Multichannel singular spectrum analysis (M-SSA) of InSAR data sets: Data-adaptive interpolation and decomposition of Sentinel-1 time-series at Pacaya Volcano, Guatemala

The goal of this paper is to present considerations, steps and tools to perform statistical analysis of Interferometric Synthetic Aperture Radar (InSAR) time-series by relying on multichannel singular spectrum analysis (M-SSA). We apply these tools to Sentinel-1 InSAR time-series processed for Pacaya Volcano in Guatemala in two steps. First, we produce, in a data-adaptive way, estimates of data points to obtain evenly sampled time-series. The resulting time-series are then decomposed using M-SSA into long-periodic nonlinear trends and oscillatory modes providing a sparse representation of the signals present in the data. Combining M-SSA that includes varimax rotation with power spectrum analysis augments the physical interpretability of the InSAR data set presented herein. Monte Carlo SSA hypothesis testing further helps estimate the statistical significance of the M-SSA modes with respect to a red-noise null hypothesis. The dominant frequencies of the main oscillatory modes retained correlate with frequency peaks of the seasonal variability of the regional hydrological system, as determined from correlograms of rainfall time-series. The spatial patterns of the significant modes correlate with three types of geological structures present at Pacaya volcano: the volcanic edifice, the 2010 and 2014 lava flows, and a collapse scarp dividing the volcanic edifice into an eastern and western part. These findings suggest that, when including the complementary tools presented herein, M-SSA is able to provide a reliable statistical picture of InSAR data sets and that the main M-SSA modes are geophysically meaningful.