Systematic assessment of atmospheric uncertainties for InSAR data at volcanic arcs using large-scale atmospheric models: Application to the Cascade volcanoes, United States

- Atmospheric uncertainties for InSAR are estimated a priori using atmospheric models.
- Uncertainties are dependent on volcano topography and geographical location.
- Detection thresholds are calculated for InSAR monitoring of Cascade volcanoes.
- Pre-eruptive unrest is most likely to be observed at low relief, shield volcanoes.
- This approach is widely applicable for near-real-time monitoring on arc-wide scales.

Satellite Radar Interferometry (InSAR) is suited to monitoring ground deformation on the scale of volcanic arcs, providing insight into the eruptive cycle over both long and short time periods. However, these measurements are often contaminated with atmospheric artefacts caused by changes in the refractivity of the atmosphere. Here, we test the use of two large-scale atmospheric models, ERA-Interim (ERA-I) and North American Regional Reanalysis (NARR), to correct atmospheric uncertainties in InSAR data from the Cascades Volcanic Arc, United States. At Lassen Volcanic Center, we find that NARR reduces interferogram standard deviation in 79% of cases by an average of 22%.Using NARR, we develop a strategy to produce a priori estimates of atmospheric uncertainties on an arc-wide basis. We show that in the Cascades, the RMS variation in range change is dependent upon volcano topography and increases by 0.7 cm per kilometre of relief. We use this to estimate detection thresholds for long-term monitoring of small magnitude (1 cm/yr) deformation signals, and short-term monitoring of ground deformation associated with pre-eruptive unrest. This new approach of assessing atmospheric uncertainties a priori is widely applicable to other volcanic arcs, and provides realistic estimates of atmospheric uncertainties suitable for use in near-real-time analysis of InSAR data during periods of volcanic unrest.