Igol version 3

4.11.Topography

A vast range of land-related applications need height information, as highlighted in this report and also by other IGOS themes (http://ioc.unesco.org/igospartners/docsTHEM.htm) such as Geohazards, Water and Cryosphere. Topographic information derives from ground or remotely acquired data processed with different procedures, depending on the original acquisition device (tying to existing geodetic networks, correcting for acquisition times and local errors, reconstructing geometric/geophysical properties of the acquisitions, interpolating, etc). This explains why formats in which topographic information might be expressed can range from height points to contour lines, to TINs, to grids, etc. This section will focus on Digital Elevation Models (DEM), which is the generic term for a representation of heights on a regular grid. Within this category, two sub-categories may be identified depending on the type of surface which is represented: Digital Terrain Model (DTM, a “bare-earth” DEM which refers to the heights of underlying Earth) or Digital Surface Model (DSM, a “top-of-canopy” or “top of buildings” DEM). Availability of information related to both surfaces allows computation of relative heights and therefore volumes, and is of particular interest to urban applications.

Currently available global topographic datasets have coarse resolution (90 m to 1Km) and were derived from spaceborne observations (SRTM), from ground observations (GTOPO30) or from a mixture of ground, airborne and satellite observations (GLOBE or ACE, the latter resulting from the combination of spaceborne altimeter data with GTOPO inputs). Finer resolution datasets already exist, for instance acquired by SRTM (C band at 30m resolution), but they have not been released worldwide. DEMs derived from the DLR/ASI X band radar onboard SRTM (30m resolution) also exist, but they do not uniformly cover the Earth.

4.11.1.Observation needs and technical requirements

Many calls have been made for improved topographic maps of the world such as the “Global Map” at a scale of 1:1 million proposed by the International Steering Committee for Global Mapping with an effective resolution of 1 km. Standards dealing with quality and accuracy of topographic datasets already exist (e.g. DTED-2 defined from NIMA (now NGA) has a post spacing of one arc second -approximately 30 meters), and a global, topographic dataset with such type of resolution would be welcomed by the IGOL community. This would in fact provide natural resources managers and decision makers, modellers and scientists with information well suited to their standard scale of work. In addition, many local topographic datasets might be improved by such global dataset, as it already happened in the past for regional DEMs, when global ACE helped correcting discrepancies. If this data set can not be released then other sources for the creation of a 30m spacing data set, such as the use of ASTER data should be explored.

It has to be noted however that in areas of low topography such as coastal zones and flood plain areas, there is a need for topographic information with much higher resolution (in the order of 1 m), because of the impact of small topographic variations on the likelihood of flooding.

Besides resolution/accuracy issues, temporal requirements for updates and delivery may vary. “Recent” information is requested from the modelling, planning and management communities (e.g. floodplain management, environmental and microclimate studies), whereas “up-to-date” information, instantaneously delivered, is necessary for disaster and relief management, mapping for humanitarian aid or market-related applications. Products directly derived from topographic datasets, are relevant for hydrologic parameters extraction, soil information systems as well as for geomorphologic analysis.

In addition, a key issue for all the land sub-themes, is the additional benefit offered by availability of distributed and accurate topographic information. This enables orthorectification of remotely sensed data, hence facilitating geographic intercomparison of various data sets and creation of products which do not need additional processing for ingestion in GIS or data fusion.

4.11.2.Current status

Current spaceborne sources of information rely on techniques exploiting optical as well as radar sensors. The principal characteristic of all such techniques is that they tend to provide information about top of canopy surface and additional processing work and data might be needed to derive information about bare-earth heights. The key-advantages reside in the repeatability of systematic observations (to generate up-to date datasets) and the independence of acquisitions from ground conditions (such as political or accessibility issues). Currently flying optical sensors for high-resolution stereo-mapping (posting of less than 10m) include SPOT5 or IKONOS which are extremely expensive, and PRISM onboard ALOS, the availability of which may be linked to data-policies issues. The key limitation of optical sensors (including also ASTER) resides on cloud cover, which may drastically affect availability of up-to-date information in tropical and subtropical areas. Radar data acquired by ERS, J-ERS1, Radarsat-1, SRTM, Envisat or ALOS can also provide DEMs, exploiting different techniques. In the case of InSAR, performances depend on geometry constraints, surface conditions or atmospheric artefacts. In addition, C and especially L-band measurements refer to an average surface defined by the first cm. of ground (or canopy), not to the top surface itself. the global SRTM data at 90m spacing is available only for 80% of the land surface, its geographic coverage extending from 60 degrees north to 56 degrees south.

On the other hand, ground-based observations may provide data for DTM (especially when based on ground surveys, e.g. leveling or GPS) or DSM (when based on airborne datasets such as radar or optical stereo-pairs or Lidar). The main issues in the case of ground-based observations reside in the cost of ground-surveys, maintenance of networks as well as accessibility and weather conditions.

4.11.3.Current plans

Radarsat-2, to be launched in late 2007, will collect very fine to fine resolution C-band SAR data. Spaceborne acquisitions of X-band data are available from Terrasar-X and will be available from the CosmoSkymed satellite constellations.

4.11.4.Necessary improvements and major gaps

There is a need to extend the coverage of the existing 90m SRTM datasets, for instance by integrating it with other existing datasets (e.g. SPOT-5 or ERS acquisitions in Tandem mode). A major gap is then represented by the lack of a medium scale global dataset, publicly available: SRTM data exist at suitable resolution over 80% of the land areas, but US data policy constraints limit their availability.

For areas of low topography such as coastal zones and flood plains, there is a need for topographic information with much higher resolution (in the order of 1 m). Local, high quality, standardized topographic information should be made available, to facilitate validation efforts of global datasets.

In addition, to facilitate the combination of elevation data from different sources and to incorporate the elevation data into other information and products, a common geodetic reference frame is required. The adoption and implementation of a global datum based on the ITRF (or WGS 84) is probably needed.

One of the important uses of topographic data is to allow the orthorectification of satellite data as was carried for the Landsat GeoCover products. For data fusion and activities such as change detection this is extremely beneficial. It is recommended that all medium and finer resolution remote sensing data is routinely made available as orthorectified products.

4.11.5.Product-specific critical issues

The SRTM data base is only openly available for the US at 30m horizontal resolution but exists for all areas for which there is SRTM coverage (58 degrees N and 60 degrees south). It is recommended that these extremely valuable data be made openly available for all areas. If not other sources (e.g ASTER) to derive such a dataset should be explored

4.11.6.Principal recommendations

• 
  Improve global coverage of 90m SRTM datasets (possibly by integration with SPOT-5 or ERS acquisitions in tandem mode).
  
• 
  Ensure public availability of the 30m spacing DEM.
  
• 
  Provide very high resolution (1m) topographic information for low/flood prone areas
  
• 
  Distribute data using a common geodetic reference frame.
  
• 
  Provide terrestrial remote sensing data in an orthorectified form so far as possible.