READ_CAR.TXT file 16-july-1997 das/hjs The CARIB97 GEOID MODELS ------------------------ The CARIB97 model is a high resolution geoid height model covering the region 9-28N, 86-58W. It has been prepared in a cooperative effort with the National Imagery and Mapping Agency (NIMA). The CARIB97 geoid model has been designed specifically for the Caribbean Sea. Due to data coverage and computational issues, one will find offsets between CARIB97 and either GEOID96 or G96SSS in regions of overlap. You have received these models on CD-ROM, or downloaded them from the National Geodetic Survey (NGS) web site, the NGS FTP site, the NGS bulletin board system, or have received the models on individual floppy disks. Among the files you have received are: GEOID.EXE the geoid interpolation program (GEOID.FOR is source code) (version 3.10 -- Written specifically for distribution with CARIB97, but equally compatable with previous geoid models) DOSXMSF.EXE 32-bit DOS extender (needed for GEOID.EXE) AREA.PAR text file of the filenames of geoid height grids CAR97W.GEO the CARIB97 grid for the Western Caribbean Sea 9-28N, 86-71W CAR97E.GEO the CARIB97 grid for the Eastern Caribbean Sea 9-28N, 73-58W GEOGRD.EXE utility program for sub-area extraction and format conversion (GEOGRD.FOR is the source code) To Install (after uncompressing the files) 1) Make a subdirectory on your hard disk (example: mkdir c:\carib97). 2) Copy the various geoid files into that subdirectory. copy *.* c:\carib97 /v (for example) 3) Repeat step 2) as required for your other sets of geoid files. (If you have installed from floppies, the files are put in the designated subdirectory.) 4) Check your AUTOEXEC.BAT and CONFIG.SYS files to insure compliance with the following notes: Note 1: DOSXMSF.EXE must either be present in the same directory as GEOID.EXE, or, it must be in a directory in your DOS PATH environment variable. (such as: c:\dos, for example) DOSXMSF.EXE may be freely reproduced and distributed, without royalty. Note 2: You must have a statement FILES=25 (or a number greater than 25) in your CONFIG.SYS file. Note 3: Do NOT put CARIB97 files in the same directory as other geoid models (such as MEXICO97, G96SSS or GEOID96). This will cause an error in program GEOID.EXE. To Execute Type GEOID , and follow the prompts. To Terminate You can stop the program at any time by the Control C key combination. BUT, PLEASE DON'T START YET. PLEASE KEEP READING THIS DOCUMENT. How Program GEOID Works ----------------------- The various geoid height grids are stored in the ".GEO" files. Program GEOID will assume that any file in your local directory with a .GEO extension is a geoid height file. You can operate with as few as one .GEO file, or as many as 15. When the program interpolates a given point, it checks the AREA.PAR file for .GEO boundaries, and uses the earliest list entry whose boundaries contain that point. The order in which the .GEO file names appear on the opening screen indicates the order in which the .GEO files are searched. When running program GEOID.EXE, the latitude and longitude of each point must be input. The GEOID96 models are heights above the NAD 83 ellipsoid, while all other recent geoid models (G96SSS, MEXICO97, and CARIB97) are heights above a GRS-80 ellipsoid, centered at the ITRF94(1996.0) origin. However, latitudes and longitudes in the ITRF94/GRS-80 system are very close to those of the NAD 83 system (with only 1-2 meters of horizontal shift.) So either type of latitude and longitude (NAD 83 or ITRF94) may be input, without affecting the interpolated geoid value. This does *not* imply that the geoid heights are heights above a different ellipsoid. Using NAD 83 latitudes and longitudes interchangably with ITRF94/GRS-80 latitudes and longitudes is merely an acceptable horizontal approximation. Each geoid model will always contain geoid heights above the specific ellipsoid, stated above. Use of NAD 27 latitudes and longitudes is really only valid for MEXICO97 and CARIB97, where no NAD 27 to NAD 83 conversion is readily available. The horizontal shifts between NAD 83 and NAD 27 can exceed 100 meters in the United States, causing a noticable difference in the interpolated geoid value in GEOID96 and G96SSS. To convert from NAD 27 to NAD 83 latitudes and longitudes in the U.S., you may use programs NADCON or CORPSCON, available from NGS. The AREA.PAR File ----------------- AREA.PAR is a plain, ASCII text file. It specifies the order in which .GEO files are to be used. If you have a favorite .GEO file, put it at the top of the AREA.PAR list. There is no problem in having overlapping .GEO files, nor is there any problem in having nested .GEO files. The AREA.PAR file specifies their priority of use. PLEASE NOTE: The AREA.PAR file we distribute with the CARIB97 model contains the names of all the CARIB97 grid files. You may not have received them all; you may not want them all. This is not a problem. If a .GEO file name is in the AREA.PAR file, but not in the local directory, then a warning message is issued, and program GEOID proceeds with the files that are available. You must have an entry in AREA.PAR for each .GEO file to be searched. An Example: You just wish to work with the CARIB97 - West file. So, load