Groundwater Energy (GWE) Transport Modeling Tutorial

This repository contains a series of Jupyter notebooks demonstrating the integration of MODFLOW-6 groundwater energy transport modeling with PEST++ for parameter estimation and uncertainty analysis. The tutorials are based on the GMDSI Open Loop Low Temperature Geothermal System tutorials. Original materials available at: https://gmdsi.org/blog/gwe_slideshow/

Overview

The tutorial demonstrates how to:

• Set up PEST++ interfaces with unstructured grid models
• Parameterize hydraulic conductivity fields using non-stationary geostatistics informed by conceptual geological knowledge
• Perform surrogate-based data assimilation and predictive uncertainty analysis using data space inversion

Model Description

The model represents a shallow, channelized alluvial deposit system near a river with the following characteristics:

• Single-layer unstructured grid model
• Historical and forecast simulation periods
• Steady-state flow model with transient heat transport
• Open loop shallow geothermal system simulation with seasonal temperature variations

Repository Structure

Notebooks

• 00_model.ipynb - Model introduction and basic setup
• 01_setup_pstfrom.ipynb - PEST++ interface configuration
• 02_priomc.ipynb - Prior Monte Carlo analysis
• 03_dsi.ipynb - Data space inversion
• 04_dsiae.ipynb - Data space inversion with autoencoder

Getting Started

Prerequisites

Create the conda environment:

conda env create -f environment.yml
conda activate gwe

Running the Tutorials

1. Start with 00_model.ipynb to understand the base model
2. Progress through the numbered notebooks in sequence
3. Each notebook builds on concepts from previous tutorials

Note: The initial model run takes approximately 5 minutes on a MacBook Pro. Runing the Prior Monte Carlo notebook can take a while (~60min).