Computer vision for improved estimates of SO2 emission rates and plume dynamics
Datasets usually provide raw data for analysis. This raw data often comes in spreadsheet form, but can be any collection of data, on which analysis can be performed.
Imaging cameras operating at ultraviolet (UV) and infrared (IR) wavelengths can measure sulphur dioxide (SO2) gas path concentrations or slant column densities. These measurements are useful in a variety of applications including the monitoring of emissions from volcanoes and also emissions from stacks at industrial plants and on ships. The usefulness of these data is increased if the emission rates (or fluxes) of the gases can also be estimated. Here we present an optical flow algorithm that allows rapid and accurate estimates of emission rates using both UV and IR camera imagery sampling at around 1 Hz or higher. Examples are provided from measurements made at Turrialba volcano, Costa Rica, and also at a ship in Hong Kong harbour. Other aspects of the properties of the fluid flow are also introduced, notably the divergence and the vorticity of the two-dimensional wind field. We demonstrate how the divergence can be used in a new method to calculate the emission rate and show how rotational effects observed in volcanic plumes and the resulting entrainment of ambient air affects plume rise and can be observed using vorticity. This is an important aspect for understanding the emplacement of gases and particles into the atmosphere that are subsequently transported by atmospheric winds, sometimes causing pollution episodes at long distances from the source.