Detecting aliens from space
Biological invasions are responsible for some of the most devastating impacts on the world’s ecosystems. Freshwater ecosystems are among the worst affected, with biological invasion posing serious threats not only to freshwater biodiversity, but to the important ecosystem services it provides. Tackling the impact of these invaders on our freshwater ecosystems is currently a major challenge, and one that requires detailed information on both invasive species distribution and spread. While these concerns are relevant globally, there is an urgent need to conduct research of this nature in countries such as South Africa, where freshwater resources are becoming increasingly pressured. This project utilises freely available satellite data and the processing capabilities of cloud computing platforms to map the distribution of invasive aquatic alien plant species across South African waterbodies and explore reasons for their spread. PhD student Geethen Singh is undertaking this valuable research in collaboration with Chevonne Reynolds, Benjamin Rosman and Marcus Byrne of the University of the Witwatersrand).
Water Hyacinth infestation at Hartebeespoort Dam in Johannesburg (a); Hartebeesport Dam viewed from an over passing satellite (b); Accurate mapping of the entire waterbody and used to limit the classification of vegetation to only the dam (c); Accurate mapping of the moving Water Hyacinth mat on Hartebeesport Dam (d)
South Africa’s surface water is frequently imaged by overpassing satellites. These valuable images can track water level changes from national extents down to individual dams. This provides scientists and water managers with near real-time monitoring of water resources. Invasive aquatic alien plant species present a serious threat to South Africa’s freshwater resources as they increase water loss through evapotranspiration. Water weeds can increase water loss by 130–180%. Water body area and the percentage cover of invasive aquatic weeds on these impoundments are crucial information needed to facilitate the removal and management of these water weeds. However, field assessments to obtain this information are often too time-consuming, costly and labour intensive. Freely available satellite images are an alternative and cost-effective approach to reliably and frequently estimate the percentage cover of water weeds on surface water. In this way, the extent of water weed infestation and its associated drivers can be identified and tracked.
Once an invasive alien species has established itself it can be difficult or even impossible to stop or slow down the invasion. Thus, the early and rapid detection of invasive alien species is needed to make management targeted, feasible and effective. Consequently, there is an urgent need for techniques that enable consistent, fast and accurate monitoring. This will allow invasion biologists the unprecedented capability to systematically monitor invasive alien species over the introduction, establishment and secondary dispersal phases of the invasion process, across large areas with a reduced risk of accidental invasive alien plant dispersal by field personnel. Geethen is making use of semi-automated satellite image analysis techniques, such as semantic segmentation of freely available Landsat and Sentinel-2 imagery to map the distribution of five prevalent invasive aquatic plants at a national level. Furthermore, he is undertaking this analysis within Google Earth Engine, a cloud computing platform that allows scientists and managers the ability to produce up-to-date maps of invasive species distribution and percentage cover on a frequent basis. Geethen is also working towards an application so that these products will be readily available to managers and decision makers when they require them. Over the last year, Geethen has made excellent progress in discriminating between our top five worst aquatic invaders; Water Hyacinth, Water Lettuce, Parrots Feather, Kariba Weed and Red Water Fern. In collaboration with the Centre for Biological Control at Rhodes University, who provided valuable ground-truthing data for these species, Geethen has been able to produce accurate and robust predictions of aquatic vegetation and its subsequent discrimination across South Africa. In the coming months Geethen will be investigating how other satellite-derived data products can provide information on biotic and abiotic drivers of these alien invasive species distributions. This will include using satellite telemetry data of southern African waterbirds to assess the contribution of this little considered vector in invasive species spread. With the ability to identify the dominant drivers within regions comes the ability for managers to tailor more effective strategies to combat invasive populations.
Activities in 2019
- Geethen began prepping a manuscript on the distribution of invasive aquatic alien species.
- Geethen and Chevonne attended the National Symposium on Biological Invasion in Cape Town and a Google Earth Engine workshop in Stellenbosch.
- Geethen attended a workshop on Synthetic Aperture Radar (SAR) at Wits rural facility and a Deep Learning Indaba in Durban.
- Geethen organised and presented three tutorials and two workshops based on the analysis of geospatial data for staff and students of Wits University.
- Geethen upgraded his MSc research to PhD.
- He has been able to map and discriminate aquatic vegetation at state-of -the-art accuracies for five invasive aquatic alien plants across the largest extent (national scale) for the longest period (5 years).
- Geethen has been approached to develop and teach geospatial techniques at Wits University.
DST-NRF CoE grant.
Research team 2019
Dr Chevonne Reynolds (APES, Wits)
Dr Benjamin Rosman (Applied Maths, Wits)
Prof. Marcus Byrne (APES, Wits)
Student: Geethen Singh (PhD, Wits)