Magma sources involved in the 2002 Nyiragongo eruption, as inferred from an InSAR analysis

Abstract

On 17 January 2002, Nyiragongo volcano erupted along a 20 km‐long fracture network extending from the volcano to the city of Goma. The event was captured by InSAR data from the ERS‐2 and RADARSAT‐1 satellites. A combination of 3D numerical modeling and inversions is used to analyze these displacements. Using Akaike Information Criteria, we determine that a model with two subvertical dikes is the most likely explanation for the 2002 InSAR deformation signal. A first, shallow dike, 2 km high, is associated with the eruptive fissure, and a second, deeper dike, 6 km high and 40 km long, lies about 3 km below the city of Goma. As the deep dike extends laterally for 20 km beneath the gas‐rich Lake Kivu, the interaction of magma and dissolved gas should be considered as a significant hazard for future eruptions. A likely scenario for the eruption is that the magma supply to a deep reservoir started ten months before the eruption, as indicated by LP events and tremor. Stress analysis indicates that the deep dike could have triggered the injection of magma from the lake and shallow reservoir into the eruptive dike. The deep dike induced the opening of the southern part of this shallow dike, to which it transmitted magma though a narrow dike. This model is consistent with the geochemical analysis, the lava rheology and the pre‐ and post‐eruptive seismicity. We infer low overpressures (1–10 MPa) for the dikes. These values are consistent with lithostatic crustal stresses close to the dikes and low magma pressure. As a consequence, the dike direction is probably not controlled by stresses but rather by a reduced tensile strength, inherited from previous rift intrusions. The lithostatic stresses indicate that magmatic activity is intense enough to relax tensional stresses associated with the rift extension.