Use this URL to cite or link to this record in EThOS:
Title: Identifying the vulnerable forests of Southeast Asia, and transforming them into a conservation and climate change mitigation priority
Author: Nomura, Keiko
ISNI:       0000 0004 8508 7087
Awarding Body: University of Edinburgh
Current Institution: University of Edinburgh
Date of Award: 2019
Availability of Full Text:
Access from EThOS:
Full text unavailable from EThOS. Please try the link below.
Access from Institution:
Facing the threat of climate change, preventing land use change in tropical forest areas has been identified as one of the main strategies to reduce carbon emissions to the atmosphere. However, the rate of tropical forest loss has increased rather than decreased over the recent decades, questioning the effectiveness of current approaches in bringing about the necessary changes. To obtain better forest protection and ensure a reduction in emissions, new approaches need to be explored. In Southeast Asia, forest loss is particularly pronounced due to the dominance of agriculture and plantation forestry. The region has experienced a total loss of 11.3% of its forest cover since the beginning of the 21st century, and the rate of loss shows little sign of slowing. Therefore I use Southeast Asia as a case study to present a pragmatic approach to identify and measure forests at risk from deforestation. My aspiration is to develop an approach applicable to the region, which can then be easily adapted globally. I present three core chapters in this thesis. After the introduction, in Chapter 2, I examine whether the current international incentive-based mechanism to reduce emissions from deforestation and forest degradation (REDD+) is well suited to identify historically vulnerable forests, and whether it is likely to lead to real emission reductions. First, I identify and measure the current areas of forests under REDD+ in the Asia and Pacific region. I compare the benchmark emissions from forests ('reference levels') submitted by the governments to the United Nations Framework Convention on Climate Change (UNFCCC) with forest area change estimates using the Global Forest Change v1.4 (GFC) dataset. The v results show consistent differences, with most countries reporting considerably less historic forest loss than the GFC-based analysis. These differences are due to: the countries' selection of activities to report; as well as their choice of forest types and land use; and the selected definitions of the forests to be monitored. Therefore, even if REDD+ is successfully implemented, it will not necessarily lead to emission reductions. In Chapter 3, I identify these vulnerable forests and the drivers of deforestation. I use publicly available satellite data (Sentinel-2) to map 13,330 ha in southern Myanmar. This area is a mixed landscape combining large areas of both natural forest and commercial plantations (mostly of oil palm and rubber). I use Google Earth Engine as a data analysis platform to conduct supervised land cover classifications using a machine learning algorithm. The classifier is able to detect the differences between visibly similar tree crops (e.g. oil palm, rubber, betel nut, and forests) with high accuracy (95.5% - 96.0%) at a 20 m resolution. Based on the results of this initial study, I then scale up the analysis to all of southern Myanmar (more than 4 million ha) and add radar (Sentinel-1 and the Shuttle Radar Topography Mission) datasets. The classifier successfully map the region, achieving a high overall accuracy of 94% against an independent test dataset (84-96% and 81-95% accuracy for oil palm and rubber respectively). In Chapter 4, the method presented in Chapter 3 is used to identify and estimate the area that is actually planted with oil palm within oil palm concession areas in southern Myanmar. The distinction between plantations and concession areas matter, as plantations have been already deforested and converted to oil palm or rubber. Meanwhile, concessions have been allocated to oil palm production, but have yet to be converted. My results show that only 17% of the total concession areas has so far been planted with oil palm (15%, 75,000 ha) or rubber (2%, 7,800 ha). Furthermore, my analyses show that approximately 25,000 ha of oil palm are planted outside formal concessions. This highlights an urgent need to clearly demarcate and enforce concession boundaries. It also reveals that about 200,000 ha of unconverted forests still exist within oil palm concessions that are at high risk of conversion in the future. Hence, these unconverted forests represent an ideal target for conservation and legal protection. The application of this approach for other regions and crop types could result in substantial protection of forests and carbon stocks. For example, in Kalimantan, Indonesia alone, more than three million ha of intact forests are estimated to lie inside oil palm concessions, mostly with little to no legal protection. It is therefore crucial to understand why some concessions remain unexploited, and to evaluate the possibility of changing the status of these areas to protect the forests. This would not affect current levels of production, yet it could considerably contribute to mitigating climate change. Overall, the methods developed and findings presented in my thesis offer a route for countries to improve their forest protection plans and reference levels. If implemented across the tropics, this approach could significantly aid policy makers in developing and implementing policies that reduce the loss of forest carbon stocks. I conclude that risk-based approaches considering tree location, land use and legal status, rather than narrowly defined forest areas, could offer a more transparent means for forest conservation, and a better route to achieving the overarching objectives of climate change mitigation.
Supervisor: Mitchard, Edward ; Patenaude, Genevieve ; Keane, Aidan Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID:  DOI: Not available
Keywords: deforestation ; carbon emission reduction ; climate change mitigation strategies ; forest protection ; REDD+ ; UNFCCC ; Myanmar ; machine learning algorithms ; oil palm ; rubber ; Shuttle Radar Topography Mission