READ.ME file for the grav and topo files accompanying the MGG-09 sea floor topography prediction. Units, reference surfaces, storage conventions: There are two data sets supplied: gravity (grav) and topography (topo). The file "grav.grd" contains floating-point information and is stored in units of milligals. The file "grav.bin" has values of (milligals * 10) rounded to the nearest integer. The gravity values reported are free-air anomalies at sea level (on the geoid) and are referred to the WGS-84 geodetic system. The values in "topo.grd" are floating-point numbers in meters. The values in "topo.bin" are in meters, rounded to the nearest integer. The values are positive above sea level and negative below sea level, so that a depth of 5000 meters appears as -5000. Latitudes are geodetic latitudes in WGS-84. File formats: "grav.grd" and "topo.grd" -- The files "grav.grd" and "topo.grd" are in the hardware-independent, NetCDF format used by the GMT graphics software. GMT is a free package of C language graphics tools designed for use in the UNIX environment. GMT tools make graytone and color PostScript output for scientific visualization and publication-quality maps and graphs. GMT is distributed free of charge over the Internet by anonymous ftp from kiawe.soest.hawaii. edu (128.171.151.16) in the /pub/gmt directory. NetCDF is a package of I/O routines that allow hardware-independent data storage using XDR plus a description and retrieval layer; it is available by ftp from unidata. ucar.edu (128.117.140.3). If you are using GMT the files grav.grd and topo.grd will be readable directly, and you will not need to write your own code to read them. "grav.bin" and "topo.bin" -- The files "grav.bin" and "topo.bin" are two-byte signed integer values in the Sun byte order (twos-complement, big-endian, or most significant byte first). Vax and Intel chips use little-endian, so IBM PCs and VAXes will have to swap the bytes to use these files; Macs, Suns, SGIs, HPs, newer DECs, etc. should be able to use them as-is. This description should enable you to write your own code to read the .bin files. The files describe arrays of 1600 rows and 7200 columns. The sequence of the array elements is in "scan-line order", which is written by rows, from the top row to the bottom row, and within each row is written from left to right. The rows span latitudes from -30 degrees south to -70 degrees south with 40 rows per degree. The columns span 360 degrees of longitude from 0 degrees east to 360 degrees east, with 20 columns per degree. The array elements are "pixel registered", meaning that the latitude and longitude, in decoimal degrees, of the center of a pixel can be described this way in C: short int data[11520000]; FILE *filepointer; int ij, jrow, icolumn; double xlongitude[7200], ylatitude[1600]; for (jrow = 0; jrow < 1600; jrow++) ylatitude[j] = -30.0 - 0.025 * (jrow + 0.5); for (icolumn = 0; icolumn < 7200; icolumn++) xlongitude[icolumn] = 0.05 * (icolumn + 0.5); if ( (filepointer = fopen("topo.bin", "r")) ++ NULL) { /* Error condition */ } for (ij = 0, jrow = 0; jrow < 1600; jrow++) { for (icolumn = 0; icolumn < 7200; icolumn++, ij++) { if ( (fread((char *)&data[ij], 2, 1, filepointer)) != 1) { /* Error condition */ } /* Here xlongitude[icolumn], ylatitude[jrow], and data[ij] correspond to one another. */ } } fclose(filepointer); Other things to know about the gravity data: The gravity data set is presented by Sandwell and Smith at the 1994 Spring Meetingof the American Geophysical Union. The data are derived from Geosat Geodetic Mission data, and stacked repeat-track data from Geosat, ERS-1, and Topex. The original data sets are measurements of the sea-surface height, from which Sandwell and Smith derive marine gravity anomalies. Details of the method are in Sandwell, Geophysical Journal International, vol. 109, pp. 437-448, 1992. The altimeter data are only available over the oceans, so on land areas the values in these files do not represent the complete gravity field. Because of the method used by Sandwell and Smith, the values on land are not zero or NaN (not a number) but are smoothed, long-wavelength approx- imations of the gravity field obtained from a spherical harmonic model. In some areas (particularly shallow seas where tides are a problem, and where currents are large and variable) the sea surface height is not a good approximation of the geoid, and in these areas the user may find that the gravity values look "noisy" or are not a good approximation of true marine gravity. In most places, however, the gravity grid is good to within 2-3 milligals and wavelength 20 km or larger. Other things to know about the topography data: Smith and Sandwell produced the topography data set which is the sea floor topography predicted from satellite altimetry and is described in the Journal of Geophysical Research, vol. 99, no. B-11, pp. 21,803 - 21,824, 1994. See this article for details of the data set. The prediction is based on the altimeter gravity, and so no prediction is available on land. The prediction grid contains values over land, but because of the method used for prediction, these will be a smoothed approximation of ETOPO-5 values. The values over land in this grid are intended only to complete the grid. Accurate land topography is available from other sources, and Smith and Sandwell may splice it in when a world-wide prediction is made in the future. As discussed in the JGR article, the process does not fit existing depth soundings exactly. Instead, the soundings are used to determine a correlation between satellite gravity and sounding data, and then applies this correlation to the gravity field to predict the short- wavelength (< 160 km) component of the sea floor topography. The long-wavelength (> 160 km) component is determined by gridding the existing soundings and low-pass filtering the grid. Thus the predicted topography is not expected to match true topography exactly, and should not be used as if it did. It may be useful in locating uncharted features, imaging the sea floor for illustrative purposes, and estimating average values of depths in regions. It should not be used to predict the precise depth at a point, such as for making decisions about hazards to navigation. At short wavelengths, the data will not match depths exactly due to limitations of the prediction method. At long-wavelengths, the data may be wrong if the low-pass filtered grid is wrong. This could occur where ship soundings were so severely in error that trouble remained even after low-pass filtering. A further source of error was the use of ETOPO5 vaules to "seed" and spped up the iterative process in regions of extremely sparse soundings: more than 100 km between soundings. It was discovered after the prediction was made that this seeding caused some shallowing to occur in the grid around the "Islas Orcadas Seamounts", which appear in ETOPO5 and published charts but are now known not to exist. The low-pass filter did not remove all of this effect, so there is a slight upwarping of the predicted sea floor in this area. Despite these problems, the prediction should be useful for many applications. Technical people to contact: David Divins National Geophysical Data Center NOAA mail code E/GC-3 325 Broadway Boulder CO 80303 303-497-6505 (phone) 303-497-6513 (fax) ddivins@ngdc.noaa.gov (Internet) Walter Smith NOAA Geosciences Lab NOAA mail code N/OES-12 1305 East-West Highway, room 8423 Silver Spring MD 20910 301-713-2860 (phone) 301-713-4475 (fax) walter@amos.grdl.noaa.gov (Internet) Address questions about the gravity and topography files to Walter Smith. Address questions about other files on the CD, or other issues relating to its use to David Divins.