Constraining the uncertainties of volcano thermal anomaly monitoring using a kalman filter technique

The activity status of a volcano can be evaluated from the remotely measured radiant power (RP). The measured RP contains noise due to reasons such as atmospheric effects and instrument characteristics. Here we first show how to estimate the uncertainty of each single RP measurement. To additionally reduce the temporal noise of the RP time series we apply a Kalman filter. The Kalman filter is able to recursively analyse an unevenly sampled time series. To validate our filtering scheme, we applied it to an eruption of Etna in 2002, as well as to the eruption of Nyamuragira in 2010. In the case of the Etna eruption, the denoised time series agrees well with in situ observations of a waxing and waning flow. For the case of Nyamuragira, the enhanced time series of RP shows more fluctuation probably due to cloud coverage. Thus, we propose a multiple instrument approach that increases the temporal resolution of the RP time series and reduces the associated noise.