Periodic Reporting for period 1 - SAV-EO (Land Degradation Neutrality: Towards an Earth Observation-based Early Warning System for Savannah Degradation)
Okres sprawozdawczy: 2022-01-01 do 2023-12-31
Drylands cover approximately 40% of the land surface of the Earth and support roughly one-third of the global human population. The savannah biome in particular covers 50% of the African continent and 20% of the global land surface, while also representing around a third of terrestrial net primary production, and comprising a critical regulating component of the land carbon sink. Over the last decades, savannahs have been under pressure from human activities, exacerbated by climate change, with dramatic shifts in vegetation distribution and, consequently, alterations of their function, threatening the ecosystem services provided to some of the continent’s most vulnerable populations. Therefore, monitoring the process and evolution of land degradation in these ecosystems is considered of great importance for the scientific community, regional and national governments, as well as international organizations and institutions, such as the United Nations Convention to Combat Decertification (UNCCD).
Land degradation in savannahs has often been associated with the encroachment or densification of its woody component at the expense of grasses. Savannahs in Africa are vital ecosystems that support millions of pastoralists whose livelihoods are increasingly threatened by woody encroachment. Monitoring the woody component of savannah land cover is, therefore, needed to enhance the understanding of broad-scale changes in savannahs and their relationship with ecosystem degradation. Over large scales, this can only be achieved in a cost-effective manner using Earth Observation (EO) technologies. The development of the Horizon 2020 SAV-EO project employed novel remote sensing methodologies based on freely-available EO optical and radar satellite data for the mapping and monitoring of fractional woody cover over South African savannah regions.
The specific objectives & final contributions of the SAV-EO project were:
1. Development of a pipeline that monitors bush encroachment over South African (SA) savannah environments through accurately mapping of fractional woody vegetation cover & its progress across the years.
2. Development of a pipeline that monitors bush encroachment at the species level over SA savannah environments through accurately mapping of fractional woody vegetation cover at the species level.
3. Development of a savannah degradation monitoring web-tool assessing vegetation maps & contributors of degradation processes in SA savannah drylands.
4. Development of an intercommunication basis between EO scientist & local stakeholders & population in SA savannah regions in order to develop & update the derived monitoring tools targeted to their current needs.
Land degradation in savannahs has often been associated with the encroachment or densification of its woody component at the expense of grasses. Savannahs in Africa are vital ecosystems that support millions of pastoralists whose livelihoods are increasingly threatened by woody encroachment. Monitoring the woody component of savannah land cover is, therefore, needed to enhance the understanding of broad-scale changes in savannahs and their relationship with ecosystem degradation. Over large scales, this can only be achieved in a cost-effective manner using Earth Observation (EO) technologies. The development of the Horizon 2020 SAV-EO project employed novel remote sensing methodologies based on freely-available EO optical and radar satellite data for the mapping and monitoring of fractional woody cover over South African savannah regions.
The specific objectives & final contributions of the SAV-EO project were:
1. Development of a pipeline that monitors bush encroachment over South African (SA) savannah environments through accurately mapping of fractional woody vegetation cover & its progress across the years.
2. Development of a pipeline that monitors bush encroachment at the species level over SA savannah environments through accurately mapping of fractional woody vegetation cover at the species level.
3. Development of a savannah degradation monitoring web-tool assessing vegetation maps & contributors of degradation processes in SA savannah drylands.
4. Development of an intercommunication basis between EO scientist & local stakeholders & population in SA savannah regions in order to develop & update the derived monitoring tools targeted to their current needs.
1. Monitoring and assessment of fractional woody vegetation cover change in SA savannah regions over the last 3 decades:
We employed spectral variability metrics of Landsat imagery spanning 32 years with the following specific objectives and scientific achievements and results:
i. Test the performance of three machine & deep learning models in mapping the fraction of savannah woody vegetation & produce the respective large-scale maps spanning across those periods.
ii. Assess the evolution of the Fractional Woody Cover (FWC) through time using the best performing model outputs & a combination of a linear trend analysis & the BFAST break detection algorithm.
2. Assessing the contribution of optical and SAR data for savannah FWC mapping & Evaluating the importance of main land degradation factors/drivers information:
In the present study, we propelled further research in the direction of FWC mapping in savannah environments at high spatial resolution by the following innovative contributions and results:
i. Provide a thorough assessment of the contribution of optical (Sentinel-2), radar (Sentinel-1) & vast ancillary (bioclimatic, soils, burn frequency, topographic/land use) information
ii. Assess the influence of using different seasons’ data on the regression mapping accuracy
iii. Benchmark the performance of various machine learning (ML) & deep learning (DL) regression techniques
iv. Provide accurate & high quality large-scale FWC maps at high spatial resolution
3. Mapping savannah woody vegetation at the species level with multispecral drone, multispectral Sentinel-2 and hyperspectral EnMAP data:
The developed pipeline innovatively contributes to research goals:
i. Evaluating for the first time the capabilities of the new-generation imaging spectrometer EnMAP for monitoring & mapping savannah FWC at the species level.
