4.1 Direct applications
4.1.1 Topography
An elevation data base consists of elevation data covering an entire territory. It is created by digitization of a topographic map or direct stereoplotting by photogrammetry (1). Few countries have a data base at a sufficiently large scale made from coverage by aerial photography. It is interesting to describe the characteristics of the contents and precision of DEMs to study what countries that do not have them could do using SPOT pairs.
We will do this by describing their specifications using the example of the IGN BDZ (2) (Illustration 1) in France:
a)
Contents of the BDZ (according to G.SAUMONNEAU, 1987)
· contours at 5, 10 or 40 m intervals depending on the class of slopes and the means used
for their input;
· points with elevations characteristic of the countryside or human activity (summits,
channels, passes, geodetic points, bench marks, etc.);
· a set of definition parameters:
- characteristics of map divisions and of each division,
- a geographic coordinate system to eliminate all connection problems within a mosaic of
mapping divisions (sometimes problems of different projections, areas or time
zones),
(2) The elevation data base made by the IGN for France is called BDZ; Z conventionally represents the third dimension.
| · a set of utilities for: - base administration, - selective extractions, - intermittent type of elevation calculations along a profile (intersections of contours with a segment), on a regular grid, etc., - graphic outputs.
|
(1) The density of points on each curve must be such that it leads to a smooth vector representation of the curves in plan, without creating an excessive number of vertices in curve segments. |
| The large number of data sources used to input information makes it
impossible to quantify an absolute uniform precision for the entire base. The precision
decreases naturally as a function of the input method as follows: · digital photogrammetric input; · scanning; · manual input. Top of the page |
(1) Rectified orthoimages, mosaics or space maps taking account of the
relief. (2) For example, SPOT image levels 2A and 2B. (3) Like level 2 in SPOT images. (4) And its consequences: |
|
|
||
Figure 24: Influence of relief on plan and geometric distorsion of images |
||
Simultaneous use of a DEM during rectification can significantly improve the quality of the geometric correction due to recalibrations taking account of local elevation differences. These are calculated by interpolation starting from the elevations of nodes in the DEM. Therefore, the following products can be supplied to users: · the orthoimage: recreation of a vertical view using digital or photographic means. This is a projection or an orthogonal perspective, corrected for: - the effect of relief, without residual parallax, - all panoramic distorsion, without any scale change. A stereo-orthoimage consists of an orthoimage and an associated stereo image, defining transverse parallax from which the elevation can easily be derived. It has the two advantages that it is a document that can be superposed on the map (useful for plotting with updating in plan) and a pair which is used in stereoscopy; · a mosaic of several orthoimages without offsets around the boundary area; offsets always hinder interpretation of network continuity, particularly roads and railways, etc.; · the space map (1) consisting of a mosaic of orthoimages divided into pages as a function of the existing mapping, together with annotations (2) necessary for locating and identifying information (road network, hydrography, etc.). Top of the page
|
(1) Please refer to Tutorial A1 for further information. (2) Marginal information consists of insets such as geographic framing and identification elements, contours and characteristic points with elevations, toponymy, miscellaneous interpreted networks, etc. |
| with erosion classes, forest area with accessibility classes, mountainous
area with risks of avalanche, etc.) referring to the inventory concept making use of
semantic attributes attached to the object (Figure 26); · how or why?: search for correlation and classification ( statistical analyses, calculation of distances as a function of the slope, etc.) in order to demonstrate the relations between all objects by doing a spatial analysis based on the topology (figure 27); · when?: description of revisions or changes that occur between two given periods, the role of the "time" factor (freezing of agricultural land as a function of the slope, impact of urbanization in areas with risks of landslides, etc), where relief becomes a controlling variable multiplying produced effects; ; · what if?: representation of projects or making assumptions about changes (flooding simulation as a function of the level difference, influence on the countryside of a motorway diversion or a TGV high speed train line, propagation of pollution in transport corridors, etc.) for prediction purposes (figure 27). |
 |
