Quantification of L-band InSAR coherence over volcanic areas using LiDAR and in situ measurements

Interferometric Synthetic Aperture Radar (InSAR) is a powerful tool to monitor large-scale ground deformation at active volcanoes. However, vegetation and pyroclastic deposits degrade the radar coherence and therefore the measurement of 3-D surface displacements. In this article, we explore the complementarity between ALOS-PALSAR coherence images, airborne LiDAR data and in situ measurements acquired over the Piton de La Fournaise volcano (Reunion Island, France) to determine the sources of errors that may affect repeat-pass InSAR measurements. We investigate three types of surfaces: terrains covered with vegetation, lava flows (a′a, pahoehoe or slabby pahoehoe lava flows) and pyroclastic deposits (lapilli). To explain the loss of coherence observed over the Dolomieu crater between 2008 and 2009, we first use laser altimetry data to map topographic variations. The LiDAR intensity, which depends on surface reflectance, also provides ancillary information about the potential sources of coherence loss. In addition, surface roughness and rock dielectric properties of each terrain have been determined in situ to better understand how electromagnetic waves interact with such media: rough and porous surfaces, such as the a′a lava flows, produce a higher coherence loss than smoother surfaces, such as the pahoehoe lava flows. Variations in dielectric properties suggest a higher penetration depth in pyroclasts than in lava flows at L-band frequency. Decorrelation over the lapilli is hence mainly caused by volumetric effects. Finally, a map of LAI (Leaf Area Index) produced using SPOT 5 imagery allows us to quantify the effect of vegetation density: radar coherence is negatively correlated with LAI and is unreliable for values higher than 7.5.