ii. Assessing the complementary use of hyperspectral EnMAP & dense multi-temporal spectro-temporal Sentinel-2 data towards the same objective
iii. Assessing the suitability of multi-spectral drone imagery to upscale species level plot/field data to spaceborne scales
4. An easy-to-use web-based platform for free-access to data and results:
Currently being finalized & updated, the web-GIS environment will provide the user community with datasets of FWC maps & trends results, FWC at the species level results & land degradation hot-spot maps including presentation of ancillary information on bioclimatic, soils, burn frequency & topographic/land use datasets, the factors/drivers that influence bush encroachment in the SA savannah regions.
Results from the project were (and are to be) disseminated through a number of peer-reviewed scientific publications in conference presentations, conference proceedings and journal papers. Additionally dissemination and communication of the project activities and results were achieved by a stakeholders’ meeting carried out in June 2023 in Pretoria (SA) as well as with project presentations at three international universities. The project website and the soon to-be-published web-based platform are also parts of the dissemination activities of the project.
We employed spectral variability metrics of Landsat imagery spanning 32 years with the following specific objectives and scientific achievements and results:
i. Test the performance of three machine & deep learning models in mapping the fraction of savannah woody vegetation & produce the respective large-scale maps spanning across those periods.
ii. Assess the evolution of the Fractional Woody Cover (FWC) through time using the best performing model outputs & a combination of a linear trend analysis & the BFAST break detection algorithm.
2. Assessing the contribution of optical and SAR data for savannah FWC mapping & Evaluating the importance of main land degradation factors/drivers information:
In the present study, we propelled further research in the direction of FWC mapping in savannah environments at high spatial resolution by the following innovative contributions and results:
i. Provide a thorough assessment of the contribution of optical (Sentinel-2), radar (Sentinel-1) & vast ancillary (bioclimatic, soils, burn frequency, topographic/land use) information
ii. Assess the influence of using different seasons’ data on the regression mapping accuracy
iii. Benchmark the performance of various machine learning (ML) & deep learning (DL) regression techniques
iv. Provide accurate & high quality large-scale FWC maps at high spatial resolution
3. Mapping savannah woody vegetation at the species level with multispecral drone, multispectral Sentinel-2 and hyperspectral EnMAP data:
The developed pipeline innovatively contributes to research goals:
i. Evaluating for the first time the capabilities of the new-generation imaging spectrometer EnMAP for monitoring & mapping savannah FWC at the species level.
ii. Assessing the complementary use of hyperspectral EnMAP & dense multi-temporal spectro-temporal Sentinel-2 data towards the same objective
iii. Assessing the suitability of multi-spectral drone imagery to upscale species level plot/field data to spaceborne scales
4. An easy-to-use web-based platform for free-access to data and results:
Currently being finalized & updated, the web-GIS environment will provide the user community with datasets of FWC maps & trends results, FWC at the species level results & land degradation hot-spot maps including presentation of ancillary information on bioclimatic, soils, burn frequency & topographic/land use datasets, the factors/drivers that influence bush encroachment in the SA savannah regions.
Results from the project were (and are to be) disseminated through a number of peer-reviewed scientific publications in conference presentations, conference proceedings and journal papers. Additionally dissemination and communication of the project activities and results were achieved by a stakeholders’ meeting carried out in June 2023 in Pretoria (SA) as well as with project presentations at three international universities. The project website and the soon to-be-published web-based platform are also parts of the dissemination activities of the project.
To date, no study has combined multi-sensor & multi-season data, machine & deep learning classification methods & dense time-series analysis to propose an innovative solution for mapping & monitoring savannah woody vegetation cover & discriminate savannah woody vegetation species. The project’s EO-based frameworks could support the UN Sustainable Development Goal (SDG) of reaching Land Degradation Neutrality by 2030, contribute towards the EC ‘The European Green Deal’ priority and also increase European research visibility in the ‘Environment & Combat Climate Change’ field, a priority area of H2020. The project’s results could be significantly useful in the research and scientific communities that address Climate Change impacts in drylands, but also to intergovernmental and governmental policy makers and local stakeholders that will use it as baseline data/tools for the development of sustainable management strategies and for assessing the effects of management decisions. SAV-EO project and derived products and tools can be employed for assessing Land Degradation and apply Land Degradation Neutrality measures on targeted areas, benefiting millions of agriculturalists and pastoralists that live in degraded southern African savannahs. SAV-EO has already developed an intercommunication basis/platform between EO scientist and local stakeholders and population in SA savannah regions through which shares the developed mapping products and land degradation monitoring tools and customizes them based on their current suggestions and needs.