Kentucky Geologic Survey research overview

Research overview:
The energy group at KGS conducts research addressing the development of energy resources (coal, oil, gas) in Kentucky, and to, some extent, more broadly in the Appalachian and Illinois basin regions. Much of this work is focused on fossil fuels and includes, for example, petroleum system analysis that seeks to characterize the source and thermal maturity controls on accumulations of oil and gas. From a non-regulatory standpoint, we also have a mandate to provide research and information that allows development of energy resources to be done with minimal impact to humans and other natural systems. Examples of this work include geologic carbon sequestration, stray gas in groundwater, and methane emissions. This effort also includes, of course, our effort to develop a risk assessment tool(s) for abandoned wells.

The abandoned well program seeks to provide a risk-based strategy and tool for assessing areas and wells that most likely represent a threat to human and natural resource systems. The goal will be to develop a risk matrix with wells being assigned a risk score potentially in a probabilistic framework. Risk scores will be portrayed spatially in an interactive map that allows users to identify areas of greatest risk for leakage and the reasons for the risk. This risk-based map would be a framework for the KY Division of Oil and Gas and Division of Water, with limited financial resources, to address abandoned wells.

In addition to the above deliverable, we envision that the risk-based abandoned well map would also serve as a framework for measurements of fugitive methane and other gases. The goal would be to characterize the range of methane fluxes and soil gases associated with a subset of abandoned wells based on their risk profile. Once the data are collected we can assess whether the distribution of these fluxes is normal or has long tails; i.e. a subset of wells accounts for a large amount of fugitive emissions. The flux and soil gas measurements could then be used to further refine the risk-based abandoned well map.

Development of the risk-based map would be helped through a collaboration with the UK Computer Science department. We plan to approach both the Agriculture and Chemistry departments about collaborative measurements of methane fluxes and soil gases.

Description:
The main focus of the current project is to convert our new LiDAR digital elevation model (~625 GB at 5 ft. resolution) into a format we can analyze. The current DEM is too large to be adequately manipulated in ArcMap. Clipping a portion of that data out is not even possible. Thus, we have moved to MATLAB and wish to convert the dataset into a Matfile. This will enable us to index the dataset while on our server and only load chosen portions of the dataset onto a computer’s RAM. This will enable the clipping of 30x30 pixel squares around each of the ~165,000 known drill pads in the state of Kentucky. These drill pads will be used in a machine learning approach to develop a neural network trained to recognize drill pads from our digital elevation model. This neural network can then be applied to the state-wide LiDAR dataset to assist in risk assessment of abandoned well locations.

We plan to undertake this process a second time in the near future utilizing the state-wide point cloud dataset (~12 TB). Moreover, the ability to analyze our data in-house will facilitate many, currently unknown, future projects.

Persons:

Jason Dortch, David Harris, Marty Parris, Tom Sparks, Brandon Nuttall, Nathan Jacobs

Computational methods:

Data conversion, machine learning

Software:

MATLAB, ArcMap 10.3, ArcPro 64, Petra 3.8.3, Excel 2016, Python 3. All software packages are commercially available. Petra is the main program to keep track of core data and logs related to drilled wells.

Grants

Publications