Massively parallel 3D inversion of gravity and gravity gradiometry data

Reliance on desktop computers limits the scale of 3D inversion of gravity and gravity gradiometry surveys, making it impractical to achieve an appropriate level of resolution and detail for geological interpretation. To begin with, airborne surveys are characterised by very large data volumes. They typically contain hundreds to thousands of line kilometres of data with measurement locations every few metres. Often, surveys cover thousands of square kilometres in area with tens of thousands of line kilometres of data. Regional surveys may be even larger and denser as the result of merging multiple and/or historic surveys. Secondly, 3D modelling of large-scale surveys exceeds the capacity of desktop computing resources. And finally, gravity data are finite and noisy, and their inversion is ill posed. Regularisation must be introduced in order to recover the most geologically plausible solutions from the infinite number of mathematically equivalent solutions. Various strategies for 3D inversion have been previously proposed but few lend themselves to truly large-scale 3D inversion. In this paper, we describe how gravity and gravity gradiometry surveys can be inverted to 3D earth models of unprecedented scale (i.e., hundreds of millions of cells) within hours using cluster computers.