Python has awesome packages for machine learning that can be coupled to groundwater models and perform automatic calibration of hydraulic parameters. This tutorial covers the procedure to implement a neural network based on a set of parameters set with corresponding head values; the study case is on a transient pumping test model with an observation point located around 10 meters away from the pump well. The analysis predicts the calibration parameters from the spatially interpolated observed head values. 

We did this tutorial since the documentation or examples on the topic were not available. Flopy can export the model grid and model attributes to shapefile with a coordinate system of reference for the three types of discretizations of MODFLOW 6. We have done an applied example to export the grid of a discretized by vertices model (DISV) to the ESRI Shapefile model. The tutorial also show options for the final grid representation in geopandas.

Having a geological model can enhance numerical models since it allows to represent higher accuracy the potential distribution of hydraulic parameters in the horizontal and vertical direction. The process to implement/insert a geological model into a Modflow model is a challenge due to restrictions on proprietary software and spatial tools; we have done the whole procedure to insert a geological model into a MODFLOW-NWT groundwater model with scripts in Python, Pyvista and others.

The code extracts the cell centroids of the modflow model and then compares its position with the different geological units exported from Leapfrog to Vtk. Once the corresponding lithology of the cell is identified a hydraulic parameter is assigned. The tutorial works from Vtks of a geological model done in Leapfrog, but it can work with any Vtk.

This tutorial shows the complete procedure to convert a geological unit as a Leapfrog mesh (*.msh) to the VTK (Visualization Toolkit) format using Python and GemGIS. Follow these steps to seamlessly transfer your geological data for advanced visualization, analysis and comparison with other model results.

We have done an applied groundwater model for a pumping test on a confined aquifer. The wellsite is located on the Clairborne aquifer (Georgia, USA) and has two observations on different layers besides the well itself. The aquifer test has two periods of pumping and recovery that last 3 days each one and the numerical model was constructed with MODFLOW-6 based on a Python script with the FloPy package. Comparison among observed and simulated wells were done with plots in Matplotlib.

Modflow 6 has changed in many ways the traditional concepts of groundwater modeling; now a group of models can run under a "simulation" and these models can interchange flow and transport. Under this structure Flopy has implemented a tool to split a groundwater model into several models where their results can be reconstructed and compared to the original entire models. Capabilities and features like this make Flopy and Modflow 6 a great tool for developing advanced groundwater models that simulate complicated tasks or requirements to the groundwater flow regime.