Thesis: Elastic velocity estimation using image-domain adjoint-state inversion of passive seismic data
The use of passive seismic arrays to monitor injection activities for CO2 geosequestration and hydraulic fracturing has dramatically increased over the past decade. The aim on my research is to obtain improved estimates of the elastic properties of the earth from data acquired with passive seismic arrays. Traditional velocity inversion techniques from earthquake data require identifying arrivals in measured waveforms. However, due to the low signal-to-noise ratio these methods are poorly suited for this problem. I propose to estimate the elastic velocity properties using full-wavefield wave-equation imaging and an image-domain tomography.
This method has three components. First, create images by propagating the compressional and shear wavefields through the current velocity models. Second, automatically assess residuals (i.e. errors) in the images. Finally, use a novel method to invert residuals for updated estimates of the velocity models. The procedure will result in improved velocity estimates that will lead to more accurate understanding of the earth and better appraisal of earthquake properties.
Why my research is important
When injecting large volumes of fluids into the subsurface there it is important to monitoring the processes taking place to ensure the program stays within the prescribed bounds and does not pose a safety risk. This is most often done through earthquake monitoring. Accurately estimating earthquake parameters, however, relies on elastic velocity models. Using incorrect velocity models can lead to inaccurate event parameters that may produce erroneous conclusions with potentially serious financial, safety, or legal consequences. My research aims to improve the elastic velocity estimations, which will provide for more robust decisions and risk assessments by operators of injection wells and could be useful in a variety of different fields within and beyond geoscience.