Anomaly separation of gravity data using filtering techniques: Case study of the Ziway and Koka sub basins, central main Ethiopian rift, Ethiopia

Abstract

This study applies three filtering techniques: frequency, upward continuation (UC), and trend surface polynomial filters, to the ground and satellite gravity data from the Ziway–Koka sub-basins to separate the regional and residual anomaly components of the complete Bougour anomaly (CBA). The estimated regional anomalies are then evaluated to identify the relatively reliable filtering technique for the study area. For the UC filter, regional anomaly grids were evaluated using successive UC to heights ranging from 0.5 to 5 km to determine the optimum height for estimating the regional anomalies. Consequently, a UC height of 5 km was determined to best represent the regional anomaly with enhanced resolution. Radial average power spectrum curves were used to determine the cut-off frequency applied in compiling the regional anomaly grid map. As a result, a cut-off frequency of 0.0970525097 cycles/km was applied to the ground-based gravity data, and 0.0918813153 cycles/km to the satellite-based data for estimating the regional anomaly component. Moreover, first, second, and third-order polynomial filters were applied to estimate the regional anomaly component to determine which order offers the best resolution. Among these, the first-order polynomial filter provides a smooth and linear representation of the regional field and was therefore selected as suitable for estimating the regional anomaly with improved resolution. The residual anomaly grids obtained from each filter are quantitatively compared by correlating the anomalies extracted along the profile AA with their corresponding analytical signal anomaly values determined from the CBA grids. For the ground-based gravity data, the correlation coefficients for the frequency, UC, and first-order polynomial filters are 0.64, 0.89, and 0.79, respectively, while for the satellite-based data, the corresponding values are 0.60, 0.86, and 0.84. These results indicate that the UC filter is relatively more reliable for estimating the regional anomaly as compared to the other filters and also effective in decomposing the CBA into its regional and residual components.