5.5. Elevation Data
When a position in a monoscopic image is used to find the corresponding ground position, elevation or height or surface shape data is often used. Elevation data is thus an input to certain
services. Such elevation data could take one of several forms, including: 1.
Single elevation value, to be used for one or more image positions 2.
List of elevation values, to be used with a corresponding list of image positions 3.
Elevation coverage, that defines the elevation as a function of ground position, to be used for one or more image positions
Each elevation could be transferred as one double precision floating point value. Note: The SRSs used for elevations are listed as separate service inputs and outputs.
The abstract model of Image Coordinate Transformation Services specified in Section 4.1 includes interfaces to surface shape or elevation data. These interfaces produce a 3-D ground position that is
the intersection of an image ray with the specified surface shape model. The ElevationModel class specified in Section 4.1.18 is expected to use an OGC Coverage to contain elevation data, and to
provide access to that data.
5.6. Elevation Error Estimates
When elevation is used and output accuracy is needed, elevation accuracy data is a needed input to certain services. The accuracy of a single elevation value, or of all elevations in a list, could be
specified by a one single precision floating point number. This value could have one of several meanings, including variance, standard deviation, or LE Linear Error. A LE value could use one
of several confidence probabilities. However, for consistency with using a covariance matrix to specify the accuracy of two or three dimensional coordinates, a covariance matrix or variance value
should be used for elevation value accuracy.
Since elevation data is used to determine a 3-D ground position that is the intersection of an image ray with the specified surface shape model, the accuracy of this position is needed when output
accuracy data is required. Well known structures for position accuracy are defined in the Accuracy Measure UML package that is specified in Section 4.4 of Abstract Specification Topic 0. This
Accuracy Measure package defines well known structures for variances and for covariance matrices.
When elevation is specified by an elevation coverage, the effect of horizontal position errors on the elevation value error should be represented. This can be done using a partial 3-D covariance matrix,
with a special interpretation of the values in off-diagonal cells. These off-diagonal cells can contain the ratio of the covariance value of that cell to the unknown variance of the corresponding
horizontal axis. The variance cell for the elevation would have the normal meaning. The other diagonal cells in the covariance matrix, for the two horizontal coordinates, would have unknown
values. The off-diagonal cells for the covariances between the horizontal coordinates would also be have unknown values.
Using this approach, elevation accuracy data types can be defined. Such a set defined using ISO standard IDL data types and structures is:
Type: Matrix Cell, of a covariance matrix struct MatrixCell {
string 2 axes; Axes of covariance matrix float value; Units: Meters squared
}; Type: Elevation Covariances
typedef sequence MatrixCell, 3 ElevationCovariances; Covariance matrix cells included only when correct
value is known Required values of “axes” string: ZZ
Optional values of “axes” string: XZ, YZ Where X, Y, and Z stand for three ground coordinates
The Open GIS Abstract Specification Page 47
Volume 16: Image Coordinate Transformation Services 00-116.doc
Matrix cell XZ contains the ratio of XZ to XX Matrix cell YZ contains the ratio of YZ to YY
5.7. Desired Image Section