5th iLEAPS Science Conference Abstracts - A4

Abstracts – Session A4

Impacts of fire on land and atmosphere

A401ORAL-0278: Vulnerability of terrestrial ecosystems to future changes in drought and fire

James Randerson, Niels Andela, Douglas Morton, Guido van der Werf, Yang Chen, Louis Giglio, Gita Lassop, Stijn Hantson, Sander Veraverbeke, Elizabeth Wiggins

Fire is an essential earth system process, influencing biodiversity, nutrient cycling, atmospheric composition, and human health. Over the past 20 years with advances in satellite remote sensing, it has become possible to investigate patterns of variability and decadal trends in fire activity at a global scale. Here we show that global burned area has changed considerably over the last two decades, with important implications for ecosystem conservation. Using several types of analysis we then assess the role of human and climate drivers in shaping the long-term trends. The creation of a high-quality fire time series also has enabled the community to develop a quantitative understanding of fire responses to El Niño and other climate modes, allowing for the development of early warning forecasting systems that predict fire season severity on seasonal timescales. These systems may create new opportunities for sustainably managing forests, and here we highlight several lessons learned from the development of an early warning system for the Amazon. We also describe how future precipitation changes may vary in sign and magnitude across different tropical continents, and the implications of these changes for the vulnerability of tropical forests to fire and other agents of global change. Finally, we will discuss how fire-climate feedbacks mediated by lightning may accelerate carbon losses from permafrost soils in the Arctic.

A402ORAL-0396: New Earth Observation based Capabilities for Regional-to-Global Landscape Fire Emissions Estimation

Martin Wooster1, Daniel Fisher1, Jiangping He1, Tianran Zhang1, Tadas Nikonovas1, Weidong Xu1

1King's College London, London, United Kingdom

Landscape burning is a globally prevalent but unpredictable phenomenon that in many regions displays large interannual variability. Fires burn on average across an area of the Earth similar to that of India every year. Landscape fires thus have a very significant impact on Earth’s atmosphere, as well as on many biogeochemical cycles, and affect human health and the climate through their substantial smoke aerosol, gaseous pollutant and GHG emissions. Observations by Earth orbiting satellites represent arguably the most successful approach to quantifying the atmospheric impacts of landscape fires.  Many highly capable Earth Observing satellite instruments have been deployed in orbit, alongside a parallel development of methods required to extract ever more quantitative fire-relevant information. The use of thermal remote sensing methods enables fire signatures to be identified and quantified even whilst the fires are still burning.  Such data has led now to a much more complete understanding of regional and global landscape fire activity and its atmospheric impact, and to the ability to assess this very rapidly where necessary. Here we will present information on how new satellites and instruments, such as Himawari-8 and NPP VIIRS, Sentinel-3 SLSTR, GOES-R are enabling continual improvement in this capability, to such an extent that we now see systems like the Copernicus Atmosphere Monitoring Service (CAMS) permitting the delivery of these data in near reat time. This enables regional and global-scale routing monitoring for fire-related air pollutant and GHG transfers to the atmosphere, as well as forecasts indicating how situations of poor air quality related to new fire activity will likely develop over the coming days.

