| Summary: | In the present work an improvement is made to the line-to-line correlations technique for the removal of errors of leveling of geophysical data obtained by aerial survey. This technique, under the assumption of continuity and strong correlation of the adjacent flight line profile registers, achieves leveling by constructing a functional estimator through a moving window in one dimension. This estimator is applied line by line recursively reducing the leveling errors and adjusted them in the sense of the minimum squares. Our improvement consists in the implementation of n-functional estimators, based on orthogonal polynomial successions of Chebyshev and Legendre. The procedure used consisted in the generation of synthetic data to which ad-hoc leveling errors of different typical characteristics were introduced to simulate the ones found in a real aeroprospection. We then apply the technique with this improved method as if it were real data and compared with the synthetic data without leveling error. We have determined the size range of the estimators, in which the Chebyshev and Legendre bases significantly remove the leveling errors regarding the implementation using the canonical form, with a percentage relative error of less than 5%.Finally, we apply the improved technique to a real aeromagnetic record acquired on the volcanic sector of the James Ross Basin in the Weddell Sea, in the northeastern end of the Antarctic Peninsula, obtaining a very good result in the removal of the leveling errors of the survey. En el presente trabajo se realiza una mejora a la técnica correlación línea a línea (line-to-line correlations) para la remoción de errores de nivelación de datos geofísicos obtenidos mediante prospección aérea. Esta técnica, bajo la hipótesis de continuidad y fuerte correlación de los registros de perfiles de líneas de vuelo adyacentes, logra la nivelación mediante la construcción de un estimador funcional mediante una ventana móvil en una dimensión. Se aplica el estimador línea a línea en forma ...
|
|---|