der Universität der Bundeswehr München
Operational Geodesy Software Packages
Autoren: H. Landau, K. Hehl, B. Eissfeller, G. W. Hein, W. I. Reilly
Universität der Bundeswehr München, Neubiberg, 1988
H. Landau, K. Hehl, B. Eissfeller, G. W. Hein:
W. I. Reilly:
H. Landau, K. Hehl, B. Eissfeller, G. W. Hein:
The Integrated Geodesy Adjustment
Features of the New OPERA 2.4 Software
The OPERA Interactive Control Program OPUS
Gravity Data Analysis Programm CHECKOB
Appendix A: The Graphic Interface
Appendix B: Output Examples
Appendix C: Plot Examples
Appendix D: A Complete Summary of OPERA Control Cards
Appendix E: A Complete Summary of OPERA Error Messages
After T. Krarup developed in 1973 his fundamental idea of integrated data processing, hins integrated geodesy, it took nearly one decade until the algorithm was realized in geodetic software. OPERA 1.0 (Hein and Landau, 1983) produced meanwhile at several institutes results which could destroy most of the criticism brought up in relation th this new way (see e.g. Argeseanu, 1986; Argeseanu and Collier, 1986, 1987; Collier, 1987; Hein et al., 1987; Müller, 1987 - to mention only some papers). Even a conversion of OPERA 1.0 into an IBM PC compatible version was prepared by Hungarian colleagues (Adam, 1987, personal communication). At other places similar (program) developments were carried out (Barzaghi, 1987; Zaiser, 1984). In the past three years new research versions of OPERA were written.
OPERA 1.1 was only a modified version with minor modifications compared to the original OPERA. Herbert Landau started to develop a completely new software system in 1984. From 1984 to 1986 he was working on OPERA versions 2.0 to 2.2. Results of these OPERA versions were published in 1985 (Landau et al., 1985) and 1986 (Landau, 1986). The youngest version 2.4 was prepared in cooperation with the other authors. New techniques were provided for covariance matrix calculation, linear equation solution, trend elimination and plotting facilities. In addition an interactive control program was written for OPERA.
With the OPERA 2.4 software version we feel now that it has reached a state which allows us to document a program which will meet - may be eith minor modifications in the next year - the requirements of a large number of applications ranging from geometrical adjustment of traditional networks via integrated adjustments to the prediction of the gravity field.
The whole software system consists now of three programs:
All the software is written in FORTRAN 77 for Digital Eqipment VAX system under VMS 4.6 control. 371 subroutines with a total of 38247 statements form the whole source code.
In comparison to OPERA 1.0 version the new system is characterized by its user-friendliness. In addition, new observation types are built in, e.g., the program is able to process satellite data coming from the Global Positioning System (GPS) as well as in combination with all other data. User-friendliness is also documented by the fact that computations can be carried out in local reference systems.
In order to increase the accuracy of results OPERA 2.4 can use topographic height information for terrain reductions of gravity-related data. Although for the first sight the principle of using original data in integrated geodesy might be disturbed by reducing data through this procedure as well as through other trend elimination procedures in collocation, one should see this in the light of moving a maximum of physical information into the deterministic part and leaving stochastics as small as possible.
OPERA 2.4 emphasizes with its capabilities in particular also the combination of GPS satellite data with gravity field information. This combination seems to be one of the major adjustment strategies in near future in order to come up with precise horizontal (ellipsoidal) coordinates as well as user-oriented height information which are still the orthometric heights in practice. This could lead to a partial replacement of traditional spirit levelling in future when in the combination of data as stated above precise orthometric heights and geoid heights can be computed simultaneously using both types as data itself. The algorithm works then like an interpolation of the geoidal surface, better: of the reference surface defined by input height information.
There are still other small details in OPERA 2.4 which are superior to its predecessor, e.g., inclusion of scale factors in distances, selection of signals, weak datum fixing, consideration of fast storage and solution algorithms, precomputation of covariance values, etc. The reader can recognize the details in the following paragraphs.
One point was already mentioned shortly. Following the concept of user-friendliness and the trend in all sciences to make computations more transparent through graphics OPERA 2.4 can provide the user with graphical output on screen and/or plotter, e.g., plots about the network geometry, data distribution, covariance functions, digital terrain, etc. This is done through PEGASUS, a software using only the original CALCOMP routines which might be available at any institution.
