While some semblance of the river channels will be present in the elevation data – often enough to infer their location – the dual effect of grid resolution and failure of the IfSAR to penetrate the water surface mean we cannot garner how deep the rivers are.
Spinning off from the collection of elevation data is the derivation of global hydrography data: that is, maps of the locations of the world’s rivers. In the Argentinian example in Figure 3, corrected striping and radar speckle have produced a flood map of much greater fidelity. Clear artefacts – the erroneous striping effect – present in previous versions along the Mekong River in Vietnam have been removed, resulting in much more realistic flood extents.Ĭomparing flood simulations using original SRTM data and MERIT DEM evidences this further. You can see from Figure 1 the effect of this new dataset on floodplain. A DEM with meritįathom scientists aided in the generation of the most recent and accurate version of global elevation data – MERIT DEM – led by Dai Yamazaki at the University of Tokyo. Corrected versions of SRTM remain the best source of elevation data for global flood modelling. While newer global datasets have been collected from spaceborne instruments since 2000, they are either commercially privileged or have not benefited from the same level of error correction as SRTM. This is dampened somewhat by aggregating to three arc seconds (~90m) resolution, where vertical errors are smoothed via averaging. The final consideration made by scientists is the raw vertical accuracy of the radar instrument, which can be the order of metres. Noise inducing radar speckle, no-data voids, and systematic ‘striping’ errors caused by the pitch and yaw of the shuttle are characteristic of raw SRTM data, but innovative algorithms have been devised to correct these and generate a “bare earth” digital terrain model (DTM). On top of urban and vegetation biases, a number of other errors have been ironed out. The SRTMDSM has undergone extensive corrections thanks to the global scientific community making it fit-for-purpose in the context of flood inundation modelling. The upshot of this is that vegetated (particularly, forested) and urban areas appear as hills in the elevation data and would never flood – providing a useless tool for managing flood risk.
Further, the grid spacing of the raw digital surface model (DSM) SRTM captured is one arc second (~30m at the equator) – while this is granular at the global scale, it is too coarse a resolution to capture the height difference between the ground surface and the tops of buildings and vegetation. However, SRTM collected is the canopies of trees and the roofs of buildings –neither of which a flood modeller is particularly interested in. We know that water flows downhill, but a flood model is of little use if it does not know what ‘downhill’ is! Transformative strides in the collation of data describing the elevation of the Earth’s surface were made in 2000 with the launch of the Shuttle Radar Topography Mission (SRTM), which consisted of an Interferometric Synthetic Aperture Radar (IfSAR) system mounted on NASA’s space shuttle Endeavour. The most important part of a flood inundation model is the quality of the elevation data over which it simulates water movement. Two things conveniently fell into place to make global flood modelling a reality: worldwide terrain and other geospatial data became available at high-resolution, and computers became fast enough to handle the calculation of water-flow equations over the tens of billions of grid cells this geospatial data constitutes.
The ability to map flood zones globally – that is, delineate areas of land susceptible to inundation when rivers burst their banks – has been revolutionised in the past decade. Satellite Positioning, Navigation & Timing (PNT).