Student information
MSc thesis topic: Estimating Deforestation Risk in Tropical Forests Through Deep Learning
Deforestation in tropical forests is a major environmental challenge that leads to habitat destruction, biodiversity loss, and escalated global warming through carbon emissions. An effective intervention policy is crucial for climate change mitigation. Advances in machine learning and remote sensing technologies have enhanced our capabilities to predict and monitor deforestation trends. Most existing methodologies employ environmental and socio-economic variables for deforestation prediction, focusing on binary classification outcomes—determining whether or not deforestation will occur within large regions, ranging from 250m x 250m to 1000m x 1000m [1][2][3]. However, this approach overlooks the amount of deforestation within these regions, disregarding the extent of deforestation, which is vital for effective intervention and policy-making. Expanding the predictive models to estimate the area likely to be deforested could offer a deeper understanding of deforestation patterns, enabling more targeted and efficient conservation strategies.
his research aims to enhance deforestation prediction methods by not only assessing the likelihood of deforestation but also quantifying its magnitude. In this thesis you will use environmental variables and satellite alerts (GLAD [4] and RADD [5] alerts) to predict the percentage of area at risk of deforestation in tropical forests using deep learning models
Software: Pytorch (python) and Google Earth Engine
Objectives
- Literature review on deforestation prediction in tropical forests.
- Develop a deep learning model to estimate the percentage of area at risk of deforestation.
- Evaluate the predictive accuracy of the model in comparison to a baseline model.
- Generate a map of future deforestation for target regions: Laos and Colombia.
Literature
- Mayfield, Helen J., et al. "Considerations for selecting a machine learning technique for predicting deforestation." Environmental Modelling & Software 131 (2020): 104741.
- Sánchez, Alexander Cotrina, et al. "Peruvian Amazon disappearing: Transformation of protected areas during the last two decades (2001–2019) and potential future deforestation modelling using cloud computing and MaxEnt approach." Journal for Nature Conservation 64 (2021): 126081.
- DE SOUZA, Rodrigo Antônio; JUNIOR, Paulo De Marco. Improved spatial model for Amazonian deforestation: An empirical assessment and spatial bias analysis. Ecological modelling, v. 387, p. 1-9, 2018.
- Hansen, Matthew C., et al. "Humid tropical forest disturbance alerts using Landsat data." Environmental Research Letters 11.3 (2016): 034008.
- Reiche J, Mullissa A, Slagter B, Gou Y, Tsendbazar N, Odongo-Braun C, Vollrath A, Weisse M, Stolle F, Pickens A, Donchyts G, Clinton N, Gorelick N & Herold M, (2021), Forest disturbance alerts for the Congo Basin using Sentinel-1, Environmental Research Letters.
Requirements
- Advanced Earth Observation course
- Machine learning course
- Geo-scripting course (Good knowledge in scripting is an asset; e.g. python, Google Earth Engine)
Theme(s): Sensing & measuring; Modelling & visualisation