DEVISE
Dune Vegetation Identification from Satellite high-resolution images
the vision of DEVISE
Recent improvements in spectral and spatial resolution of satellite imagery open new and exciting prospects for large-scale environmental monitoring. Still this potential is largely unused in dune ecogeomorphology, due to the challenges related with the small size and density of dune plants and the complexity and heterogeneity of the existing species. Machine learning techniques and subpixel classification methodologies, like the Random Forest Soft Classification (RFSC), have shown promising results in similarly challenging environments in terms of plant size and heterogeneity, with high accuracies in subpixel fractional abundance of marsh-vegetation species. Even though subpixel classification could improve monitoring biodiversity from satellite imagery, similar approaches have never been tested for dune environments.
These challenges and gaps inspired the DEVISE exploratory project, built around the idea of testing subpixel classification methods for dune plant species identification using high-resolution satellite imagery. Building on the demonstrated capacity of RFSC to identify plant species distribution in marsh environments, we plan to transfer and adapt the methodology to the more highly mixed and challenging environment of Mediterranean coastal dunes, and more specifically to the barrier system of Ria Formosa, a wetland in South Portugal with high ecological and socio-economic significance (Natural Park (1987), Ramsar and Natura 2000 site). Two data-collection campaigns will be performed, corresponding to low and high plant growth phases (autumn and late-spring), including: a) on-demand acquisition of high-resolution WorldView 2 imagery and b) extensive fieldwork on plant reflectance measurement and species mapping.
objectives of the exploratory project
Test hypothesis
Test initial scientific hypothesis that RFSC methods can be successfully used to identify dune plant species from high-resolution satellite imagery.
Optimise classification algorithm
Optimise the subpixel classification methodology, assess its predictive capacity and identify potential limitations.
Improve dune monitoring
Improve current capacity in monitoring coastal dune vegetation through satellite data.
Duration: 18 months - March 2023 to August 2024
project structure
time and resources management
team communication
administration and reporting
duration: 01/03/2023 - 31/08/2024
(18 months)
T2.1 satellite image acquisition
tasked image correction and preprocessing
past images compilation
T2.2 field data collection training/validation
datasets: plant mapping & radiometry
past plant mappings
duration: 01/04/2023 - 30/05/2024
(14 months)
T3.1 algorithm preparation
adaptation and testing to species classes
test with past data and troubleshooting
T3.2 classification & performance evaluation
tasked image classification goodness of fit indicators
error assessment
duration: 01/03/2023 - 31/07/2024
(17 months)
dissemination of results
scientific publications
communications in conferences
duration: 01/11/2023 - 31/08/2024
(10 months)