Water Vapor Tomography of the Lower Atmosphere from Multiparametric Inversion: the Mt. Etna Volcano Test Case

Abstract

Space techniques based on GPS and SAR interferometry allow measuring millimetric ground deformations. Achieving such accuracy means removing atmospheric anomalies that frequently affect volcanic areas by modeling the tropospheric delays. Due to the prominent orography and the high spatial and temporal variability of weather conditions, the active volcano Mt. Etna (Italy) is particularly suitable to carry out research aimed at estimating and filtering atmospheric effects on GPS and DInSAR ground deformation measurements. The aim of this work is to improve the accuracy of the ground deformation measurements by modeling the tropospheric delays at Mt. Etna volcano. To this end, data from the monitoring network of 29 GPS permanent stations and MODIS multispectral satellite data series are used to reproduce the tropospheric delays affecting interferograms. A tomography algorithm has been developed to reproduce the wet refractivity field over Mt. Etna in 3D, starting from the slant tropospheric delays calculated by GPS in all the stations of the network. The developed algorithm has been tested on a synthetic atmospheric anomaly. The test confirms the capability of the software to faithfully reconstruct the simulated anomaly. With the aim of applying this algorithm to real cases, we introduce the water vapor content measured by the MODIS instrument on board Terra and Aqua satellites. The use of such data, although limited by cloud cover, provides a two-fold benefit: it improves the tomographic resolution and adds feedback for the GPS wet delay measurements. A cross-comparison between GPS and MODIS water vapor measurements for the first time shows a fair agreement between those indirect measurements on an entire year of data (2015). The tomography algorithm was applied on selected real cases to correct the Sentinel-1 DInSAR interferograms acquired over Mt. Etna during 2015. Indeed, the corrected interferograms show that the differential path delay reaches 0.1 m (i.e. 3 C-band fringes) in ground deformation, demonstrating how the atmospheric anomaly affects precision and reliability of DInSAR space-based techniques. The real cases show that the tomography is often able to capture the atmospheric effect at the large scale and correct interferograms, although in limited areas. Furthermore, the introduction of MODIS data significantly improves by ∼80% voxel resolution at the critical layer (1,000 m). Further improvements will be suitable for monitoring active volcanoes worldwide.