The provision of regular, high-cadence, high-detail water resource maps will support accurate and timeous monitoring of the status of local, regional and national water resources. Water surface extents are automatically generated from imagery recorded by the European Space Agency’ s( ESA) Sentinel2 satellite, which has been operational since June 2015. Sentinel2 provides multiple imaging overpasses per month, minimising information loss due to cloud cover, and supports a minimum water feature mapping size detection capability of between 0.1 and 0.4 ha., depending on water and surrounding landscape characteristics. Each month, a set of three total surface water products are generated that describe:
( a) the long-term maximum surface water extent( since Sentinel2 operational commencement at the end of 2015);
( b) the maximum surface water extent from the preceding 6 months; and( c) the current month’ s total water surface extent. In all cases the mapped water datasets are based on 20-metre raster cells, equivalent to the base resolution of the Sentinel2 imagery used. The mapped total surface water area represents the combined extent of both natural and manmade water features, as observed within each of the assessment timeframes.
The image data modelling procedures that have been developed remove any confusing water identification issues relating to other landscape characteristics, such as dark wildfire burn scars, terrain shadows, and cloud shadows, all of which have similar spectral characteristics to water and can lead to classification confusion and mapping errors.
The GeoTerraImage water identification algorithms have been developed using in excess of 100 000 sample points, distributed across South Africa, that represent a wide range of seasonal and regional water and non-water reference characteristics.
The spectral data associated with these reference samples has been used to derive a complex set of( image classification) rules that can be applied repeatedly to cloud-based satellite image archives. This allows for mapping and monitoring of the total surface water extent across South Africa on a regular and repeatable basis. Since the rules are both generic and standardised, the water information outputs are directly comparable over time and have the same mapping accuracies and information content, resulting in a very reliable monitoring procedure.
Examples of modelled water extents over the Vaal Dam region are illustrated in Figures 1 to 4, which represent the Vaal Dam( and surrounding smaller dams), surface water extents for the months of 12-2015, 09-2016 and 09-2017. These three dates represent pre-drought, mid-drought and postdrought conditions, and clearly show the timedependent fluctuations in Vaal Dam surface water area over this period. If these time-series water extents are combined into a single overlapping illustration( Figure 4), then the fluctuating water levels are highlighted even further.
Note: Figures 1 to 3 have a generic image background, whilst Figure 4 has a hill shaded-terrain backdrop. u
Vaal Dam surface water extent 12-2015.
Vaal
Dam surface water extent 09-2016.
Vaal Dam surface water extent 09-2017.
Combined Vaal Dam water surface extents for 12-2015, 09-2016 and 09-2017. netwroking contributor industry debate environment infrastruture municipalities
Water Sewage & Effluent November / December 2017 15