Resolution measures for 3D magnetic inversions
Inversion of magnetic data into 3D models is becoming commonplace. The theoretical basis of the method is well established and has been extended to include constraints based on physical properties and geologic information. Nevertheless, only limited attention is paid to assessing the reliability of computed models, which usually involves deriving some measure of how well features within the models can be resolved. Resolution lengths determined from resolution matrices can be unrealistically small, suggesting that the recovered models are more reliable than they really are. One cause of this effect is the calculated model itself, which directly influences the character of the resolving functions. We have developed an approximate resolution measure not affected by the calculated model, and we found it to give more realistic resolving lengths. This approximation is computationally less demanding and can be calculated prior to inversion. It suggests that the resolution length is equal to the depth of the parameter or model feature of interest.