A403ORAL-0257: Remote sensing supporting fire monitoring

Susanne Mecklenburg1

1European Space Agency, Frascati, Italy

Remote sensing offers an efficient tool to monitor land surface condition that might lead to a risk of fires. This paper will focus on data products derived from the European Space Agency’s (ESA) Soil Moisture and Ocean Salinity (SMOS) and the Copernicus Sentinel-3 missions providing complementary land surface parameters that are either an indicator for the risk or a direct measurement of fires.
The SMOS mission, in orbit since 2009, carries a passive microwave interferometric radiometer measuring in L-Band, providing accurate global observations of emitted radiation originating from the Earth’s surfaces since the atmosphere is almost transparent in this spectral range. In addition, over land the effect of vegetation on the measurements is smaller than for shorter wavelengths. SMOS provides global measurements of surface soil moisture. These measurement provide the basis of root zone soil moisture data which form the basis for drought predictions, an important monitoring and prediction tool for plant available water but also fire risk. In addition, SMOS provides a vegetation optical depth (VOD), a key parameter for assessing plant water stress and monitoring plant health, an indicator of wet/dry conditions. SMOS surface soil moisture measurements have been assimilated into numerical weather prediction, carbon assimilation schemes and evaporation models. This paper will give a synthesis of the current work.
The Copernicus Sentinel-3 mission, in orbit since 2016, carries as an optical payload the Ocean and Land Colour Instrument (OLCI), providing information about vegetation status and health, and the Sea and Land Surface Temperature Radiometer (SLSTR), delivering in addition to land surface temperatures also a fire radiative power (FRP) product, thanks to its new fire channels compared to its predecessor AATSR.

The paper will present the various SMOS and Sentinel-3 data products, explore their complementarity and – if available – show their use in model predictions.

A404ORAL-0301: Assessing fuel consumption in FireMIP models

Stéphane Mangeon2, 1, Apostolos Voulgarakis1, Stefanos Mousafeiris1, Gerd Folberth3, Niels Andela4, Guido van der Werf5, Dominique Bachelet6, Matthew Forrest7, Stijn Hantson8, Gitta Lasslop9, Fang Li10, Joe Melton11, Chao Yue12

1Imperial College, London, United Kingdom 2Singapore-MIT Alliance for Research and Technology, Singapore, Singapore 3Met Office Hadley Centre, Exeter, United Kingdom 4NASA Goddard Space Flight Center, Greenbelt, The United States of America 5VU University Amsterdam, Amsterdam, The Netherlands 6Oregon State University, Corvallis, The United States of America 7Senckenberg Biodiversity and Climate Research Institute, Frankfurt am Main, Germany 8Karlsruhe Institute of Technology, Institute of Meteorology and Climate research, Atmospheric Environmental Research, Garmisch-Partenkirchen, Germany 9Max Planck Institute for Meteorology, Hamburg, Germany 10International Center for Climate and Environment Sciences, Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing, China 11Environment Canada, Victoria, Canada 12Laboratoire des Sciences du Climat et de l’Environnement, LSCE CEA CNRS UVSQ, Gif Sur Yvette, France

Landscape fires are a phenomenon rooted in physics and chemistry that occur on small spatial scales, while on larger scales they depend on climate, ecosystem type, and human interventions. Global fire models have been developed mainly in the last decade, and typically focus on estimating burnt area and carbon emissions. These models follow either semi-empirical approaches based on large-scale observations, or more detailed mechanisms such as the Rothermel spread model to estimate key quantities such as burnt area and emissions. Yet, a purely physical quantity that is less studied is Fuel Consumption (FC), i.e. the amount of fuel consumed during a fire event. Examining FC provides insight into whether models are capturing processes underlying landscape fires accurately. Even when models show a good skill in capturing burnt area and emissions, if their FC is found to be problematic, there is a risk of making unreliable predictions of fire-related quantities under e.g. future scenarios. We investigated FC as part of the Fire Model Intercomparison Project (FireMIP), the first effort to intercompare global fire models. We found that some models have an alarming tendency to predict extreme FC. We will also show that models systematically overestimate FC in lightly vegetated areas (grasslands/savannahs), while underestimating that in forested areas. This underestimate is particularly pronounced for tropical forests, where FC is in reality very high, largely due to human deforestation practices. This reinforces one of FireMIP’s key overall findings: human-fire interactions should not be ignored when simulating large-scale fire activity and emissions. We also stress that the skill of the underlying land/vegetation models in which fire models are typically built plays an important role. Finally, it is promising that the models which seemingly best capture FC use a simple and transferable approach.

