Gravity empirical covariance values for the continental United States

Abstract

Variances and correlation distances have been determined for terrain corrected Bouguer anomalies in all 30 x 30 arc min blocks covering the continental United States. Also linear trend values were subtracted from the data. In order to make the determination independent of the frequently irregular local data distribution, gravity values predicted in a regular grid were used. The prediction was made using least squares collocation, constructing local surfaces representing the gravity anomalies for each 30 x 30 arc min block. For this purpose 544,000 terrain-corrected Bouguer anomalies were available in the U.S. National Geodetic Survey gravity data base. Therefore in order to limit the computational effort, a sequence of surfaces was determined by selecting in each block an increasing number of data points until a limit of the absolute value of the error equal to 4 mGal was reached. This procedure reduced the number of gravity observations used per block to a very few (in areas of little gravity variation) to maximally 188. The correlation distance was found to be smallest in areas of high topography. Large variances are associated with the midcontinental gravity high and tectonic features on the Pacific coast.

A review of non-stationary spatial methods for geodetic least-squares collocation

“…• Goad et al (1984) use a linear polynomial (tilted plane) to remove the trend from Bouguer gravity anomalies over the continental United States;…”