iGW-TU-Dresden/CFPy

CFPy - a Python package for the generation of MODFLOW-CFP specific input files (.cfp, .crch, .coc). CFPy is the FloPy-equivalent package for MODFLOW-CFP.

CFPy

CFPy - a Python package for the generation of MODFLOW-CFP specific input files (.cfp, .crch, .coc). CFPy is the FloPy-equivalent package for MODFLOW-CFP.

Check the CFPy paper here! Please cite us if you are using CFPy in your research.

Contributing:

If you want to contribute to the package (e.g., changing the code or adding examples), please generally use the dev branch for commits (especially for changes to the source code!). New examples can also be directly committed to the main branch.

Dependencies:

• python >= 3.9, < 3.11 (without pyKasso), == 3.9 (with pyKasso)
• numpy >= 1.18.5, <1.25.0
• matplotlib >= 3.3.4, <3.6.0
• pandas >= 1.2.1
• flopy >= 3.3.3
• watermark >= 2.3.0
• pykasso == 0.1.0 (OPTIONAL)

See the beginning of the individual example notebooks for more information on Python and package versions.

Installation

The easy way

pip install CFPy-TUD

This installs CFPy quickly and easily, however, without pyKasso. For instructions on how to install pyKasso as well, see below.

The custom way

The installation is described specifically for using the Anaconda distribution of Python / using conda environments. If you encounter an OpenSSL-related error during one of the following steps, go to .../anaconda3/lib/bin, copy libcrypto-1_1-x64.dll and libssl-1_1-x64.dll and paste them to .../anaconda3/dlls, and continue regularly afterwards.

Installation of CFPy only (without pyKasso)

For experienced users

Install CFPy from source in a (new) environment with Python >= 3.9.

For inexperienced users

• Download the CFPy source code here
  • Make sure that you are in the main branch (see upper left of the page)
  • Download the files by pressing the green "Code" button in the upper right of the page and selecting "Download ZIP"
  • Unpack the ZIP (it should be called CFPy-main.zip) on your machine
  • Get the directory path where the unpacked CFPy source files are now stored (e.g., C:/Users/.../CFPy-main) by checking the directory of the file setup.py in the CFPy-main directory
• Create a new environment for CFPy
  • Open Anaconda PowerShell Prompt on your machine
  • To create a new environment, type conda create -n cfpy_env python=3.9 and press [Enter] (if user input is asked - typically for confirming the installation, press [y] and [Enter] to confirm)
  • Activate the environment by typing conda activate cfpy_env and pressing [Enter] (and do NOT close Anaconda PowerShell Prompt)
• Install CFPy
  • Remain in Anaconda PowerShell Prompt with the cfpy_env environment activated
  • To install CFPy, type pip install -e <YourPathToCFPy> and replace <YourPathToCFPy> with your path to CFPy source files from earlier, e.g., C:/Users/.../CFPy-main, and press [Enter] (if user input is asked - typically for confirming the installation, press [y] and [Enter] to confirm)
  • Optionally, you can install Jupyter Notebooks / Jupyter Lab (all CFPy examples are in this format) in the cfpy_env environment by typing conda install -c conda-forge jupyterlab and pressing [Enter] (if user input is asked - typically for confirming the installation, press [y] and [Enter] to confirm); make sure that the cfpy_env is still activated

You are now ready to use CFPy! To use CFPy in a Jupyter Notebook (typical application) at any time:

• Open Anaconda PowerShell prompt
• Activate the environment via the command conda activate cfpy_env
• Open Jupyter Lab via the command jupyter lab

Installation of CFPy with pyKasso

For experienced users

• create a new / activate an existing environment environment with Python == 3.9.
• Download pyKasso source files from the cfpy branch here
• Create a new environment from the environment.yml
• Install karstnet in this environment (from source)
• Install CFPy in this environment (from source)

