XFires

Summary

Extreme fires are the most damaging in terms of their effects on climate and the wider environment, yet they remain ill defined. The XFire project aims to gain a holistic understanding of extreme fires and their impact in the Earth system. We use a broad selection of relevant Essential Climate Variables to help define and differentiate extreme fires from ‘normal’ fires. We then determine the impact of extreme fires using case studies that will address the following questions:

1. What variables would allow to give a broader and more global definition for an extreme fire?
2. Have extreme fires increased in recent years? Where and why?
3. Can they be predicted?
4. What are impacts of extreme fires on the atmosphere and climate, vegetation, lakes, the cryosphere and on human health?

Background

Fire plays an important role in the Earth system, affecting atmospheric composition and climate, vegetation, soil and societal resources. It is usually seen as a threat, but it is one of the natural Earth system processes and can help rejuvenate ecosystems by leading to succession in different biomes. The balance between positive and negative effects therefore greatly depends on perspective (natural versus anthropogenic for example). The characteristics of fires play a key role; whether they are small or large, intense or calm, frequent or rare, etc. The concept of fire regimes is used to describe these characteristics and becomes even more relevant now that changes in fire regimes are increasingly observed. In regions where fires are becoming more intense or more frequent, for instance, the damages will tend to be higher. Extreme fires are particularly relevant, as they entail the most severe damages, both in terms of social and ecological values. According to the European Commission, within Europe “most damage caused by fires is due to extreme fire events, which only account for about 2% of the total number of fires”.

In addition to the intrinsic importance of extreme fires, their occurrence and impacts are closely linked to climate change, related to different climatic variables, such as soil and vegetation moisture content, biomass, temperature, etc. Fires impact the atmosphere and thus aerosol, greenhouse gases, and ozone concentrations, while the indirect effects of fire-related particles affect also water bodies and ice sheets.

The project first aims to use a cross-Essential Climate Variable (ECVs) approach to define and characterise extreme fires. We will use machine-learning approaches to model extreme fires, and to quantify the impact of extreme fires on the climate and wider environment. These advances will help inform IPCC, the Global Carbon Budget. In addition, Fire ECV has been selected by GCOS as a study case for climate adaptation policies.

Work breakdown structure

The XFires cross-ECV project is designed around 5 main Tasks. The objectives of the first task are to identify and analyse the knowledge gaps on extreme wildfires, and improve the definition of and characterise extreme fires, thereby differentiating them from ‘normal’ fires. We will then explore the spatio-temporal dynamics in extreme fires over recent decades, to address a central question as to whether extreme fires have increased.

Task 2 then assembles and pre-processes a range of ECVs for further analyses. Specifically, we inventorise risk factors, intercompare fire-related ECVs with other datasets and build a community database for exploitation in subsequent Tasks.

In Task 3, we will use these vegetation, climate, atmospheric, fire and land surface data from remote sensing to build machine learning (ML) models for predicting extreme fire burned area and fire radiative power. High-resolution stand-level forest data will be used to assess the combustion fraction, mortality and regrowth of fire-affected vegetation, providing the combustion completeness and hence aboveground carbon emissions rates. These huge improvements to observational constraints will be fed back into the ML model to provide predictive relationships between climate, vegetation, land use and burned area and fire intensity, from which further predictive relationships between these and aboveground biomass emissions can be concretely estimated. We will then use a combination of these outputs and remotely sensed CO data to provide improved constraints and estimates for soil smouldering and combustion emissions. By combining these we will leverage the state-of-the-art across this palette of technologies to provide a first realistic picture of total pre-fire conditions and post-fire impacts of extreme fires on the terrestrial carbon cycle.

Task 4 is dedicated to the characterisation of uncertainty across individual ECVs and error propagation to quantify:

1. uncertainty across different burned area and emissions products; and
2. uncertainty in the climate impact of extreme fires related to the temporal resolution of input emissions datasets, and scaling assumptions applied to emissions.

Task 5 we investigate the impact of extreme fires on land vegetation and soils, and the carbon cycle applying our ML model. We will also generate new emissions associated with extreme fires, and contrast those with ‘normal’ fires for use as input into an Earth System Model, to quantify their impacts on atmospheric composition and climate. Finally, we explore the wider impact of extreme fires on human health, lakes, and via black carbon affecting melt-rates on the Greenland ice-sheet.