Articles | Volume 11, issue 2
https://doi.org/10.5194/amt-11-781-2018
https://doi.org/10.5194/amt-11-781-2018
Research article
 | 
08 Feb 2018
Research article |  | 08 Feb 2018

Improved optical flow velocity analysis in SO2 camera images of volcanic plumes – implications for emission-rate retrievals investigated at Mt Etna, Italy and Guallatiri, Chile

Jonas Gliß, Kerstin Stebel, Arve Kylling, and Aasmund Sudbø

Abstract. Accurate gas velocity measurements in emission plumes are highly desirable for various atmospheric remote sensing applications. The imaging technique of UV SO2 cameras is commonly used to monitor SO2 emissions from volcanoes and anthropogenic sources (e.g. power plants, ships). The camera systems capture the emission plumes at high spatial and temporal resolution. This allows the gas velocities in the plume to be retrieved directly from the images. The latter can be measured at a pixel level using optical flow (OF) algorithms. This is particularly advantageous under turbulent plume conditions. However, OF algorithms intrinsically rely on contrast in the images and often fail to detect motion in low-contrast image areas. We present a new method to identify ill-constrained OF motion vectors and replace them using the local average velocity vector. The latter is derived based on histograms of the retrieved OF motion fields. The new method is applied to two example data sets recorded at Mt Etna (Italy) and Guallatiri (Chile). We show that in many cases, the uncorrected OF yields significantly underestimated SO2 emission rates. We further show that our proposed correction can account for this and that it significantly improves the reliability of optical-flow-based gas velocity retrievals.

In the case of Mt Etna, the SO2 emissions of the north-eastern crater are investigated. The corrected SO2 emission rates range between 4.8 and 10.7 kg s−1 (average of 7.1  ±  1.3 kg s−1) and are in good agreement with previously reported values. For the Guallatiri data, the emissions of the central crater and a fumarolic field are investigated. The retrieved SO2 emission rates are between 0.5 and 2.9 kg s−1 (average of 1.3  ±  0.5 kg s−1) and provide the first report of SO2 emissions from this remotely located and inaccessible volcano.

Download
Short summary
The paper focusses on gas-velocity retrievals in emission plumes using optical flow (OF) algorithms applied to remote sensing imagery. OF algorithms can measure the velocities on a pixel level between consecutive images. An issue of OF algorithms is that they often fail to detect motion in contrast-poor image areas. A correction based on histograms of an OF vector field is proposed. The new method is applied to two example volcanic data sets from Mt Etna, Italy and Guallatiri, Chile.