High-altitude data are used to calibrate a least-squares recursion filter that estimates the continually changing primary field of an airborne electromagnetic (AEM) system. The coupling changes in fixed-wing towed-bird systems generate “geometry noise” that in the on-time can be much larger than the ground secondary response. The LSQ filter accurately predicts the high-altitude primary field of a fixed-wing system. The filter is then applied to survey-altitude data to estimate the primary field for subsequent subtraction. After removing the primary field, a spatially consistent difference is detected over a range of delay times, as would be expected from geologic responses. A map of decay constants is produced for the survey area using the data corrected by the predicted primary field. Comparing these time constants with those computed from the conventional method, the maximum decay constant detectable was seven times larger. Thus, the new process can characterize conductors that are seven times more conductive than the conventional processing method. The residual primary field occurs at relatively high frequencies compared to targets of interest. At low frequencies class="inline-formula" id="inline-formula-1">class="math mml" alt="Formula" src="F97/embed/mml-math-1.gif" />, we estimate that 28% of the survey-altitude primary field remains whereas only 1% of calibration flight primary is not predicted.