Multi-level continuous active source seismic monitoring (ML-CASSM): Application to shallow hydrofracture monitoring

Induced subsurface processes occur over a wide variety of time scales ranging from seconds (e.g. fracture initiation) to days (e.g. unsteady multiphase flow) and weeks (e.g. induced mineral precipitation). Active source seismic monitoring has the potential to dynamically characterize such alterations and allow estimation of spatially localized rates. However, even optimal timelapse seismic surveys have limited temporal resolution due to both the time required to acquire a survey and the cost of continuous field deployment of instruments and personnel. Traditional timelapse surveys are also limited by experimental repeatability due to a variety of factors including geometry replication and near-surface conditions. Recent research has demonstrated the value of semi-permanently deployed seismic systems with fixed sources and receivers for use in monitoring a variety of processes including near-surface stress changes (Silver et.al. 2007), subsurface movement of supercritical CO2 (Daley et.al. 2007), and preseismic velocity changes in fault regions (Niu et. al. 2008). This strategy, referred to as continuous active source seismic monitoring (CASSM), allows both precise quantification of traveltime changes on the order of 1.1 x 10^-7 s and temporal sampling on the order of minutes. However, as previously deployed, CASSM often sacrifices spatial resolution for temporal resolution with previous experiments including only a single source level. We present results from the first deployment of CASSM with a large number of source levels under automated control. Our system is capable of autonomously acquiring full tomographic datasets (10 sources, 72 receivers) in 3 minutes without human intervention, thus allowing active source seismic imaging (rather than monitoring) of processes with short durations. Because no sources or receivers are moved in the acquisition process, signal repeatability is excellent and subtle waveform changes can be interpreted with increased confidence. This technique, which we refer to as ML-CASSM, was deployed at a DNAPL contaminated site undergoing bioremediation through hydrofracture emplacement. ML-CASSM was used to image fracture propagation in two crosswell profiles located several meters from the initiation point. Since the entire fracturing process occurred over a 45 minute period, traditional timelapse acquisition methods would have been incapable of capturing the stages of fracture growth. With ML-CASSM, 12 complete datasets were acquired over the course of fracturing allowing estimates of fracture propagation through the two imaging planes. In addition to these datasets, hundreds of sequential surveys acquired before and after the fracturing procedure allowed estimation of system stability at baseline and the combination of fracture consolidation and pressure dissipation afterwards. The fracture zone was visible seismically as a localized area of reduced P-wave velocity and increased P-wave attenuation. In addition to changes in the primary arrival, diffracted events and scattering from the fracture were observed leaving open the possibility of fracture characterization using later components of the wavefield.