A405ORAL-0042: Consideration of wildfires as a source of mineral dust emitted into the atmosphere – Investigation of the conceptual model using

Large-Eddy-Simulations (LES)

Robert Wagner1, Kerstin Schepanski1, Michael Jähn1

1Leibniz Institute for Tropospheric Research (TROPOS), Leipzig, Germany

Wildfires are a common phenomenon in semi-arid regions nearly all over the world. Especially the Sahel in Sub-Sahara Africa is a very active fire spot during the dry season. Most of these fires are caused by human activity making them to an important anthropogenic aerosol source, which modulates the global atmospheric aerosol load.
Fires can impact significantly on the surface conditions making the fire area prone to wind erosion after the fire event. Besides soot particles, mineral dust particles are entrained into the atmosphere by the fire, which ultimately contribute to the aerosol composition and alter the aerosol’s optical, physico-chemical and microphysical properties. However, the quantity of mineral dust emitted during fire events is still undetermined.
Here, high resolving Large-Eddy Simulations (LES) were performed using the All Scale Atmospheric Model (ASAM) in order to investigate the impacts of fires on the near-surface wind pattern that drive fire-associated dust emissions. By means of case studies the influences of different fire properties (fire intensity, size, and shape) and different atmospheric conditions on the strength and extent of fire-related winds and finally their relevance for dust emissions were investigated. First results show that in the surrounding of the fires a strong increase in the occurrence of high wind velocities takes place. This results in exceeding typical threshold velocities for dust emission and allows an efficient entrainment of dust particles in the atmosphere. This process and thus the amount of emitted particles are very sensitive to the fire and atmospheric properties.
Results from this study will support the development of a parameterization of fire-related dust aerosol entrainment for atmosphere-aerosol models. This will allow for an estimate of such fire-related dust emissions at continental scale and will finally contribute to the reduction of the uncertainty in the aerosol-climate feedback.

A406 ORAL-0330: Human-caused fires do not limit convection in tropical Africa – a reinterpretation of the data

Sally Archibald1, Paul Laris2

1University of the Witwatersrand, Johannesburg, South Africa 2California State University, Long Beach, The United States of America

There is evidence that particulates from fires reduce cloud cover and that this has implications for the timing of rainfall in fire-prone regions. These results have been interpreted as evidence that human ignitions in Africa reduce rainfall and cause aridity. There are several reasons why this is likely to be incorrect. Here we discuss the controls on climate and fire in these grass-fueled fire regimes, and present evidence that human ignitions tend to bring the fire season earlier, rather than later in the dry season: their burning practices are linked more to vegetation moisture than weather conditions. Misunderstanding of how and why people light fires in these regions has led to very inaccurate representations of land-atmosphere feedbacks. We reinterpret past analyses in the light of these results and suggest that earth system scientists and social-ecologists need to work more closely together to ensure accurate understanding of human impacts on the earth system.

A407ORAL-0315: Impact of biomass burning on air quality in South China and Mainland Southeast Asia

Carly Reddington1, Dominick Spracklen1, Stephen Arnold1, Luke Conibear1

1School of Earth and Environment, University of Leeds, Leeds, United Kingdom

Rapid economic growth combined with inadequate environmental legislation has led to serious air quality problems across Asia. Efforts to improve air quality are hindered by poor understanding of pollutant sources and processes that lead to unhealthy air. Previous studies have focussed heavily on the contributions of local anthropogenic emissions e.g., traffic, industry, shipping etc. to air pollution. Whilst these emissions are certainly important, observations suggest that many cities in Asia are also influenced by regional-scale emissions from other sources such as agricultural waste burning, residential fuel combustion and forest fires. Our previous research suggests that these sources may offer considerable, yet largely unrecognised, options for rapid improvements in air quality. In this work, we use a combination of models and observations to quantify the contribution of biomass burning (including wildfires, agricultural fires and deforestation fires) to air quality degradation in major cities in South China and Mainland Southeast Asia. Specifically, we use a global aerosol microphysics model (GLOMAP) to examine long-term trends in particulate matter concentrations from fire emissions in the region; and a higher resolution, regional model (WRF-Chem) to quantify the contribution of biomass burning to air pollutant concentrations and calculate the negative impacts on human health.

