DipReader (Beta)

The DipReader algorithm, based on multiple AI tools developed in Python, is designed to efficiently digitize and analyze big-data from legacy dipmeter charts. These charts, which predate the digital era, record the dip and azimuth of geological layers measured at various depths within a well. Historically, a significant volume of well logging data exists solely in paper form, necessitating labor-intensive manual transcription for inclusion in computational analyses. DipReader streamlines this process by using advanced techniques such as convolutions and mathematical morphology, leveraging the capabilities of the OpenCV and Scipy libraries. It accurately identifies and extracts pertinent information from scanned images of these documents, thereby enabling automated data transcription and facilitating easier integration into modern geoscientific studies.

SmartSolo code

Numerous libraries and repositories are available to address the challenges of processing seismic data. While these resources are designed for broad applicability, they often do not cater specifically to the unique devices used in our lab experiments, such as the 3-channel SmartSolo geophones. Each SmartSolo unit is equipped with a compass, thermometer, GPS, and other sensors, collecting extensive metadata along with seismic records. This metadata, which includes details on orientation, placement, frequency response, and timing, greatly facilitates the deployment of SmartSolos for precise seismic recording, offering convenience and flexibility.

However, the SmartSolo metadata requires significant processing to be in a format suitable for effective utilization. As a result, we developed a new script to automate the analysis and formatting of seismic files with metadata logs from a SmartSolo array. This script prepares the data for use with other packages like Phasenet, ensuring accurate test results.

SmartSolos have the capability to sample at much higher rates compared to earlier devices, which is invaluable for accurately picking seismic arrival times, especially when devices are deployed closely and the array aperture is short. This high sampling rate, while beneficial, also results in a large volume of data that can be challenging to process. Fortunately, not all signal processing procedures need to be applied to every geophone in the array. Only certain processes benefit from high-frequency sampling, and often, only specific time intervals around seismic events are required for further analysis.

For instance, Phasenet effectively picks seismic arrivals from high-frequency data when it’s downsampled to 100Hz for a single geophone. Subsequently, only the time series spanning the actual seismic events are needed from the other geophones for processing each arrival. These shorter intervals can be managed more efficiently, even at a high sample rate. This selective approach to data processing optimizes the utility of the SmartSolo’s capabilities while managing the challenges posed by large data volumes.