AREA.PAR into your favorite line editor, and delete the lines referring to the other geoid regions. You may now delete those .GEO files without receiving the warning messages on the opening screen of program GEOID. Save the updated AREA.PAR as plain ASCII text. Data Input ---------- You can key data by hand, point by point, or you can create an input file using a text editor. Several file formats are provided, including the NGS "Blue Book" format. These formats are detailed in a "Help" menu option which appears if you specify an input file name. That file doesn't need to exist if you are only going to look at the supported formats in the "Help". Data Output ----------- Results are collected into an output file. The default name of these files is GEOID.OUT, but you can use any legal file name you choose. (A word of advice: Don't use misleading extensions such as .EXE, .GEO, .BAT, etc.) The format of the output file is linked to the format of the input file to maintain consistency. The CARIB97 Model ----------------- The CARIB97 model was computed on June 20, 1997 using over seven hundred thousand terrestrial and marine gravity values. The method of computation uses a Fast Fourier Transform (FFT) technique to compute the detailed geoid structure, which is then combined with an underlying EGM96 geopotential model. The result is a gravimetric geoid height grid with a 2' X 2' spacing in latitude and longitude, referred to the Geodetic Reference System 1980 (GRS 80) normal ellipsoid in an International Terrestrial Reference System 1994 (ITRF94(1996.0)) frame. The EGM96 model was evaluated to conform to a "tide-free" system, insofar as the Earth's permanent tide effect is concerned. When comparing the CARIB97 model with GPS ellipsoidal heights in the ITRF94 reference frame and tidal benchmarks (in various individual island datums), one may discern individual offsets, on an island by island basis, with an average offset at the -50 cm level. Many different signals contribute to these island-wide offsets. First, the difference between Mean Sea Level (MSL, which is one possible sea level measured at a tide guage) and the geoid level is non-zero, and depends partly on the permanent ocean circulation around each island. Also, the actual sea level measured at each tide guage (mean sea level, mean low water, etc) was not clearly identified, which will contribute to relative differences between differing tide guages. Finally, the treatment of offshore gravity data in the CARIB97 model will contribute to these offsets. In addition to the -50 cm offset, localized systematic errors are evident on some islands. These systematic errors may appear as extreme (30+ ppm) tilts, over very short ranges, island by island. These sort of short of localized extreme tilts are generally too large to be geoid error. GPS height error is in the 2.5 cm RMS range, indicating that these systematic errors are probably caused by relative errors between neighboring tide gauge measurements. Finally, some long-wavelength systematic errors are evident in the comparisons. These errors are a composite of error in the tidal benchmark elevations, error in the GPS ellipsoidal heights, and error in the CARIB97 model itself. Since the errors are long-wavelength, they can be modeled locally as a plane; perhaps at a 5 to 6 part-per-million level. One may expect to see larger tilts in the CARIB97 model near Cuba and Venezuela due to a lack of gravity coverage. The National Imagery and Mapping Agency (NIMA) ---------------------------------------------- The National Imagery and Mapping Agency (NIMA) has often been a significant contributor of data in previous NGS geoid models. For CARIB97, NIMA joined NGS as a full partner, contributing time, data and personnel to help ensure the success of this program. NIMA has provided a major portion of the NGS land gravity data set. NIMA has also been instrumental in the creation of the various 30" and 3" elevation grids in existence. And, NIMA was a partner with NASA in the joint project to compute the new global geopotential model, EGM96. Although the work of the NIMA generally precludes public recognition, their partnership with NGS is gratefully acknowledged and both agencies are looking to the future for further partnerships. GSFC/NIMA Geopotential Model, EGM96 ----------------------------------- The Goddard Space Flight Center (GSFC) and the National Imagery and Mapping Agency (NIMA) have been engaged in a joint project to compute an improved global spherical harmonic model of the Earth's geopotential. This model incorporates the latest satellite tracking data, as well as altimeter data from TOPEX/Poseidon, ERS-1, and the Geosat Geodetic Mission. EGM96 also incorporates new surface and marine gravity data covering the globe, including the former Soviet Union. EGM96 is a global geopotential model expressed as spherical harmonic coefficients complete to degree and order 360. Therefore, the shortest wavelength this model can exhibit is one degree, and its resolution is one-half degree (about 50 km). Although this model does not reproduce geoid structure at very fine resolution, it is global. We thank the many members of the project team for making this model available. Deriving