Probabilistic detection of volcanic ash using a Bayesian approach

Abstract

Airborne volcanic ash can pose a hazard to aviation, agriculture, and both human and animal health. It is therefore important that ash clouds are monitored both day and night, even when they travel far from their source. Infrared satellite data provide perhaps the only means of doing this, and since the hugely expensive ash crisis that followed the 2010 Eyjafjalljökull eruption, much research has been carried out into techniques for discriminating ash in such data and for deriving key properties. Such techniques are generally specific to data from particular sensors, and most approaches result in a binary classification of pixels into “ash” and “ash free” classes with no indication of the classification certainty for individual pixels. Furthermore, almost all operational methods rely on expert-set thresholds to determine what constitutes “ash” and can therefore be criticized for being subjective and dependent on expertise that may not remain with an institution. Very few existing methods exploit available contemporaneous atmospheric data to inform the detection, despite the sensitivity of most techniques to atmospheric parameters. The Bayesian method proposed here does exploit such data and gives a probabilistic, physically based classification. We provide an example of the method's implementation for a scene containing both land and sea observations, and a large area of desert dust (often misidentified as ash by other methods). The technique has already been successfully applied to other detection problems in remote sensing, and this work shows that it will be a useful and effective tool for ash detection.Key PointsPresentation of a probabilistic volcanic ash detection schemeMethod for calculation of probability density function for ash observationsDemonstration of a remote sensing technique for monitoring volcanic ash hazards

Decision Analysis for Management of Natural Hazards

“…Environmental modelling has its own set of uncertainties which have been discussed at length elsewhere (Beven, 2009). Understanding one hazard event and its potential consequences may require a multitude of models, each with uncertainties; for example in predicting volcanic activity different models are needed for gas emissions, tephra fallout, debris avalanches, and lahars (Vecchia, 2001, Mackie andWatson, 2014) The ability to simulate hazards, and the resources required to run the models, varies between hazard types and will influence the kind of risk and decision analysis that is appropriate: for example high resolution climate modelling can be very computationally expensive, prohibiting the ability to represent statistics of extreme events using supercomputers (Allen, 2003), and leading some authors to emphasise representation of uncertainty which does not rely on complex models (Dessai et al, 2009, Blazkova and Beven, 2009, Brown et al, 2011b.…”

Spectrally Enhanced Cloud Objects—A generalized framework for automated detection of volcanic ash and dust clouds using passive satellite measurements: 1. Multispectral analysis

“…In an effort to sufficiently capture the complicated multispectral relationships over a wide range of conditions and reduce the many pieces of spectral information into a single objective metric, a Bayesian approach is utilized. Bayesian approaches have been successfully applied to several satellite‐based classification problems [ Uddstrom et al ., ; Merchant et al ., ; Heidinger et al ., ; Kossin and Sitkowski , ; Cintineo et al ., ; Mackie and Watson , ]. As discussed by Kossin and Sitkowski [] and Heidinger et al .…”

“…In an effort to sufficiently capture the complicated multispectral relationships over a wide range of conditions and reduce the many pieces of spectral information into a single objective metric, a Bayesian approach is utilized. Bayesian approaches have been successfully applied to several satellite-based classification problems [Uddstrom et al, 1999;Merchant et al, 2005;Heidinger et al, 2012;Kossin and Sitkowski, 2009;Cintineo et al, 2014;Mackie and Watson, 2014]. As discussed by Kossin and Sitkowski [2009] and Heidinger et al [2012], the classical Bayesian approach is not practical when more than just a few features are used, as the size of the class conditional probability density functions can easily grow to an unmanageable size and become very difficult to sufficiently populate.…”

Atmospheric dependency of the sensitivity of infrared satellite observations to volcanic ash clouds

“…The threshold is often placed on the signal in order to identify the presence of ash, and the strength of the signal is often used to infer ash column loading, from which concentration can be estimated using an assumed (or observed through other means) geometric cloud thickness, 8,[12][13][14] although recently more sophisticated techniques have become available (e.g., see Refs. [15][16][17][18].…”