Ash detection in the thermal infrared: towards best practice using a combination of high temporal resolution multispectral data, hyperspectral data and radiative transfer forward modelling

The split window algorithm, or more accurately its application to volcanic ash cloud detection, is twenty years old this year. Often applied, and much maligned, the split-window algorithm has changed little since its first application, due primarily to radiometric limits on infrared satellite sensors. However, the advent of hyperspectral imagery, for example the Atmospheric Infrared Sounder (AIRS), has the capacity to improve our ability to detect and quantify ash. Although the trade off between spectral and spatio-temporal resolution is far too high a cost to make AIRS directly useful to aircraft hazard managers, it has the capacity to illuminate the validity of the multispectral split-window algorithm. In concert with a coupled plume-atmosphere radiative transfer model, hyperspectral data can be used to delimit failings in high temporal resolution retrievals. I will present a comprehensive investigation of the split-window algorithm through (i) validation of multispectral data using near-coincident AIRS images, (ii) forward modelling of detection sensitivities to ash including composition, size and number density and (iii) quantifying the effects of environmental variables including background surface temperature and atmospheric water vapour content. The aim of this research is to provide insights into best practice for application of the split-window algorithm and to look forward to the next generation of IR-enabled research and meteorological platforms.