The following paragraphs present a documentation of the software package.
W. I. Reilly:
The Integrated Geodesy Adjustment
Method of Solution
Formats of Input-Data Files
Description of the Output Listing
The programme system GEONET has its origin in 1980. I had been following my interest in the geometric structure of earth's gravity field by testing the use of least-squares collocation for the interpolation of gravity, and of the higher derivates of the gravity field (Reilly, 1979), with a comprehensive trial of the method in the area of Canterbury, New Zealand (Reilly, 1980a,b, 1981a). To have continued on this path would have been to delve more deeply into what are broadly called "geoid studies".
I was, however, anxious to see a more immediate application of this work on the geometry of the gravity field to a practical geodetic problem, that of network adjustment. This desire coincided with a parallel interest in the determination of earth deformation from geodetic observations, developed so succesfully in New Zealand by Bibby (1973, 1982). It seemed obvious that earth deformation studies would need a precise, physically rigorous adjustment method to extract the maximum of information from a wide range of geodetic observations available (including modern EDM measurements), and that traditional two-dimensional adjustment methods, with subsequent comparison of coordinates, would not be sufficient. What was needed was an expansion of Bibby's method of simultaneous adjustment, with a continous strain-model, to a fully three-dimensional adjustment system which also considered changes in the gravity-field. At the same time, the method should also be available for time-invariant network adjustment, offering a three-dimensional alternative to the two-dimensional methods currently in use.
The first step, in 1980, was to develop the theoretical basis for a three-dimensional network adjustment method (Reilly, 1980c), and to write a corresponding FORTRAN programme for the PDP/11 computer at Geophysics Division, DSIR, Wellington. This programme had as its object the derivation of the geocentric Cartesian coordinates, the astronomic longitude and latitude, and the gravity potential, for each network point, from observation of angles, distances, levelling, and satellite Doppler positioning.
In 1981, this programme was transported to the Technische Hochschule, Darmstadt, and built into a unified programme system on the IBM/370 computer, in a way not possible on the smaller PDP/11. Some results of this stage are given in (Reilly, 1984). In 1982 the programme returned to New Zealand, and the time-variable, three-dimensional strain model was added to it (Reilly, 1979, 1981b, 1982b). It was applied to the analysis of geodetic surveys in southern New Zealand, and a further theoretical development was that of the analysis of heterogeneous strain (Reilly, 1986a), in an attempt to clarify the phenomenon of the variation of strain from point to point. Programming was now on a VAX computer, still using the FORTRAN language. The programme system, going under the generic name of GEONET, was now becoming somewhat unwieldy and hard to follow, having been carpentered together over a period of five years on three disparate computers. In 1985 I therefore undertook a complete "top-down" revision, translating the entire programme into the newly-available VAX-PASCAL, and exploiting the security offered by PASCAL's rigorous type-definition facility to the encoding of multi-dimensional vector geometry.
In 1986, the programme travelled again, this time to the Universität der Bundeswehr München (University FAF Munich), where the first task was to translate VAX-PASCAL into Burroughs-PASCAL to run on the University's Borroughs B7800 computer; a rather lenghtly task, given the differences in each manufactorer's additional features. The programme was then further expanded by the addition of (a) a model of faulting, either continuous or discontinuous in time, (b) a model of time-varying gravity field, as envisaged in the original concept, and the inclusion of the intensity of gravity in the set of coordinate parameters in each station (Reilly, 1985, 1986b). The range of observational data admitted was also extended to include (a) the type of relative vector measurements to be expected from using GPS techniques, and (b) both absolute and differential measurements of the intensity of gravity. (In April 1987 a VAX-PASCAL version was re-translated from the revised Burroughs version.)
It is the PASCAL version of GEONET, current in March 1987, that is the subject of this report (with some notes on VAX modifications in an Appendix). It is hoped that this version will now be stable for some time, though it is rare to find a programme system in an actively developing field that is not subject to revision. The purpose of this report is to provide a guide to enable a qould-be user both to obtain some results, and to understand something of the process by which they have been obtained. For the theory, reference should be made to the literature already cited; and the programme listing may also be consulted; it is quite liberally embellished with comments, though the main lines of the structure may be difficult to discern at first sight.