A408ORAL-0250: Fire pollution for preindustrial, present day and future conditions in an interactive Earth System Model

Keren Mezuman2, 1, Susanne Bauer2, 3, Kostas Tsigaridis2, 3

1Earth and Environmental Sciences, Columbia University, New York, NY, The United States of America 2NASA Goddard Institute for Space Studies, New York, NY, The United States of America 3Center for Climate Systems Research, Columbia University, New York, NY, The United States of America

A climate model with prognostic biomass burning allows us to study the drivers, feedbacks, and interactions of fire in time periods outside of the satellite era. As recent works have shown (e.g. Westerling et al., 2006; Veira et al., 2016) a region’s fire activity is sensitive to changing temperatures and the arrival of spring, i.e. a changing climate. Other than regulating the atmospheric carbon monoxide budget, fires release to the atmosphere a suite of reactive gases and aerosol particles that affect air quality. We set out to study fire pollution of different regions in the world under different climate conditions by further developing the GISS fire model (Pechony and Shindell, 2009, 2010). We correlated the modeled flammability with MODIS fire counts, in a vegetation specific parameterization, which allowed us for the first time to interactively simulate climate and fire activity with GISS-ModelE2.1. Biomass burning occurrence was driven by environmental factors such as vapor pressure deficit and precipitation, as well as natural and anthropogenic ignition. With this new method we were able to attribute the source of the fire to either natural or anthropogenic origin. Present day results were evaluated against GFED4s data. Our results indicate that fire pollution is high in all time periods, but expected to play a bigger role in the future. We also show that humans play an important role in the spatial distribution of fire activity, and in curbing fire pollution.

A409 -ORAL-0035: Landscape fires dominate terrestrial natural aerosol – climate feedbacks

Catherine Scott1, Stephen Arnold1, Sarah Monks2, 3, Ari Asmi4, Pauli Paasonen4, Dominick Spracklen1

1School of Earth and Environment, University of Leeds, Leeds, United Kingdom 2Cooperative Institute for Research in Environmental Sciences, University of Colorado, Boulder, Colorado, The United States of America 3Chemical Sciences Division, NOAA Earth System Research Laboratory, Boulder, Colorado, The United States of America 4Department of Physics, University of Helsinki, Helsinki, Finland

The terrestrial biosphere is an important source of natural aerosol including landscape fire emissions and secondary organic aerosol (SOA) formed from biogenic volatile organic compounds (BVOCs). Atmospheric aerosol alters the Earth’s climate by absorbing and scattering radiation (direct radiative effect; DRE) and by perturbing the properties of clouds (aerosol indirect effect; AIE). Natural aerosol sources are strongly controlled by, and can influence, climate; giving rise to the potential for natural aerosol climate feedbacks.
Earth System Models (ESMs) include a description of some of these natural aerosol–climate feedbacks, predicting substantial changes in natural aerosol over the coming century with associated radiative perturbations. Despite this, the sensitivity of natural aerosols simulated by ESMs to changes in climate or emissions has not been robustly tested against observations.
Here we combine long-term observations of aerosol number and a global aerosol microphysics model to assess terrestrial natural aerosol-climate feedbacks. We find a strong positive relationship between the summertime anomaly in observed concentration of particles greater than 100 nm diameter and the anomaly in local air temperature. This relationship is reproduced by the model and driven by variability in dynamics and meteorology, as well as natural sources of aerosol.
We use an offline radiative transfer model to determine radiative effects due to changes in two natural aerosol sources: biogenic SOA and landscape fire. We find that interannual variability in simulated global natural aerosol radiative effect (RE) is negatively related to the global temperature anomaly. The magnitude of global aerosol-climate feedback (sum of DRE and AIE) is estimated to be -0.15 W m‑2 K-1 for landscape fire aerosol and -0.06 Wm-2 K-1 for biogenic SOA. These feedbacks are comparable in magnitude, but opposite in sign to the snow albedo feedback, highlighting the need for natural aerosol feedbacks to be included in climate simulations.

