Probability forecast verification in petroleum exploration: insights from the Norwegian Continental Shelf

Biased forecasts lead to biased decisions. An essential pre-drill input in petroleum exploration is the geological probability of success (PoS). If the PoS is biased and lacks accuracy and reliability, the result is value erosion. We use the Brier score, skill score, bias and the attribute diagram to assess the quality of PoS forecasts for prospects on the Norwegian Continental Shelf (NCS) from 1990 to 2022. Pre-drill and post-drill information about the prospects are reported to the Norwegian Offshore Directory. As the reported PoS is obtained by multiplying the probability of source, reservoir and trap, verification measures are also applied to these three factors. Overall, NCS forecasts tend to exhibit pessimism and overconfidence, with some improvement in bias over time. The trap forecasts consistently exhibit a negative skill score over time, indicating a performance no better than the standard reference class. Furthermore, PoS forecasts are not reliable. Biased forecasts indicate that forecasters need to revise their judgments, while negative skill scores imply that a forecast may not add value. Poor reliability indicates that prospects estimated to have high PoS may not be successful, and vice versa. These shortcomings can lead explorers to prioritize not viable prospects while missing the more promising ones.