Volcanic ash cloud detection from space: a comparison between the RSTASH technique and the water vapour corrected BTD procedure

Abstract

Volcanic eruptions can inject large amounts (Tg) of gas and particles into the troposphere and, sometimes, into the stratosphere. Besides the main gases (H2O, CO2, SO2 and HCl), volcanic clouds contain a mix of silicate ash particles in the size range from 0.1 μm to 1 mm or larger. The interest in volcanic ash detection is high, particularly because it represents a serious hazard for air traffic. Particles with dimensions of several millimetres can damage the aircraft structure (windows, wings, ailerons), while particles less than 10 μm may be extremely dangerous for the jet engines and are undetectable by the pilots during night or in low visibility conditions. Furthermore, ash detection represents a critical step towards quantitative retrievals of plume parameters. In this paper two different satellite techniques for volcanic cloud detection and tracking are compared, namely a water vapour corrected version of the brightness temperature difference (BTD-WVC) procedure and an implementation of the robust satellite technique, specifically configured for volcanic ash (RSTASH). The BTD method identifies volcanic ash clouds on the basis of the brightness temperature difference measured in two infrared spectral bands at around 11 and 12 μm. To account for the atmospheric water vapour differential absorption in the 11–12 μm spectral range, which tends to reduce (and in some cases completely mask) the BTD signal, a water vapour correction procedure has been developed (BTD-WVC), based on measured or synthetic atmospheric profiles. RSTASH instead, is based on the analysis of a time series of satellite records, aimed at identifying signal anomalies through an automatic unsupervised change detection step. To assess the performance of the BTD-WVC and RSTASH methods in detecting volcanic ash clouds, some eruptive events of Mt Etna, observed by the Advanced Very High Resolution Radiometer (AVHRR) sensor, have been analysed. The obtained results show a good agreement between the BTD-WVC and RSTASH techniques for all the considered images, in terms of pixels detected as ‘ash affected’ (i.e. the ash cloud area). In particular, compared to the traditional BTD procedure, the BTD-WVC and RSTASH techniques significantly improve volcanic ash cloud detection, both in daytime and night-time data, especially in the case of low ash loading.