Water Observations from Space Help
The Water Observations from Space (WOfS) viewer provides a summary of surface water observations across Australia overlayed on a Google Earth background, which provides basic location search functionality.
To zoom to a location
- Type its name in the Enter location name box at the top-left of the viewer. Suggestions of which location you are looking for will appear.
- Click a suggested location to zoom to the location. The location will be indicated with a yellow diamond place icon.
The information displayed over the Google Earth background is the WOfS surface water summary. This shows how often water was seen across Australia as a percentage of how often the Landsat satellites were able to clearly see the Earth at each point. An area is considered to have been "clearly seen" if the satellite sensor was working correctly, and the location was not covered by cloud or shadows at the time of observation.
WOfS is shown in a colour scheme indicating frequently observed water (high percentages) in purple and blue, down through greens to infrequently observed water (low percentages) in yellows, and finally to very low percentages in red.
This means that water that is often found in a location, such as in a dam or lake, will appear purple or blue. Water that is found in areas that might dry up or in areas that the satellite can't always see (such as narrow river channels), will be in the green shades. Water that is found in an area very rarely (such as from flooding) will show up in red colours, with the biggest, rarest events as the darkest red.
Occasionally the detection system mistakes shadows from clouds and hills as water, so there are some errors where a shadow is mistakenly detected as water, that also appears in red.
The best way to use WOfS
- Determine how often water was seen in a location. Red and yellow colours are where water was seen the least often, and so contain the most errors. Green and blue results are where water is seen often, and so contain the least errors.
- Compare the result for a location with the surrounding area, to get some context from other water features. If the shown area is related to a river, lake or other known water feature, then the displayed result is less likely to contain errors.
- Click on a point location to get more information, describing the actual number of times water was seen, how often the area was seen clearly and how confident we are in the result. More detail on understanding the point location results is given below.
WOfS works poorly in dense urban areas such as cities, where the medium resolution of the Landsat satellites and frequent shadows from large buildings creates errors in water detection.
WOfS also contains more errors in areas where clouds are more frequent, such as along mountainous and coastal areas. This is due to the inability for an automated process to correct for shadows 100% of the time. Occasionally cloud and terrain shadows will be mistaken for water and appear in the WOfS result.
WOfS works best on large, open flood plains such as along Australia's inland rivers.
WOfS point location results
More information about a point location can be obtained by left-clicking on the point in the viewer. This brings up a results "bubble" that provides four pieces of information:
- Water observations: How many times was water actually seen at this point.
- Clear observations: How many times could the satellite actually see this point clearly, without problems from clouds, shadows or a sensor malfunction.
- Water summary: The ratio or water observations to clear observations, as a percentage of time that water was seen at the point.
- Confidence: How likely is it that the water observations at the point are correct.
What is confidence?
The confidence probability is a measure of how confident we are that a location flagged as water is truly water. The confidence information was created by comparing the WOfS with other data, to see how well it matched known water areas from other sources. These sources are:
- Multi-resolution Valley-Bottom-Flatness (MrVBF). This uses the terrain across Australia to work out whether an area is likely to be on or near a valley bottom or flat area where water could accumulate.
- MODIS Open Water Likelihood (MODIS OWL). This is a product similar to WOfS, derived from the low resolution MODIS satellites, showing the likelihood that an area has contained water between 2000 and 2012.
- Australian Hydrological Geospatial Fabric (Geofabric). This is a spatial database of Australian water features, including rivers, lakes, flood plains and other water features.
- Terrain slope. This is a second factor created from the terrain. The steeper the terrain, the less likely that water will accumulate in the location.
- Built-up urban areas. The way light is reflected from large buildings and their shadows causes considerable water detection errors in cities. The more built-up an area, the less likely that water will be correctly detected there.
- Observation frequency. Areas where the observation frequency is low typically have poor conditions for water detection. The higher the observation frequency, the more likely that a good water/non-water classification can occur.
The higher the confidence at a location, the more the water result correlates with the confidence factors. However the confidence factors also contain errors, and hence can produce an overstated confidence in some areas such as cities, where the confidence factors are not as useful.