Use of motion estimation algorithms for improved flux measurements using SO2 cameras

Nial Peters,Alex Hoffmann,Talfan Barnie,Michael Herzog,Clive Oppenheimer

doi:10.1016/j.jvolgeores.2014.08.031

Abstract

SO2 cameras are rapidly gaining popularity as a tool for monitoring SO2 emissions from volcanoes. Several different SO2 camera systems have been developed with varying patterns of image acquisition in space, time and wavelength. Despite this diversity, there are two steps common to the workflows of most of these systems; aligning images of different wavelengths to calculate apparent absorbance and estimating plume transport speeds, both of which can be achieved using motion estimation algorithms. Here we present two such algorithms, a Dual Tree Complex Wavelet Transform-based algorithm and the Farnebäck Optical Flow algorithm. We assess their accuracy using a synthetic dataset created using the numeric cloud-resolving model ATHAM, and then apply them to real world data from Villarrica volcano. Both algorithms are found to perform well and the ATHAM simulations offer useful datasets for benchmarking and validating future algorithms.

Highlights

SO2 cameras are the latest addition to the family of UV spectroscopy techniques used to measure sulphur dioxide (SO2) emissions
It is not our intent to conduct an in-depth comparison of the many different motion estimation algorithms available, and we acknowledge that more suitable algorithms may exist for use with SO2 camera images than the two which we present here
The SO2 fluxes through a horizontal line of pixels 170 m above the crater, calculated using the velocities from Active Tracer High-Resolution Atmospheric Model (ATHAM) and the two motion estimation algorithms, are plotted against time

Summary

Introduction

SO2 cameras are the latest addition to the family of UV spectroscopy techniques used to measure sulphur dioxide (SO2) emissions. Since the first volcanological demonstrations (Mori and Burton, 2006; Bluth et al, 2007) they have been used in numerous studies of SO2 fluxes from volcanoes Using SO2 cameras, estimates of SO2 flux at ~1 s temporal resolution are possible, enabling studies of highly dynamic degassing events such as Strombolian eruptions (Dalton et al, 2009; Mori and Burton, 2009; Tamburello et al, 2012) and periodic fluctuations in passive degassing (Tamburello et al, 2013; Pering et al, 2014). The cameras operate on the principle that the apparent absorbance between the two bands will vary only as a function of atmospheric species that absorb unequally across them (Mori and Burton, 2006). The calibration of apparent absorbance to SO2 column amount is non-trivial (Kantzas et al, 2010; Kern et al, 2010, 2013; Lübcke et al, 2012)

Methods

Results

Discussion

Conclusion