A410ORAL-0017: Impact of land surface modification on fire hazard in the humid tropics

Taufik Muh1, Henny Van Lanen1

1Wageningen University, Wageningen, The Netherlands

Vast areas of wetlands in Southeast Asia are now into a transformation process to a human-modified ecosystem. Expansion of agriculture cropland and forest plantations changes the landscape of wetlands in this region. Here we present observation-based modeling evidence of increased fire hazard due to land surface modification by changing the vegetation cover, incl. canalization in tropical wetland ecosystem. We tested two contrasting wetlands, namely a natural wetland and a canalized wetland for the period 1980-2015. Our results show that uncontrolled canalization causes very low groundwater levels during the dry season leading to an increase of high fire hazard from around 4% of the time under natural drainage to an extreme level of about 40%. We found that improved water management through controlled drainage reduces high fire hazard (17% of time lower threat). However, controlled drainage still triggers the fire season to come 1-2 months earlier than under natural wetland conditions, indicating that the canal water regime is a key variable to control high fire hazard. The findings suggest that improved water management can reduce fire susceptibility with noticeable effects on the environment even far beyond the fire-burnt areas.

A411ORAL-0361: Relationship between Fire Hotspots and Forest Loss  in Riau Province, Indonesia between 2000 and 2013

Hari Agung Adrianto2, 1, Dominick Spracklen2, Stephen Arnold2, Wolfgang Buermann2

1Bogor Agricultural University, Bogor, Indonesia 2School of Earth and Environment, University of Leeds, Leeds, United Kingdom

Forest and peatland fires occur regularly across Indonesia resulting in large greenhouse gas emissions and causing major air quality issues. Over the last few decades Indonesia has also experienced extensive forest loss and conversion of forest to plantations. Previous studies suggest that fire is connected to this land cover change through a multi-year process, However, it is not known what fraction of fire occurs before, during or after land-use change. Using long-term remote sensing observations, it is possible  to explore spatial and temporal relationships between forest loss and fire frequency. In this study, we combine MODIS MCD14ML fire hotspot data with Hansen Global Forest Change datasets at 250m cells. We focus on the Riau Province in Central Sumatra, one of the most active regions of fire in Indonesia. Over the period 2000 to 2013, there were 111770 hotspots and 31% of the province experienced forest loss. Only 20 percent  cells have two or more hotspots, indicating a low reoccurrence of fire at the same location. At the provincial scale, there is relatively low correlation (r=0.44) between annual extent of forest loss and annual hotspot count. The number of hotspots in areas with forest loss (forest loss extent >75%) is 3.2 times greater than in areas with no forest loss (loss <25%). Furthermore, we calculated the time difference between the year of forest loss (2005-2007) and the year with the maximum number of hotspots (2000-2013). Both events occurred in the same year in 59% of cells, whereas in 13% of cells maximum hot spot year occurred before forest loss and 28% cells it occurred after forest loss.   This shows a strong connection between fire and forest loss at the local scale.

A412ORAL-0138: Modelling Fire Danger Rating over South Africa - Now and into the Future

Roland Schulze1, Stefanie Schütte1

1University of KwaZulu-Natal, Pietermaritzburg, South Africa

As a point of departure, a fire danger rating over South Africa, based on ambient daily weather conditions of maximum temperature and minimum relative humidity, is used to classify each of 5838 agro-hydrologically relatively homogenous response areas into four classes of “unlikely”, “unfavourable”, “favourable” and “very likely” fire conditions on a daily basis. This is done using 50 years of daily climate data for each of the nearly 6000 areas. The number of days with fire conditions in the four classes are then mapped on a seasonal basis. Modifications to the climate driven fire danger rating are then made,
  • first in regard to the vegetative biomass’s fuel load for both present and projected future climatic conditions based on attributes of 70 natural vegetation types identified in South Africa and their likely shifts, 
  • then with respect to soil wetness and vegetation wetness and
  • finally taking likely wind conditions into account.
In order to assess possible impacts of climate change, the simulations are repeated using daily climate outputs for present and future conditions from four GCMs and seasonal averages of fire danger in the four classes are mapped. Major spatial changes in fire danger ratings are evident into the future and these are assessed, also with some thoughts given to possible adaptation options.

