Correction of Ionospheric effects in Coseismic ALOS data of the 2008 Wenchuan (Mw7.9) earthquake

The Mw7.9 Wenchuan earthquake (12 May 2008) ruptured the middle segment of the Longmenshan (LMS) thrust belt, located at the eastern margin of the Tibetan Plateau. Several research groups have produced maps of the coseismic deformation using ALOS PALSAR data. However, the deformation maps and the estimated fault slip distributions differ significantly from each other in these studies. This difference is mainly because of anomalous InSAR signals, caused by ionospheric effects, which the researchers have dealt with in various different ways or ignored altogether. The ionospheric variations in the Wenchuan coseismic data are unusually strong and affect both the InSAR phase and the azimuth offset tracking measurements. Despite the various research efforts, important questions remain to be adequately addressed, e.g. (1) How do ionospheric effects distort InSAR observations of the coseismic deformation and how can we correct for them in the interferometric phase and azimuth offset measurements? (2) What are the differences between the Wenchuan fault slip models estimated from contaminated and ionospheric-corrected ALOS data? Some researchers have tried to correct for the ionospheric effects seen in the ALOS data using various different methods and external data sources, but their improvements have been very limited. Here we propose a new method utilizing additional interferogram pairs that only include the ionospheric effects, but no coseismic deformation, to remove the ionospheric signals. We applied this method to the InSAR data to produce "ionospheric free" coseismic interferograms and azimuth offset measurements. We validated our InSAR results against GPS measurements, which show that our corrected InSAR data are in a very good agreement with the GPS observations, confirming a successful removal of the ionospheric signals. Out of the 6 parallel ascending ALOS tracks we used to map the coseismic deformation, we find that data from 3 tracks are severely distorted by ionospheric effects. Unfortunately, these are the tracks where the most important fault slip patterns are located, such as the largest surface fault rupture and a transition from pure reverse faulting to a combination of reverse and right-lateral strike slip. This means that previous fault slip models estimated from contaminated ALOS data are likely biased and should be be reevaluated. Beside InSAR data, azimuth offsets often prove to be a useful supplementary information when studying large earthquakes, e.g. for locating the fault rupture. However, until now these data have been mostly ignored in the case of the Wenchuan earthquake due to the strong ionospheric effects. For our final fault model, we combine the corrected ALOS InSAR and azimuth offset data with GPS observations and descending Envisat (both image mode and ScanSAR) data to constrain better the spatial details of the fault slip.