False negatives and positives in the interpretation of EM data

Fanavoll, Jonny; Danielsen, Jens; Hesthammer, Jonny; Stefatos, Aris

doi:10.1190/1.3513874

Cited by 4 publications

(1 citation statement)

References 4 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In 2010 several papers were published analyzing the performance of CSEM technology. Three such papers present the database shown in Figure 1, hereafter referred to as "the Hesthammer database" Fanavoll et al, 2010). Two other papers, published by Buland et al (2010Buland et al ( , 2011, are based on similar NAR distributions for surveys over hydrocarbon discoveries, with the mean value of 37% for the calibration surveys and 39% for exploration surveys, and STDs of 25% and 27%, respectively.…”

Section: Introductionmentioning

confidence: 98%

Estimates of the anomaly threshold in controlled-source electromagnetics

et al. 2012

View full text Add to dashboard Cite

SPECIAL SECTION: M a r i n e a n d s e a b e d t e c h n o l o g yEstimates of the anomaly threshold in controlled-source electromagnetics C ontrolled-source electromagnetics (CSEM) has been commercially applied for hydrocarbon exploration for more than 10 years. The normalized amplitude response (NAR), a common measure for CSEM anomaly strength, is the ratio of the CSEM response at the target to the background response. Use of this measure allows anomaly strength to be quantified but leaves open the question of how strong the NAR level needs to be before it can be considered significant for hydrocarbon exploration. We apply three statistical approaches to determine the appropriate NAR threshold for defining a CSEM anomaly and for estimating the overall significance of the CSEM method in hydrocarbon exploration.The analysis is based on the published data in Hesthammer et al. (2010aHesthammer et al. ( , 2010b, referred to here as "the Hesthammer database." This database reports NAR levels and hydrocarbon exploration outcomes versus the depth to the target for a set of drilled prospects. The database authors used an NAR level of 1.15 (or 15%) for their anomaly threshold, while all three of our analyses indicate that this threshold should be higher.The first approach is rooted in classic statistics and is a straightforward comparison of false positive and false negative errors. The criterion of equalizing the number of false positives and false negatives suggests than an anomaly threshold of 20-25% is more appropriate. The second approach uses a Bayesian method and seeks the NAR threshold that maximizes the Bayes factor. For this data set, an NAR threshold of 35% maximizes the Bayes factor at a value of 2.67. The third method is a value of information (VOI) approach based on Bayesian statistics. This approach depends on specific economic factors and gives maximal VOI of CSEM data also at an NAR threshold of 35%.Jeffreys (1961) provided a useful scale for evaluating the informational significance of the Bayes factor. Using his scale, Bayes factors between 3 and 10 are characterized as "substantial." The maximum Bayes factor of 2.67 achieved with this database using an NAR threshold of 35% approaches Jeffreys' "substantial" information category.In our study, we use the statistics of the Hesthammer database itself to estimate the threshold at which a CSEM anomaly is significant for hydrocarbon exploration. Our overall conclusion is that the NAR threshold for this database needs to be higher than 15%. However, the Hesthammer database presumably was acquired over approximately a decade, so it contains a mix of older data with low signal-to-noise ratio as well as more recent higher-quality surveys. Because our conclusions are based on the entire data set, our results do not reflect advancements in technology.Corresponding author: ifilina@hess.com Figure 6. The VOI of CSEM data as a function of CSEM anomaly threshold for fixed NPV of $150 million. Maximal VOI occurs at an NAR threshold of 35%.

show abstract

Section: Introductionmentioning

confidence: 98%