InSAR cumulative displacement cubes and Bayesian change-point probability cubes for Suswa (Kenya), Reykjanes (Iceland) and Domuyo (Argentina), Sentinel-1

This dataset supports the study "Probabilistic detection of deformation episodes in InSAR time series using Bayesian change-point analysis" (Bouygues, Albino and Pinel, submitted to JGR: Solid Earth). It contains InSAR cumulative displacement cubes and the associated Bayesian change-point probability cubes for three real volcanic cases, together with the synthetic cubes used for validation and the files needed to regenerate them.

Real cases are cumulative line-of-sight displacement time series derived from Sentinel-1, using one COMET LiCSAR frame per volcano:

  Suswa (Kenya), frame 130A_09212_131313, ascending, 87 acquisitions
  Reykjanes (Iceland), frame 155D_02579_100800, descending, 219 acquisitions
  Domuyo (Argentina), frame 083D_12636_131313, descending, 179 acquisitions

Each cube covers 500 x 500 pixels on the LiCSAR geographic grid (EPSG:4326, 0.001 degree spacing), centered on the volcanic edifice. Values are cumulative line-of-sight displacements in meters, referenced to the first acquisition of each series. For each case, a probability cube P_CP(x, y, t) gives, for every pixel and every date, the posterior probability that a change point occurs at that date in the displacement time series, following the method described in the paper, which adapts the Bayesian change-point detection algorithm of Ruggieri (2013, https://doi.org/10.1002/joc.3447) to three-dimensional InSAR cubes.

The displacement time series were produced from COMET LiCSAR products (Lazecký et al., 2020, https://doi.org/10.3390/rs12152430). The code that generates the synthetic cases and computes the probability cubes is archived separately at DOI logiciel. 

See README.md for file structure, variables, georeferencing, acquisition dates, line-of-sight vectors and loading examples.

Released under CC-BY-4.0. This work contains modified Copernicus Sentinel data, processed by COMET.