Articles | Volume 35
Adv. Geosci., 35, 105–113, 2013
https://doi.org/10.5194/adgeo-35-105-2013
Adv. Geosci., 35, 105–113, 2013
https://doi.org/10.5194/adgeo-35-105-2013

  04 Sep 2013

04 Sep 2013

Atmospheric corrections in interferometric synthetic aperture radar surface deformation – a case study of the city of Mendoza, Argentina

S. Balbarani1, P. A. Euillades1, L. D. Euillades1, F. Casu2, and N. C. Riveros1 S. Balbarani et al.
  • 1CEDIAC Institute, National University of Cuyo, 5500 Mendoza, Argentina
  • 2IREA Institute, National Research Council, 80124 Naples, Italy

Abstract. Differential interferometry is a remote sensing technique that allows studying crustal deformation produced by several phenomena like earthquakes, landslides, land subsidence and volcanic eruptions. Advanced techniques, like small baseline subsets (SBAS), exploit series of images acquired by synthetic aperture radar (SAR) sensors during a given time span.

Phase propagation delay in the atmosphere is the main systematic error of interferometric SAR measurements. It affects differently images acquired at different days or even at different hours of the same day. So, datasets acquired during the same time span from different sensors (or sensor configuration) often give diverging results. Here we processed two datasets acquired from June 2010 to December 2011 by COSMO-SkyMed satellites. One of them is HH-polarized, and the other one is VV-polarized and acquired on different days.

As expected, time series computed from these datasets show differences. We attributed them to non-compensated atmospheric artifacts and tried to correct them by using ERA-Interim global atmospheric model (GAM) data. With this method, we were able to correct less than 50% of the scenes, considering an area where no phase unwrapping errors were detected. We conclude that GAM-based corrections are not enough for explaining differences in computed time series, at least in the processed area of interest. We remark that no direct meteorological data for the GAM-based corrections were employed. Further research is needed in order to understand under what conditions this kind of data can be used.

Download