A413 ORAL-0184: Determinants of fire intensity, severity and greenhouse gas emissions for savanna fires in West Africa

Paul Laris1, Rebecca Jacobs1, Moussa Kone2, Fadiala Dembele3

1California State University, Long Beach, The United States of America 2Université Félix Houphouët-Boigny , Abidjan, Côte d'Ivoire 3Institut Polytechnique Rural de Formation et de recherché Appiquée , Katibougou, Mali

African Savanna fires emit large quantities of greenhouse gases.  While it is increasing recognized that these fires play an important role in the global carbon cycle, there are few accurate estimates of their emissions and none from West Africa which is that continent’s most active fire region. Most estimates of emissions from savannas contain high levels of uncertainty because they have been based on very broad generalizations of complex landscapes and burning practices. To improve emissions estimates, this study used a novel approach to develop a model based on the actual burning practices of people who set fires in the mesic savanna of Mali. To determine the factors that most influence fire emissions of three key gases--methane, carbon dioxide and carbon monoxide--we conducted over 200 experimental fires and used a portable gas analyzer to measure emissions at the plot level. Burn plots (vegetation type) and season of burning (early, middle or late) were selected using two methods; the first based on local burning practices and the second based on a random fire regime. Data were collected for fire season, savanna type, grass type, biomass consumed, scorch height, speed of fire front, fire type and ambient air conditions at two sites in Mali. We used multiple regression analysis to determine the key factors effecting the fire intensity, severity and emissions of CO2, CO and CH4. Preliminary results suggest that fire type and fuel load and type are important determinants of fire intensity, severity and emissions. The implications of traditional burning practices on these factors are discussed.

A414ORAL-0408: Drivers of past biomass burning in the central European lowlands during the Holocene

Elisabeth Dietze1, Martin Theuerkauf2, Michał Słowiński 3, Achim Brauer1

1GFZ German Research Centre for Geosciences, Section 5.2 Climate Dynamics and Landscape Evolution, Potsdam, Germany 2University Greifswald, Institute of Botany and Landscape Ecology , Greifswald, Germany 3Polish Academy of Sciences, Stanisław Leszczycki Institute of Geography and Spatial Organization, Warsaw, Germany

The central European lowlands (CEL), south of the Baltic Sea, are characterized by clear gradients in modern climate. In particular, a more continental climate prevails in northern Poland compared to northern Germany and the Baltic states. The CEL share a similar geological history with the last retreat of the Scandinavian Ice Sheet leaving a diverse geomorphological and pedological setting that provided the scene for the migration of natural vegetation since the late Glacial. Also human settlement history during the Holocene was regionally different, as evidenced by palynological records, leading to a  complex pattern of CEL land cover history.
Wildfires in the region are today largely restricted to pine forests as the temperate deciduous forests are not fire-prone. Furthermore, today’s forest fires are largely suppressed by forest management, whereas humans used fire in the past as land management tool. However, little is known about the contribution of fire to past land cover change as well as about the past (and potential future) interaction between human activity, climate, vegetation and fire in this region.
To address this gap in knowledge, we will present a first synthesis of Holocene fire history based on c. 60 sedimentary charcoal records from lakes and peat archives from N Germany, N Poland and the Baltic states. By comparing charcoal composites of several regional and climatic clusters, we discuss a) potential climatic drivers and b) the impact of different human land use histories. By a comparison with quantitative vegetation reconstructions and other available proxy and climate model data, the relationship of fire with land cover/land use, and climate will be discussed.
The study shows the great importance of pine as fuel and human hand as ignition source that challenge the clear fire-climate relationship known from other biomes.

