Estimation of Undiscovered Resources by Incorporating Spatial Models of Degree of Exploration into Data-Driven Models such as Weights of Evidence and Logistic Regression
Data-driven models such as weights of evidence and logistic regression are useful for predicting the occurrence of known resources, but they do not specifically address the undiscovered resource base. The missing link between known and undiscovered resources is the degree of exploration, i.e., the thoroughness of past exploration efforts and its variation in space. If a degree-of-exploration map can be created, it becomes possible to estimate undiscovered resources. Such an exploration model can be quantitatively intersected with a data-driven resource occurrence model using relatively simple mathematical expressions to spatially estimate the probability of occurrence of undiscovered resources. It also becomes possible to assess, and compensate for, a potential bias in the distribution of known resources (i.e., training points) caused by spatially biased past exploration efforts.We show an example of a degree-of-exploration model built using fuzzy logic for geothermal systems in the state of Nevada, USA. In building this exploration model, consideration was given to where geothermal systems are most likely to remain hidden (i.e., without surface expressions) as well as to where such geothermal systems have been most efficiently explored with subsurface methods such as drilling. The degree-of-exploration model was then intersected with a weights-of-evidence model of geothermal systems in order to produce a predictive map of undiscovered geothermal resources. We used this map to estimate that undiscovered geothermal systems in Nevada outnumber known resources by a ratio of more than 2:1.The evidential layers used in the weights-of-evidence model included gravity gradient, crustal strain, earthquakes, and isostatic gravity. These layers were selected in part because of their perceived independence from exploration factors. However, after construction of the degree-of-exploration model, all evidence layers were found to correlate with degree of exploration, with a systematic upwards bias in the positive weights of as high as 23％. This bias is likely caused by the natural tendency of past exploration efforts to focus on the same favorable areas identified by the weights-of-evidence model. The degree-of-exploration model provides a means of approximately compensating for this exploration bias.
M. F. Coolbaugh G. L. Raines
University of Nevada, Reno, MS 172, Reno, NV 89557 USA
国际会议
The 12th Conference of the International Association for Mathematical Geology(第12届国际数学地质大会)
北京
英文
90-93
2007-08-26（万方平台首次上网日期，不代表论文的发表时间）