Are There Thermal Precursors to Eruptions Detectable by ASTER? Evaluating 22 years of Global Medium Resolution Satellite Thermal Observations at 200+ Volcanoes

Abstract

Measurements of temperature changes at volcanic features can reveal trends that could forecast eruptions. Using nighttime advanced spaceborne thermal emission and reflection radiometer observations with a spatial resolution of 90 m/pixel, we collect thermal measurements for 195 subaerial volcanoes that erupted at least once between 2000 and 2022 (810 eruptions) to evaluate for thermal precursors. Roughly 28% of eruptions (225) have sufficient nighttime cloud-free observations to construct a robust time series. We use our deviation from thermal time series baseline algorithm to identify temperature observations that vary from the preceding average by a specified number of standard deviations. In our data set, 23% of eruptions were preceded by an increase in temperature, 6% had a decrease in temperature, 15% had an increase-then-decrease trend, 55% had no detectable thermal precursor to eruption, and 69% of volcanoes had a volcanic thermal change not associated with an eruption. Thermal precursors can start as early as 8 years before an eruption to only a week; however, most precursors last less than a year. We also perform chi-square tests to compare precursors with volcanic and eruptive characteristics. We find that all kinds of thermal precursors are present in all volcanic and eruptive characteristics and the size and duration of the precursor do not indicate size of the subsequent eruption. Finally, we find several examples where more temporally dense satellite thermal observations at spatial resolution of 90 m/pixel or better are needed.