A415 POSTER-0144: Evaluation of fire-vegetation-climate relations

Gitta Lasslop, Thomas Moeller, Donatella D'Onofrio, Stijn Hantson, Silvia Kloster

Fire has a strong impact on the global distribution of vegetation, especially in the tropics. Fire is strongly driven by climate but also modulated by vegetation. With global warming fire occurrence is often expected to increase. This may lead to increased carbon emissions and in consequence an amplification of human induced climate change. However, increased drought and higher temperatures might also lead to a reduction in fuel load and a reduction of fire emissions. To increase our confidence in projections of fire occurrence we evaluate the relation between tropical climate, vegetation and fire in a global vegetation model which incorporates both a simple and a complex fire scheme. We aim at identifying potential for model improvement with respect to the interaction between climate, vegetation and fire based on the evaluation. Model simulations are used to investigate the effect of land use on differences in climate-vegetation-fire relations between continents.
We use two global vegetation datasets based on remote sensing and one site level dataset for Africa to analyze the relationship between precipitation, vegetation cover and fire for the tropical area. We compare these relationships to model simulations of the land surface model JSBACH coupled to both a simple and a complex (SPITFIRE) fire algorithm.
The complex fire algorithm strongly improves the spatial pattern of burned area compared to the simple fire scheme. JSBACH shows too high tree cover for low precipitation in comparison to the satellite data and the site-level dataset. We find that the relationship between maximum tree cover and precipitation depends on the spatial scale. Differences in the climate-fire-vegetation relationship between Africa and South America, e.g. a lower fire occurrence in South America can partly be attributed to land use.

A416 POSTER-0147: Global wildfire patterns modeLling induced by climate change

Chao Wu1, Sergey Venevsky1, Stephen Sitch2, Lina Mercado2, Chris Huntingford3

1Ministry of Education Key Laboratory for Earth System Modeling, Department of Earth System Science, Tsinghua University, Beijing, China 2College of Life and Environmental Sciences, University of Exeter, exeter, United Kingdom 3Centre for Ecology & Hydrology, Wallingford, United Kingdom

Wildfire is an important and necessary natural disturbance in forest ecosystems, helping to shape global biome distribution and maintain the structure and function of fire-prone communities. Meanwhile, it is also seen as a risk to human societies. Therefore, the frequency and spatial distribution of wildfires must be better understood. Especially in the background of climate change, understanding how fire patterns would change and its impacts on terrestrial ecosystem composition and carbon cycle in the future are important and diverse questions ranging from wildfire management strategies and climate change mitigation. Although the global burnt areas are expected to increase in response to climate warming, there exist great uncertainties. Uncertainties are associated with different representations of vegetation and fire process in the models on the one hand, and differences in projected changes in climatic patterns on the other hand.  Here we used simulations with 34 climate models and a Dynamic Global Vegetation Model: SEVER-DGVM, including a process-based fire module: SEVER-FIRE to explore the change of global burnt area and terrestrial ecosystem carbon cycles in response to climate change based on four Representative Concentration Pathways (RCPs) and their uncertainties.  The results will benefit for ecologists and policy makers.

A417POSTER-0201: Evaluating aerosol emissions from vegetation and peat fires in Equatorial Asia

