Upscaling Blue Carbon: Harnessing Earth Observation and machine learning to finance seagrass restoration

Abstract

We are in the midst of a climate and biodiversity crisis. Upscaling nature-based solutions could cut emissions by over a third of what is required to meet Paris Agreement targets yet funding for nature-negative activities continues to surpass investment into nature protection and restoration by 140 times. Earth Observation and machine learning methods are considered promising tools to support the sustainable and responsible finance of nature. Their potential is explored here with a focus on seagrass ecosystems as a natural climate solution. This thesis takes an interdisciplinary approach lying at the intersection of data science, environmental science, finance and responsible innovation. Firstly, it explores the challenges and barriers in the use of Earth Observation data to support sustainable financial decision making. A baseline of needs across stakeholders is provided and five challenges to the uptake of Earth Observation data are identified. From these insights, responsible innovation principles are proposed to support responsible uptake within the sector. Secondly, a novel data-driven method is developed using Earth Observation and machine learning to predict carbon stocks in Z. marina seagrass beds. These findings provide improved methods for predicting carbon stocks in seagrass at scale that could be used to inform and upscale investment in seagrass conservation and restoration decision-making. The application of this model to proposed areas of seagrass restoration within south-west UK, demonstrates significant seagrass restoration potential that could contribute to climate mitigation activity in the UK. Such a tool is shown to have relevance for future financing to upscale seagrass restoration, although crucial evidence gaps remain making large scale financed restoration of seagrass nascent but not yet scalable. Ultimately this thesis demonstrates the value and opportunity of Earth Observation data to quantify ecosystem services in blue carbon systems and the importance of taking an interdisciplinary approach to ensure the responsible development of technology to support nature restoration.