Investigation into the effect of silicate ash on the 8.6 μm sulfur dioxide retrieval in the North Pacific region

The North Pacific region contains over 150 active volcanoes which may produce large eruption plumes. Volcanic plumes are composed of various amounts of silicate ash, ice, SO42-, SO2 and other gases, all of which vary in abundance due to initial concentrations, fallout and atmospheric interactions. These cloud components all have specific transmission signatures within the TIR; however no single wavelength is unique to SO2 due to the overlaps in absorption and scattering features with silicate ash. We propose an ash correction model for the 8.6 μm SO2 algorithm that uses a previously developed ash retrieval to generate corrected at-sensor radiances and test it on the 2001 eruption of Mt. Cleveland. The ash retrieval is based on the split window technique and uses thermal infrared wavelengths unaffected by SO2 to determine the effective radius and total mass of the ash particles. The size distribution is then estimated to generate a series of optical depths for the given distribution and series of wavelengths. MODTRAN models the at-sensor radiance values based on the optical depths, atmospheric conditions at the time of the image acquisition and the vertical location of the plume in the atmosphere. The new at-sensor radiance values are used by the 8.6 um algorithm to estimate corrected SO2 total column abundances. The 19 February 2001 eruption plume of Mt. Cleveland contained approximately 60 kt of SO2 and extended northwest-southeast for 1000 km. A comparison of the uncorrected to corrected MODIS SO2 retrieval for the northwestern eruption lobe indicates an overestimate of 9.4% by the uncorrected retrieval. 3892 pixels in the northwestern lobe contain silicate ash with a total ash mass of 13 kt and SO2 column abundance of 33 kt. Approximately 19 kt of SO2 was estimated for the pixels with ash based on the original at-sensor radiances and 17 kt for the corrected at-sensor radiance values. The 2001 Mt. Cleveland case study indicates the 8.6 μm SO2 retrieval overestimates SO2 column abundance due to the presence of ash, however, it is minimal and the original model is robust.