Much like peat (click here to learn more about Adatos’ peat model), mangroves are extremely important in carbon sequestration. From storing carbon within themselves, to preventing coastal erosion that could release carbon into the atmosphere; from mangrove soils acting as carbon sinks, to providing habitats for various animals and organisms, mangroves play a crucial role in the global carbon cycle. Methods to monitor and protect mangroves are hence of utmost importance, which is where we come in.
Adatos specializes in combining AI with remote sensing, and the advancement of these technologies in recent years has resulted in increasingly reliable digital monitoring, reporting, and verification (DMRV). Specific to mangroves, Adatos uses these technologies to accurately identify and map the extent of mangroves, as well as their health. This is possible due to the unique spectral reflectance profiles of mangroves as captured by satellite imagery, which is distinguishable from other types of vegetation.
Different land cover types have unique spectral reflectance profiles. As seen in this graph analyzing major land cover types within Moreton Bay, Australia, mangroves (solid dark green line) have a visibly distinct spectral reflectance profile. The unique spectral reflectance is even more distinguishable to Adatos’ mangrove prediction models that use AI and high dimensionality datasets to distinguish between mangrove and non-mangrove pixels. Image Source: Kamal, Phinn, & Johansen, 2015
While these unique spectral reflectance profiles may not be easily identifiable to humans, AI is able to use a multitude of datasets to both identify and distinguish between different spectral reflectance profiles. Adatos’ mangrove models assess the large amount of data that is fed into it, to accurately and precisely distinguish different types of pixels, with a particular emphasis on mangrove pixels.
Even comparing between different vegetation types, mangroves still have a distinct spectral reflectance profile. Granted, within mangrove varieties (all italicized legend items) there are variations in the reflectance spectral profiles, but there is still a visibly distinguishable difference between mangroves and non-mangrove vegetation types. Image Source: Zhen, Liao, & Shen (2018)
For mangrove areas, Adatos can train bespoke models to:
Map mangrove extent, for producing mangrove maps and to detect mangrove loss
Identify mangrove health according to custom health indices, for identifying areas of mangrove degradation
These maps can be produced in just a few weeks, and can cover a range of years to track changes in mangroves. This is a stark contrast to traditional MRV methods that usually require months of ground truthing to produce just one mangrove map of the current situation. We have produced these models for areas in Asia and Africa, and our customers have used our mangrove maps as a pre-feasibility assessment, and to identify potential project areas.
Extract of an Adatos mangrove map produced to identify areas of mangrove loss (red areas) between 2010 and 2022. Due to the availability of satellite images, the 2010 map is produced at a 30 meter resolution, while the 2022 map is produced at a 10 meter resolution. Image Source: Adatos
If you would like to know more about our mangrove capabilities and harness the advantages of AI and remote sensing, contact us on LinkedIn. For more nature-based content, please click here.