| Figure 25: Creation of a complex theme The "complex theme" drawing obtained is characterized by a sequence of logical operations (and, or, not, "=", etc.) made on drawings that may or may not be derived from the DEM. In this case we have given preference to simple thresholding on the amplitude of slopes and on a raster drawing of altitudes, although we could make it very complex |
 |
In a second level, we start again from the previously created complex theme drawing. We add a land occupancy map to it, which may be obtained by interpolation of SPOT images, as in the IGN BDCarto map. This gives a definition of a "status map" identifying types of crops only in a type of countryside analysis (See CORINE Land Cover/Land Use type studies) |
Figure 26: Status map |
|
 |
We can immediately see the importance of the DEM in the reasoning which
led to the result. It was used: · to provide a map of slopes (calculation of their amplitude), · to study a floodable area theme (thresholding of elevations) · to specify the danger of an avalanche (thresholding of slopes and study of second order curves) |
Figure 27: Decision map |
|
| Data quality The intrinsic qualities of satellite data that enable use of the third dimension are evaluated according to five criteria (classified for the purposes of this document only): 1 - The precision directly related to stereoscopy and the geometric resolution obtained by use of space-triangulation modeling (and therefore indirectly to the orbitography of satellites and the geometry of sensor modules). It confirms that the highest resolution available satellite images give topographic mapping standards of 1:50 000 (standard deviation 15 m, namely 90 % of points closer than 0.5 mm, or 25 m on the ground). 2 - The spatial and temporal uniformity of the covered surface area, and planimetric precision standard satisfied by the data. This is essential in domains in which synoptic vision combined with relief vision has an overriding importance: for example geological studies (Smithsonian Institute, University of Nevada, ITC, Université Curie). 3 - Repetitiveness, linked to the possibility of setting points, contemporary criterion but which depends on cloud cover and programming constraints. This progress is essential to make fast studies of quickly changing phenomena such as forest fires, landslides, etc. 4 - The possibility of obtaining stereoscopy and oblique images (sometimes with innovative or spectacular specific advantages) for oceanography (study of waves) or analysis of the countryside (study of stereoradiometry of vegetation cover, studies of vineyards and orchards, land registry and archeological applications). 5 - The identification resolution derived from spatial resolution. Top of the page
|
Measured energies depend entirely on local ground illumination conditions, and therefore cannot be characteristic of a single thematic formation which can be represented by a single class in the map legend. The illumination of a point on the ground depends on: · firstly, the position of the sun and atmospheric conditions; · secondly, the geomorphology of the ground (1), and particularly the topographic position of the point (amplitude and orientation of the slope, which are the main factors controlling the exposure). This demonstrates topographic effects on image radiometry. The most frequent assumption for the calibration of SPOT images is Lambert ground (2), but the reflectance actually depends also on sighting conditions (bidirectional reflectance concepts). When images have been calibrated, the difference between the two views at different angles demonstrates these stereoradiometry effects which are sensitive in the three SPOT channels but particularly in the red channel [H.LE MEN, 1987]. They cannot be neglected when doing multitemporal studies using automatic classification or segmentation methods. A great deal of research is being done and there are many applications in this field; with about twenty correction models proposed during the last 15 years. In general, they can only take account of the effect of the angle of solar incidence (as in the Lambert model), sometimes with diffuse illumination. A more detailed distinction can be made between the following six types of topographic effects: · effect of the angle of solar influence (azimuth and elevation); · effect of shade; · effect of elevation; · effect of diffuse illumination; · effect of the environment; · effect of bidirectional reflectance. Since the relative part of the influence of each effect is directly dependent on the condition of the atmosphere (very difficult to know when taking the photo). An a posteriori estimate of this condition can be obtained using satellite data and their sensitivity to corrections by modeling the six effects mentioned. When the atmosphere is clear, the visible domain is particularly sensitive to the three predominant effects, namely the angle of solar incidence, shade and elevation. On the other hand, the close infrared is less sensitive to the effect of elevation and
more sensitive to the effect of bidirectional reflectance. |
(1) See detailed explanations later in 4.2.1. (2) Lambert ground is a surface with a uniform diffusion spectrum and no preferred direction (see Tutorial B1: Physical Principles of Remote Sensing). |
