Benefits of GNSS Local Observations Compared to Global Weather-Based Models for InSAR Tropospheric Corrections Over Tropical Volcanoes: Case Studies of Piton De La Fournaise and Merapi

Abstract

From repeat-pass interferometry, tropospheric signals often prevent the detection of ground deformation signals. In recent years, tropospheric corrections derived from global weather-based models have been implemented in several InSAR processing chains. In this study, we evaluate the performance of two weather-based models (ERA5 and GACOS) on two tropical volcanoes: Piton de la Fournaise and Merapi. For Piton de la Fournaise, the reduction of the tropospheric noise is efficient for $\sim$30% and $\sim$60% of the data sets for GACOS and ERA5, respectively. For Merapi, the performance reaches $\sim$40% for GACOS and $\sim$50% for ERA5. Although GNSS local stations provide real-time information about tropospheric delays, their potential for improving InSAR corrections on active volcanoes is under-exploited. Here, we produce local GNSS-based tropospheric corrections and compare their performance to global weather-based models. For Piton de la Fournaise, the gain of performance with $\sim$34 GNSS stations is about 25% compared to ERA5 models. GNSS-based corrections increase the signal-to-noise ratio in InSAR time series allowing the detection of ground displacements between July and December 2021. For Merapi, GNSS-based models with only 5 stations spatially distributed at different elevations are as efficient as ERA5 models. GNSS-based corrections induce a decrease in the noise level from values $>$1–0.5 cm in a period of quiescence. Here, we show that GNSS-based models are an efficient alternative to global weather-based models for instrumented volcanoes. The proposed approach paves the way toward near real-time InSAR monitoring of volcanic unrest and other processes (landslides, groundwater extraction).