T8 – Modeling the impacts of tropical cyclones on mangroves in the Philippines

Short description

Mangrove forests are fundamental for the sustenance of a series of marine and terrestrial species [1], for the preservation of coastlines [2], for the sustenance of local communities [3], and for capturing carbon. In fact, mangrove forests are one of the most important carbon sinks in the world [4]. Historically, mangroves over-exploitation and deforestation had been the main drivers of the loss of mangrove forest extent. Although increasing efforts put in conservation policy-making are mitigating this threat, extreme climate events are becoming more and more of a threat. Among these, the increasing number of tropical cyclones is the main contributor to mangrove forest losses [5]. Understanding the dynamics of the long-term damage caused by these events requires in-depth analyses, which up to now have been done only for isolated case studies. Unfortunately, close to nothing is available in the literature regarding the impacts of tropical cyclones in Southeast Asia, and specifically the Philippines, even though they boasts one of the largest mangrove forests in Southeast Asia.

This project will seek to systematically analyze and characterize the short- and long-term impacts of tropical cyclones on mangrove forests in the Philippines. The starting point for the project will be a previous thesis that laid the groundwork by assessing the long-term impacts of two notable cyclones, Idai and Kenneth, on the mangrove forests in Mozambique. This thesis will take a significant step further by extending the analysis to all the cyclones on record that have made landfall in the Philippines, with the ultimate goal of understanding the drivers of permanent damage to mangroves after a cyclone hits, as well as the compounding effects of multiple cyclones hitting in short sequence. Optionally, the students may also seek to develop a predictive model (e.g. based on machine learning) that can estimate, immediately after the impact of a cyclone, which areas of a mangrove forest will die in the short- and long-term, so that targeted interventions may be carried out efficiently.

PROPOSED ACTIVITIES

Data acquisition
The main source of data for this project will be Google Earth Engine, where several large satellite datasets (e.g. Sentinel, Landsat) can be accessed and processed on the cloud. However, these data will likely need to be curated extensively, for example to remove cloud cover.

Mangrove segmentation
To estimate the damage to mangrove forests, it will first be necessary to identify from satellite images the extent of these forests. In this phase, the student(s) will adopt well-established classification methods (random forests), adapted from the previous thesis. The goal of this phase is to obtain binary maps (mangrove / non-mangrove) for each month after a cyclone strikes, up to X months, for all cyclones on record.

Statistical / Machine learning analysis
In this phase, the student(s) will thoroughly analyze the short- and long-term effects of cyclones on mangrove forests, identifying key drivers of permanent damage vs recovery, and potentially also developing a predictive model for the recovery of mangrove forests.

References

  1. Ilka C Feller, Catherine E Lovelock, U Berger, Karen L McKee, Samantha B Joye, and MC Ball. Biocomplexity in mangrove ecosystems. Annual review of marine science, 2(1):395–417, 2010.
  2. How Mangrove Forests Protect The Coast – https://www.youtube.com/watch?v=aoMrLYJOdA4
  3. Kanika Bimrah, Rajarshi Dasgupta, Shizuka Hashimoto, Izuru Saizen, and Shalini Dhyani. Ecosystem services of mangroves: A systematic review and synthesis of contemporary scientific literature. Sustainability, 14(19):12051, 2022.
  4. Alongi, Daniel M. “Carbon cycling and storage in mangrove forests.” Annual review of marine science 6.1 (2014): 195-219.
  5. Yu Mo, Marc Simard, and Jim W Hall. Tropical cyclone risk to global mangrove ecosystems: potential future regional shifts. Frontiers in Ecology and the Environment, 21(6):269–274, 2023.

Relevant courses and knowledge

Natural Resources Management 1

Number of students

1/2

Requisites

The student should be comfortable with coding or learning how to code in Python and/or JavaScript.