Development of High-Resolution Groundwater Models for Predicting Water
Levels
Task 1: Establish a hydrogeologic database for Louisiana, Alabama, and Mississippi
Task 1-1: Collection well log data from the Louisiana Department of Natural Resources and the Mississippi Department of Environmental Quality and design a database. More than 120,000 well logs in 64 parishes of Louisiana state have been collected. The well-logs include wireline electrical logs, drillers’ logs, and geotechnical borings. The project offered an opportunity to integrate the work of thousands of undergraduate student helpers who have worked over the past five years to transcribe Louisiana’s drillers’s logs and geotechnical borings. Lithologic descriptions were typed and stored in Excel worksheets separated by parishes. Easson (Co-PI) is working with students to build a geological database for Mississippi. They are using various types of logs, which can be converted into coordinates for a spatial database. Once the database is built, the research team will meet to integrate the data into a large dataset that will be used for developing MODFLOW models.
Task 1-2: Develop machine learning algorithms to digitize typed and handwritten image-based drillers’ logs into CSV format. This is a challenging task as the image files are not in a uniform format. The study found that Google Vision performs well for typed text but has difficulty recognizing handwritten text.
Task 1-3: Co-PI Tsai hired a digitization company to digitize wireline electrical logs in Louisiana. Only short-range resistivity, spontaneous potential, and gamma-ray curved were digitized into the LogASCII Standard (LAS) formatted files. The LAS file can be read by many software.
Task 1-4: Electric logs were interpreted into sand facies and shale facies. Depth values were typed and stored in Excel worksheets by Co-PI Tsai’s graduate students and student workers. The digitized drillers’ logs and electrical logs in Louisiana are displayed to the public through a Google site.
Task 1-5: A 3D stratigraphic model for Louisiana based on the existing well log dataset was recently published. The public can access the model result through ArcGIS Online. The project will generate 2D cross-sections to display to the public. If successful, we will develop Mississippi and Alabama models with the aforementioned well-log data set.
Task 1-6: Generate KML files to display well log data via Google Earth. The well-logs displayed in the Louisiana Well Log Portal are KML file of 64 Louisiana parishes.
Task 1-7:Generate VTU files to display cross-sections and stratigraphy models to ParaView and Aquaveo’s GMS. We have successfully tested this code in the New Orleans groundwater model. We will generate VTU file to visualize the stratigraphies of Louisiana, Mississippi, and Alabama.
Task 2: High-Resolution MODFLOW model development
Task 2-1: Co-PI Tsai has created an ArcGIS shapefile of the proposed groundwater model domain including the continental shelf. The model will include Mississippi River alluvial aquifer (MRAA), Wilcox aquifer, and Claiborne aquifer in the MERAS domain, and MRAA, Chicot, Evangeline, and Jasper aquifers in the CLAS domain.
Task 2-2: GCP groundwater model will be built on the USGS National Hydrogeologic Grid. The NHG cell size is 1 km. Using the NHG will allow us to compare our groundwater model to the USGs groundwater models. Co-PI Tsai has developed an NHG shapefile of the GCP groundwatermodel.
Task 2-3: GCP groundwater model will be built on the USGS National Hydrogeologic Grid (NHG) The NHG cell size is 1 km. Using the NHG will allow us to compare our groundwater model to the USGS groundwater models under their Regional Groundwater Availability Studies.
Task 2-4: Convert the 3D stratigraphy model into a MODFLOW 6 unstructured grid. (LSU and SU) We imported VTU files into Aquaveo’s GMS to generate DISU files for MODFLOW-USG and MODFLOW 6 unstructured grids.
Task 2-5: Add water use data, stream data, and recharge data (MODULE 1) into MODFLOW 6 model. (LSU, UM, UA) Co-PI Tsai will work with UA and UM teams to add water-use data, stream data, and recharge data to the MODFLOW 6 model.
Task 2-6: Use groundwater level data and GRACE-derived groundwater storage changes data to calibrate the MODFLOW 6 model. The next step is to remove surface water storage changes in the GRACE data. Then, we will analyze GRACE-based groundwater storage change in the GCP model domain.