisture factor (BTRAN, 0-1, unitless, smaller in dryer soil), (e) diffuse photosynthetically active radiation (PAR_diff; units: W m?2), and (f) direct PAR (PAR_dir). Regions are striped where the difference between FIRE and NOFIRE(i.e.,impacts of fire aerosols)passes the Student’s t-test at the 95%confidence level.Global totals of fire aerosols’impacts are also given.

To understand the reasons for fire aerosols’ impacts on GPP, the impacts of fire aerosols on several major variables that regulate GPP and are likely affected by fire aerosols are analyzed, both in observations and the CESM experiments. According to previous studies (Dai,Dickinson, and Wang 2004; Chen et al. 2010; Bonan 2015), GPP is increased in the tropics and decreased in the extratropics over cooler surface air,increased under wetter soil conditions,and increased with higher diffuse and direct photosynthetically active radiation(PAR).

As shown in Figure 4(b),fire aerosols significantly cool surface air over all land grids.They also dry the root-zone soil over land except in the Amazon and some regions in southern and eastern North America and the African savannas (Figure 4(d)), which mainly respond to the change in precipitation(Figure 4(c)).In addition,fire aerosols generally increase the diffuse PAR (Figure 4(e)) but with a stronger decrease in direct PAR (Figure 4(f)), thus leading to a decrease in total PAR.Therefore,fire aerosols significantly reduce GPP in most vegetated areas, mainly because fire aerosols cool and dry the surface in the extratropics and attenuate direct PAR in the tropics where fire aerosols are majorly distributed(Figure 2).In the Amazon,fire aerosols significantly increase GPP due to the induced cooler and wetter land surface and increased diffuse PAR.

4. Conclusions and discussion

This study has quantified and analyzed the influence of fire aerosols on the annual GPP of global terrestrial ecosystems during 2003-11 based on the fully coupled Earth system model CESM1.2.The results show that fire aerosols significantly decrease GPP in vegetated areas, except for the Amazon.The present study shows that fire aerosols at low-to-moderate concentrations increase diffuse PAR and thus tend to increase GPP, and fire aerosols cool the surface air and decrease direct incident solar radiation in the temperate broadleaf forests of the USA.These findings are in good agreement with earlier field- and model-based studies (Kanniah, Beringer, and Hutley 2012, 2013). This study also found that the role of fire-aerosol-induced cooling and drying and light attenuation are dominant on the global scale using the fully coupled CESM, which lead to a decrease in GPP rather than an increase as shown in Yue and Unger (2018) using offline runs. The difference suggests the importance of the feedbacks of fire aerosols to the climate system in quantifying global-scale impacts of fire aerosols on terrestrial ecosystem productivity.

Earlier studies reported that changes in ecosystem functioning and structure due to biomass burning and fire-induced plant mortality can decrease global GPP by 5 Pg C yr?1(Li,Bond-Lamberty,and Levis 2014)and 1 Pg C yr?1(Lasslop et al. 2019). Here, the present results show that fire aerosols decrease global GPP by 1.6 Pg C yr?1,comparable to the impacts of fire-induced direct changes in ecosystem functioning and structure.

Four main sources of uncertainty in the present estimates are worth noting. First, CESM overestimates global GPP (Figure 1), which likely leads to an overestimated influence of fire aerosols on GPP. Second, CESM may underestimate fire aerosol burdens and their impacts,according to the evaluation of AOD in Figure 2. This is partly because small fires are missed in emission product GFED3.1 used in the present study,and surface fire aerosol emissions were not scaled according to AOD before estimating the influence of fire aerosols in this study. Third,the modeling of aerosol-cloud interaction in CESM is still associated with large uncertainties. The radiative effect due to aerosol-cloud interaction estimated by CESM is higher than that in many other climate/Earth system models and observational constraints (Malavelle et al. 2017).Lastly, because the biogeochemical module in CLM4,which models the changes in plant structure (e.g., leaf area) as well as carbon and nitrogen cycles, except for photosynthesis (Oleson et al. 2013), is not open, most responses and feedbacks of terrestrial ecosystem structure and functioning to fire aerosols are not included in the present study.For example,fire itself and respiration may be affected by fire-aerosol-induced surface climate changes. The surface anomalies subsequently change the vegetation canopy,soil heat and water status,vegetation distribution, and thus GPP (Kanniah, Beringer, and Hutley 2012; Li, Bond-Lamberty, and Levis 2014; Li and Lawrence 2017; Li, Lawrence, and Bond-Lamberty 2017),which is not considered in the present study.

The present study focuses on the multi-year average of fire aerosols’impacts.However,a quantitative understanding of the influence of fire aerosols on temporal changes of GPP is also important for Earth system research.This can be achieved using simulations with and without temporal changes in the input data of fire aerosol emissions, and identifying the dominant contributory factor using multiple linear regressions,as in Piao et al.(2013).

Acknowledgments

Thanks to Yiquan JIANG for providing the dataset of CESM simulations;Yiquan JIANG,Zhongda LIN,and Xu YUE for helpful discussions; the two anonymous reviewers for their valuable comments and suggestions; and the editor for handling this paper.

Disclosure statement

No potential conflict of interest was reported by the author.

Funding

This study was co-supported by the National Key R&D Program of China [grant number 2017YFA0604302], the National Natural Science Foundation of China [grant numbers 41475099 and 41875137], and the Chinese Academy of Sciences Key Research Program of Frontier Sciences [grant number QYZDY-SSW-DQC002].