- Use cutting-edge data from earth observation satellites and models to quantify the relationships between droughts and forest fires;
- Quantify the effects of climate-induced fires on the carbon cycle of tropical forests;
- Apply this new knowledge of drought-fire feedbacks in new remote sensing methods of forest fire detection;
In 2019, at least 125,000 hectares, equivalent to 172,000 soccer fields, were burned in the entire Brazilian Amazon (https://www.bbc.co.uk/news/world-49515462). The majority of these fire occurrences were observed in the northern state of Amazonas, which alone accounted for 39,100 hectares, or about 30 percent of the total cleared-and-burned areas. Significant deforested-and-burned lands were also detected in the northern states of Rondônia and Pará, where fire hotspots were also prominent in August. However, although most of these fires relate to slash-and-burn agriculture, droughts from previous El Niño events may increase the occurrence of forest fires (McDowell et al., 2018). The combination of land-use activities and El Niño events between 2015 and 2016 in Central Amazon resulted in large wildfire areas around of the Tapajos National Forest (Fig. 1).
Tropical fires represent one of the largest net sources of CO2 to the atmosphere, as well as impacting local and regional air quality (Cochrane et al., 2003). Given that average global temperatures are projected to rise between 1.4 and 5.8°C over the next century, it is likely that small scales of tropical forest fires will undergo dramatic changes in the future (Chen et al., 2011). While land-use and climate models that inform the IPCC have progressed markedly since the late 1980s, the response of tropical fires to climate change is a notable missing process (Poulter et al., 2010). Currently, future tropical forest fire scenarios follow predetermined spatial patterns rather than explicitly responding to the climate changing around them, and there is no feedback between, say, CO2 emissions and climate change. Although there exists a large dataset from remote sensing for tropical forest fires, these remain underexploited for investigating the relationship of fires to climate change. This project will address this knowledge gap and provide new ways to evaluate and develop land surface models. The study will focus on the Brazilian Amazon, where several rainforest areas were burned in 2019, resulting in a severe loss of trees.
The proposed project will quantify and predict tropical forest fires by combining remote sensing and climate modelling, in particular investigating how the reliability of large-scale climate models or data could improve future scenarios tropical forest fire emissions. The project will make use of an extensive archive of remote sensing images (historical Landsat and MODIS satellite images and Sentinel-2 from the European Space Agency) and climate data (Coupled Model Inter-comparison Project) to be incorporated in a process-based dynamic vegetation-terrestrial ecosystem model designed for regional and global studies, the LPJ-GUESS. The student will incorporate well-established methods and algorithms from generalized linear models, classification trees, generalized additive models and random forest models to predict regional scales of tropical forest fires using climate models and remote sensing products. Outputs include changes in vegetation composition and cover in terms of species or plant functional types (PFTs), leaf area index (LAI), biomass, net primary production (NPP), net ecosystem carbon balance and carbon emissions from wildfires.
Training and Skills
The student will be able to access several modules on the new post-graduate programme on “Satellite Data Science” launched this year at the University of Leicester (https://le.ac.uk/courses/satellite-data-science-msc/2020) with the collaboration of Space Park Leicester (https://le.ac.uk/spacepark) and potentially the Catapult Satellite Applications (https://sa.catapult.org.uk/). This will allow the student to use cutting-edge data from earth observation satellites and models. Specifically the student will attend several modules on remote sensing and computer programming languages such as R and Python at the University of Leicester. They will develop skills in big data handling and machine learning, along with quantitative ecosystem modelling, scientific code development and data analysis. All supervisors have expertise in remote sensing and ecosystem modeling and will support the student with the project.
The Doctoral Researcher will be supported in developing research communication skills, including presenting their work at international conferences and publishing their findings in high-impact journals.
Year 1: Meta-analyses of the effects of droughts and climate shifts on fire frequency in the tropics. Characterize the frequency distributions of the forest fires associated with droughts, land-use types and the stage of forest fragmentation in the Amazon tropical forests. Design preliminarily scenarios to measure impacts of essential climate variables (e.g. precipitation and dry season length) on fire regimes. Familiarisation with remote sensing and modelling tools and perform initial simulations for forest fire distributions of the Amazon.
Year 2: Use remote sensing and other earth observation data to quantify forest fires in the tropics using machine learning for image processing. Plan and conduct model simulations for forest fires encapsulating remote sensing data, ground observations, and climate data — presentation of the initial findings in national and international meetings.
Year 3: Produce new products from remote sensing and modelling outputs related to changes in vegetation composition, plant functional types (PFTs), leaf area index (LAI), biomass, net primary production (NPP), net ecosystem carbon balance and carbon emissions from wildfires. Use these products in real cases of forest conservation practice in the tropics, such as forest recovery. Publication of papers and conference presentations at international meetings. Thesis preparation.
Partners and collaboration (including CASE)
The studentship will be offered in collaboration with Leicester Space Park and potentially Catapult Satellite Enterprise. This collaboration offers an excellent opportunity to analyse new remote sensing datasets and models with several experts from academia and industry, expanding the methods of fire detections in a wide range of datasets of earth observation.
Also, the student will benefit from an extensive network of international collaborators from the UK (Universities of Leicester and Birmingham) and Brazil (INPE, the Brazilian Space Agency). They will be part of a much larger team working on the broad topic of forest fires and climate.
Dr Fernando Espírito-Santo (email@example.com) and Prof. Heiko Balzter (firstname.lastname@example.org) in the School of Geography, Geology and the Environment, University of Leicester (https://www2.le.ac.uk/departments/geoggeolenv).
Dr Thomas Pugh (T.A.M.Pugh@bham.ac.uk) in the School of Geography, Geology and Environmental Sciences, University of Birmingham (https://www.birmingham.ac.uk/schools/gees/index.aspx).