Abstract
In petroleum studies quantitative tools such as basin modelling are widely used when ranking exploration targets and assessing associated geologic risk. It is therefore vital to assess the confidence that can be associated with the results of a given model based on available measured (control) data. Model validity is checked by an often time consuming (trial and error) calibration against measured data attempting complicated modifications of the model to satisfy the highly non-linear relationships between model parameters. These procedures often result in over-stated confidence in model results. Inversion procedures, however, aim at finding the simplest constellation of model parameters that best agree with calibration data taking the known uncertainties into account. A rapid pseudo-inverse method is presented for 1D deterministic forward models. A simple model for parameter behaviour is used to map the “real” behaviour into the simple model space. Then, by approximating the residual surface with a polynomial or cubic spline interpolation for a few data points only, the resolution limits, sensitivity and uncertainty can be easily assessed and the residual surface can be converted to a pseudo-probability density function for use in later risking procedures. From a measure of the goodness of fit to observed data, in this case vitrinite reflectance, the significance of variations of different input parameters, in this case heat flow, can be rapidly evaluated and uncertainties assessed. The method is illustrated on a data set from a detailed 1D modelling study of a small region of the German Variscan Rhenish Massif.