Orthometric Heights From GPS ------------------------------------- One key problem is that each Caribbean island may have it's own datum, or no defined datum at all. In addition, for the continental US, NGVD 29 is not a sea-level datum, and the heights are not true orthometric heights. The datum of NAVD 88 for conterminous North America is selected to maintain reasonable conformance with existing height datums, and its Helmert heights are good approximations of true orthometric heights. In addition, CARIB97 contains local and regional vertical datum offsets, itself. This leads to a warning: Do not expect the difference of a GPS ellipsoidal height at a point and the associated CARIB97 height to exactly match the vertical datum you need. However, one can combine the precision of differential carrier phase GPS with the relative precision of CARIB97 height differences, to approach that of leveling. Include at least one existing benchmark in your GPS survey (preferably many benchmarks). The difference between the published elevation(s) and the height obtained from differencing your adopted GPS ellipsoidal height and the CARIB97 model, could be considered a "local orthometric height datum correction". If you are surveying an extensive area (100+ km), and you occupy a lot of benchmarks, then you might detect a trend in the corrections perhaps a six part-per-million level. This may be error in the CARIB97 model. The United States does not currently consider geoid-corrected GPS orthometric heights as a substitute for geodetic leveling in meeting our Federal Geodetic Control Subcommittee(FGCS) standards for vertical control networks. Studies are underway, and many less stringent requirements can be satisfied by geoid modeling. Widespread success has been achieved with the preceeding models, GEOID96, GEOID93 and GEOID90. The GEOGRD Utility Program -------------------------- GEOGRD -- This converts to and from ".GEO" binary files and ASCII text files. It can also be used to extract subgrids in the process of conversion. For example: one can make a .GEO grid for the island of Jamaica by using CAR97W.GEO, "converting" from binary (.GEO) into binary (.GEO) and specifying an area that surrounds Jamaica. Remember two things: 1) If you do create new .GEO grids, you must update your AREA.PAR file. And, 2) a total of only 15 geoid grid files can be open at once with the PC version of GEOID. If you construct many different grid pairs, consider creating different sub- directories with different combinations of grid pairs in them. A Technical Note on Program GEOID --------------------------------- Some users prefer to write their own interpolation software. If you do, please be aware that there is a loss of precision in the grid file headers for grid spacings of 2'. This is accomodated in program GEOID 3.00 (and later versions, including the 3.10 version supplied with CARIB97) by internally re-computing the grid spacing in subroutine GRIDS. You might need to place similar code in your interpolation software, depending upon how it was written. ------------------ (Example Fortran 77 code) ----------------------------- *** patch for inexact headers (due to 2' spacing) idx1=idnint(DX1*3600.d0) DX(NAREA) = dble(idx1)/3600.d0 idy1=idnint(DY1*3600.d0) DY(NAREA) = dble(idy1)/3600.d0 ***** DX(NAREA) = DX1 **** old code ***** DY(NAREA) = DY1 **** old code ---------------------------------------------------------------------------- Future Plans ------------ A research effort is underway to improve geoid height estimates in the future, perhaps at the 1-cm accuracy level. One important direction is integrating gravity data with GPS and geodetic leveling measurements, and the study of error in GPS ellipsoid heights and in the various island datums in use. Additionally, more gravity data is being collected, and we are investigating the use of additional NIMA gravity data for an improvement to CARIB97. It is likely that this research will yield a significant improvement to our geoid model in 1999. Acknowledgements ---------------- We would like to thank all of the agencies that have contributed to the NGS databases. We thank the United Nations Environment Programme (UNEP) for their GTOPO30 global 30 arcsecond digital terrain data, which was used everywhere but in Florida. We also thank the National Geophysical Data Center (NGDC) for their TOPO30 digital terrain data (TOPO30) which was used in Florida. For More Information -------------------- For Products Available From the National Geodetic Survey: Information Services Branch National Geodetic Survey, NOAA, N/NGS12 301-713-3242 fax: 301-713-4172 For Products Available From the National Imagery and Mapping Agency: NIMA(GIMG), Mail Stop L-41 Geodesy and Geophysics Department National Imagery and Mapping Agency 3200 South Second Street St. Louis, MO 63118-3399 314-260-1174 fax: 314-260-1071 For Information on CARIB97 and Future Research: Dr. Dru A. Smith National Geodetic Survey, NOAA, N/NGS5 301-713-3202 Internet: dru@ngs.noaa.gov Mr. Howard J. Small NIMA Attn: H. Small 3200 S. Second Street St. Louis, MO 63118-3399 314-260-1052 Internet: smallh@nima.mil Visit the CARIB97 web site: http://www.ngs.noaa.gov/GEOID/carib97.html README file 16-July-1997 das/hjs