A holistic inversion algorithm has been developed for time-domain airborne electromagnetic (AEM) data. The algorithm simultaneously recovers a layered earth conductivity structure as well as unmeasured elements of the system geometry. It inverts a complete flight line of data in one inversion. This allows us to take advantage of the expected along-line continuity of conductivity and system geometry, which cannot be exploited when each sample is inverted independently. The conductivity and thickness of each layer and geometry variable is parameterised by the node coefficients of separate cubic-spline basis functions, which implicitly represent smooth continuous along line variations. Each of the cubic splines may have different node spacings that are chosen to adequately represent the expected scale length of lateral variability of conductivity and system geometry. The regularised inversion scheme is formulated to minimise an objective function comprised of data misfit, reference model misfit, and vertical and horizontal roughness terms. The minimisation is implemented via a gradient-based iterative scheme in which a sparse linearised system is solved by the conjugate gradient method within each iteration. The method has been applied to fixed-wing and helicopter AEM data. The results demonstrate that the method produces conductivity models that are geologically credible and consistent with downhole conductivity logs. They also show improved continuity and interpretability in comparison to sample by sample inversions. We found that the estimation of transmitter-receiver separation and receiver pitch geometry parameters was stabilised by the implicit along line continuity constraints.
