Abstract
Salt marshes, a coastal wetland habitat found in temperate climates, have experienced a shift in perception over the past century. This once marginalised habitat is now highly valued for its rich biodiversity, in addition to the many ecosystem benefits it provides. This increased value has led to widespread policies of protection and restoration. These policies create a requirement for monitoring. However, traditional in situ monitoring is highly time and resource intensive, while also challenging in this highly protected and hazardous habitat. This has led to many monitoring schemes falling short of advocated standards, in turn resulting in a lack of data regarding the status of this habitat. Consequently, novel techniques are required to improve monitoring practices. One technique with the potential to address this data sparsity is remote sensing, which involves collecting data at a distance (remotely) using sensors mounted on platforms such as drones and satellites. In addition, the data collected from these sensors can be combined with machine learning algorithms to provide automated monitoring. There have been numerous coastal wetland assessment studies that have illustrated the effectiveness of this technology. How-ever, to date there has been no review comparing this technology with in situ monitoring across remote sensing platforms, nor on how in situ monitoring may be integrated with drone surveys and satellite data to provide insights into marsh condition. This study aims to address these knowledge gaps.In this study, species composition was monitored in situ across an established and a newly restored marsh in the Adur Estuary. In addition, drone surveys were completed to map vegetation communities across the site using machine learning. Drone surveys were also completed in two additional sites in southeast England to assess the wider applicability of the developed methods. Finally, the outputs from the drone models were used to train further machine learning models for satellite data as a step towards larger, regional-scale mapping of marsh zonation across southeast England. This study finds that remote sensing technology can provide very effective data for monitoring and should be incorporated into monitoring practices to facilitate low-cost, repeatable and spatially comprehensive surveying. However, trade-offs exist between scale and data resolution which must be considered, with larger scales resulting in coarser and lower accuracy data.
| Date of Award | Jun 2025 |
|---|---|
| Original language | English |
| Awarding Institution |
|
| Supervisor | Corina Ciocan (Supervisor), Sarah Purnell (Supervisor) & Raymond Ward (Supervisor) |