Improving Long-Range Ground-Based Thermal Data: A Study of the 12 August 2011 Paroxysmal Event of Mount Etna.

Abstract:

Ground-based thermal remote sensing is a valuable tool for the study and monitoring of volcanoes and their hazards. Unlike satellite-based sensors, ground-based thermal cameras can be placed and operated as situations demand, covering restricted areas in high detail, or enabling broad but continuous monitoring of activity. While ground-based sensors have been used extensively at short distances (e.g. <1000 m), there have been few instances where cameras have been used to capture data at the substantially long ranges which facilitate permanent installation and monitoring. This is due primarily to factors such as atmospheric attenuation and across-image variations in the target path-length resulting in substantial uncertainty in the derived surface temperatures.

Here we present a methodology for correcting and analysing long-range thermal data using MODTRAN transmissivity values calculated for path-lengths >4 km and under different atmospheric conditions (temperature and relative humidity). These corrections have been applied on a per-pixel basis to selected data from the 12 August 2011 paroxysmal event at Mount Etna, Sicily, to calculate lava flow area, flow volume and radiant heat flux. We examined the sensitivity of low, medium, and high apparent temperature pixels to uncertainty in the atmospheric conditions. We then examined how this variability affects the resulting calculations for flow area, volume and radiant heat flux. This was achieved by introducing a fractional error into the measured values for relative humidity. Fractional errors representing 5–25% (with a step value of 2.5%) of the measured relative humidity value were used. Over this range a maximum change of 9.8% was seen in total number of detected hot pixels, 9.1% for values for area and volume, and 8.8% for maximum radiant heat flux. We also examined the effect of changes in the atmospheric temperature has on these calculations. Using a minimum value of 10 C and a maximum value of 26.5 C (maximum recorded value for the date of the eruption) we found a maximum change of 1.3% in total number of detected hot pixels, 1.2% for values of area and volume, and 13.9% for maximum radiant heat flux.