Laura Kiely, Dominick Spracklen, Stephen Arnold, John Marsham

Equatorial Asia contains large areas of forested peatland which are undergoing deforestation and drainage to make way for plantation agriculture. The area has seen increased burning as fire is used to clear land, and drained peatland is more susceptible to fires. Vegetation and peat fires release large amounts of aerosol particles with substantial impacts on air quality and climate. Aerosols from fires alter cloud properties, which can impact the climate and rainfall. Understanding the full extent of these emissions is essential for assessing the impact of fires.
Emissions from these fires are calculated using satellite data on fire location with information on the available fuel and emissions factors for that fuel type. However, peat can burn underground to varying depths, does not regenerate between fires, and can smoulder for long periods of time. Therefore, emissions from peat fires are particularly uncertain and may be misrepresented in fire emissions datasets. The aim of this work is to evaluate aerosol emissions from vegetation and peat fires over Equatorial Asia.
We have used three datasets of aerosol emissions from fires (GFED, GFAS and FINN) and satellite aerosol optical depth (AOD). For each dataset, the aerosol emissions to AOD ratio for peat fires has been compared to the ratio for vegetation fires. We used the Weather Research and Forecasting model coupled with chemistry (WRF-Chem) to simulate AOD over the same time period and region, then analysed the model in the same way as the observations. We used these comparisons to assess whether there is evidence that emissions from peat fires are being misrepresented in emissions datasets.

A418 -POSTER-0206: The impact of human fire starts and land use on burnt area

Douglas Kelley1, Rhys Whitley2, Ioannis Bistinas3, Chantelle Burton4

1NERC center for Ecology and Hydrology, Wallingford, United Kingdom 2Suncorp Group, Personal Lines Pricing Research, Sydney, Australia 3Vrije Universiteit Amsterdam, Amsterdam, The Netherlands 4Met Office Hadley Centre for Climate Science and Services, Exeter, United Kingdom

Global fire models typically describe fire as a consequence of fuel load, moisture, natural and anthropogenic ignitions, and land use suppression. A lack of information on the temporal and spatial distribution of these controls has meant that their effects on simulating burnt area are largely untested. Despite this, there is a pervasive assumption that fire is proportional to the number of ignitions, with many models simulating burnt area mainly as a result human fire starts. Here, we map the limitation and sensitivity of burnt area to each control using a simple framework whereby limitations are imposed by: fuel discontinuity; fuel moisture and atmospheric drying potential; lightning and human ignitions; and land use. Controls are described from remote sensed and meteorological observations and optimized against observed burnt area.
Fuel moisture and fuel production are shown to be the main limitations of fire over much of the world. This is followed by land use, with agriculture shown to reduce burnt area in adjacent, none agricultural areas. Ignitions only limit in dry season savanna, where rapid drying of fuel built up during the wet season removes all natural limitations. Human ignitions only contribute a small increase in global burnt area, dramatically offset by the impact of suppression through land use. The assumption that burnt area is a result of human fire starts at a global scale is clearly incorrect, and adequate description of suppression through land use should become a priority to correctly simulate burnt area. However, some areas of the world, including the Amazon and Boreal forests, are still sensitive to small changes in ignitions despite fuel moisture being the main limiting factor. In these regions, correct modulation of fire regimes by potential changes in ignitions sources will be vital in assessing the impact of fire on future ecosystem function and services.

A419 -POSTER-0342: Interactive INFERNO

Chantelle Burton2, 1, Richard Betts2, 1, Chris Jones1, Ted Feldpausch2, Andy Wiltshire1, Eddy Robertson1

1Met Office Hadley Centre, Exeter, United Kingdom 2University of Exeter, Exeter, United Kingdom

Fires play an essential role in vegetation dynamics, the carbon cycle, hydrological cycle and atmospheric chemistry, and are one of the most important disturbance factors globally. Over the coming decades we are likely to see increasing shifts in global climate conditions such as rising temperatures, heat extremes, droughts and altered precipitation patterns as a result of climate change. Coupled with socio-economic drivers of deforestation, these shifts could have a substantial impact on the risks and resilience of ecosystems to fire. INFERNO (INteractive Fire and Emission algoRithm for Natural envirOnments (INFERNO), a diagnostic fire model for burnt area and emissions, was implemented in the Dynamic Global Vegetation Model JULES (Joint UK Land Environment Simulator) last year. It uses fuel flammability and ignition to calculate potential burnt area for the JULES plant functional types. Since then work has been ongoing to develop the model capability by coupling to dynamic vegetation and soil pools in order to model interactive fire-vegetation processes on a global scale, and to improve understanding of fire land-surface interactions. Here we present the first results from Interactive-INFERNO.