302 PETROLEUM GEOLOGY/Exploration Figure 11 Time lapse (four dimensional) seismic survey over a North Sea field; the changes in amplitude result from a significant depletion of the oil due to extraction OWC, oil water contact Reproduced with permission of Millenium Atlas Company Limited Subjective Methods Subjective methods result from the implicit intuitive thought processes of a single individual or group of individuals and can be formidable tools when a substantial body of expertise is available However, there is the problem of bias: usually people are more certain about the unknown than they can afford to be on the basis of the information available, and some people are consistently pessimistic, while others are optimistic There is also the problem of consistency of knowledge: local knowledge can override possibly more relevant but less well-known worldwide information Statistical Methods Statistical methods attempt to extrapolate past experience using a variety of statistical techniques Some depend on the validity of an essentially linear extrapolation, for example a hydrocarbon richness per unit volume of sediments in explored basins, which can then be applied to unexplored basins Other methods attempt to improve this extrapolation by considering past exploration results, such as the decline of success ratio through time and the decline of field size as a function of exploration effort In the example shown in Figure 15, the larger fields are discovered first (the so-called ‘creaming’ effect), but, although the average size of the fields found is decreasing, the number found is still increasing One of the best known of such studies is that of M King Hubbert, who in 1956 predicted that US oil production would peak in the early 1970s; this proved to be correct (the so-called ‘Hubbert’s Peak’; Figure 16) Disadvantages of these methods are that a certain degree of exploration maturity is required for the necessary historical data to be available, the results may be strictly applicable only to the particular basin or province analysed, and the interrelationships of the variables chosen are not always known and correlations may be fortuitous Deterministic Models Deterministic models attempt to integrate all the relevant geological concepts and research findings on the processes of hydrocarbon generation, migration, accumulation, and retention into a coherent model that will provide quantitative estimates of the hydrocarbon volumes likely to be present and the risks involved These methods are now widely employed in the industry The problem is that the information available is usually insufficient for precise estimates of the required variables to be made, and, in such