NEWS
SAEON Fynbos Node launches the Global Overberg Renosterveld Watch application
Overberg Renosterveld Conservation Trust
The Renosterveld of South Africa ’ s Overberg is one of the world ’ s most species diverse mediterranean type shrublands , but it is also one of the most threatened ecosystems worldwide .
Only around 5 % of the original extent of this vegetation remains . One of the main threats is illegal ploughing due to expanding agricultural lands into the surrounding Renosterveld .
Research by Glenn Moncrieff and colleagues at the
Fynbos Node of the South African Environmental Observation Network ( SAEON ) found that a total of 478.6 hectares of Renosterveld vegetation was lost to illegal ploughing from 2016 to 2020 , representing a total of 0.72 % of the remaining natural vegetation in the region .
With such a significant loss of Renosterveld over such a short time , an effective monitoring solution that can detect potential land cover change events such as illegal ploughing is needed .
Using a combination of satellite imagery and machine learning , Glenn Moncrieff and Marcel Gietzmann-Sanders have developed the Global Overberg Renosterveld Watch application , which starts to address this urgent need for an effective remote sensing solution that can be used to detect changes so that resources can be assigned as needed to start to tackle the problem on the ground .
The technology is developed based on the
Global Forest Watch application , which uses Google Earth Engine and satellite imagery to document deforestation in Brazil . The satellite imagery which is used is now globally available , documenting land cover change around the world over several decades .
With further work and the application of machine learning to build algorhythms that teach the software to identify the differences between Renosterveld vegetation and transformed agricultural land , this technique can now be used to quickly identify illegal ploughing events within days of their occurrence .
Whenever new satellite imagery becomes available , these models can then be run to automatically highlight areas where land cover change may have taken place . At present , these models are being run every five days . These changes in Renosterveld vegetation can even be detected prior to complete transformation has taken place . Often in areas where illegal ploughing is about to take place , the veld is brush cut .
This means that even these changes can be detected prior to the area being ploughed , which means that areas , where interventions are needed , can
Figure 1 . These models are being run every five days and can highlight areas where land cover change may have taken place . It is available online via cloud services .
Grassroots Vol 21 No 4 December 2021 36