Decision theory is a Bayesian framework that allows the best rational decision to be made in when identifying hazardous regions by considering the probability of the hazardous state given a signal, the prior probability of that hazard occurring, and also the risk associated with a misclassification of the hazard. More generally, decision theory is a rational decision making process whereby a signal is interpreted in order to make the optimum choice of action under uncertainty considering the associated risk. This allows the classification of an atmospheric state, given a signal, to be taken away from the scientist and transferred to that of the decision maker who needs only to show their degree of aversion to the risk of misclassification.
Volcanic ash poses a risk to infrastructure, aviation and health. Economic losses can result from both false positive and false negative classification of airspace contaminated with volcanic ash, however the overall risk associated with a false negative classification of ash-contaminated airspace is less severe than that of a false positive classification. The identification of volcanic ash using infrared remote sensing is a textbook example of where decision theory should be applied to interpreting a signal with associated risk.
This paper demonstrates the use of decision theory to classify pixels as containing volcanic ash, desert dust (a common false positive for volcanic ash) or free from both volcanic ash and desert dust. The example uses the SEVIRI instrument and the ‘reverse absorption’ effect of silicate particles to demonstrate the strengths of decision theory in the classification of hazards under uncertainty and their associated risk.
The implications of decision theory stretch beyond volcanic hazards and can be applied to the classification of any atmospheric or meteorological hazard.