For inexperienced users

• Download the pyKasso source code from here
  • Make sure that you are in the main branch (see upper left of the page)
  • Download the files by pressing the green "Code" button in the upper right of the page and selecting "Download ZIP"
  • Unpack the ZIP (it should be called pyKasso-cfpy.zip) on your machine
  • Get the file path of the ènvironment.yml from the unpacked pyKasso source directory (e.g., C:/Users/.../pyKasso-cfpy/environment.yml
  • Get the directory path where the unpacked pyKasso source files are now stored (e.g., C:/Users/.../pyKasso-cfpy) by checking the directory of the file setup.py in the pyKasso-cfpy directory
• Create a new environment for pyKasso
  • Open Anaconda PowerShell Prompt on your machine
  • To create a new environment from the environment.yml file, type conda create -f <YourPathTo/environment.yml> (where you replace <YourPathTo/environment.yml> with the path you got before) and press [Enter] (if user input is asked - typically for confirming the installation, press [y] and [Enter] to confirm)
  • Activate the environment by typing conda activate pykasso2D and pressing [Enter] (and do NOT close Anaconda PowerShell Prompt)
• Download and install karstnet
  • Download the karstnet yource files from here by pressing the green "Code" button in the upper right of the page and selecting "Download ZIP"
  • Unpack the ZIP (it should be called karstnet-master.zip) on your machine
  • Get the directory path where the unpacked karstnet source files are now stored (e.g., C:/Users/.../karstnet-master) by checking the directory of the file setup.py in the karstnet-master directory
  • Remain in Anaconda PowerShell Prompt with the pykasso2D environment activated
  • To install karstnet, type pip install -e <YourPathTokarstnet> and replace <YourPathTokarstnet> with your path to karstnet source files from earlier, e.g., C:/Users/.../karstnet-master, and press [Enter] (if user input is asked - typically for confirming the installation, press [y] and [Enter] to confirm) and do NOT close Anaconda PowerShell Prompt
• Install pyKasso
  • Remain in Anaconda PowerShell Prompt with the pykasso2D environment activated
  • To install pyKasso, type pip install -e <YourPathTopyKasso> and replace <YourPathTopyKasso> with your path to pyKasso source files from earlier, e.g., C:/Users/.../pyKasso-cfpy and press [Enter] (if user input is asked - typically for confirming the installation, press [y] and [Enter] to confirm) and do NOT close Anaconda PowerShell Prompt
• Download and install CFPy
  • Download the CFPy yource files from here by pressing the green "Code" button in the upper right of the page and selecting "Download ZIP"
  • Unpack the ZIP (it should be called CFPy-main.zip) on your machine
  • Get the directory path where the unpacked CFPy source files are now stored (e.g., C:/Users/.../CFPy-main) by checking the directory of the file setup.py in the CFPy-main directory
  • Remain in Anaconda PowerShell Prompt with the pykasso2D environment activated
  • To install CFPy, type pip install -e <YourPathToCFPy> and replace <YourPathToCFPy> with your path to CFPy source files from earlier, e.g., C:/Users/.../CFPy-main, and press [Enter] (if user input is asked - typically for confirming the installation, press [y] and [Enter] to confirm)
  • Optionally, you can install Jupyter Notebooks / Jupyter Lab (all CFPy examples are in this format) in the pykasso2D environment by typing conda install -c conda-forge jupyterlab and pressing [Enter] (if user input is asked - typically for confirming the installation, press [y] and [Enter] to confirm); make sure that the pykasso2D is still activated

You are now ready to use CFPy together with pyKasso! To use CFPy with pyKasso in a Jupyter Notebook (typical application) at any time:

• Open Anaconda PowerShell prompt
• Activate the environment via the command conda activate pykasso2D
• Open Jupyter Lab via the command jupyter lab

CFPy_Examples:

Note: make sure you have pyKasso (and its dependencies) installed if you want to use the full functionality of CFPy!

Example 1

• Notebook: EX01_CFPy_FloPy_I.ipynb
• Files: CFPy_EX01_RUN
• Description:
  • simple example for CFP mode 1 from the MODFLOW-CFP documentation, coupling CFPy (MODFLOW-CFP input-file generation) and FloPy (MODFLOW input-file generation)
  • generating CFP input files from a user-defined node network structure
  • model computation with cfpv2

Example 2

• Notebook: EX02_stochastic_pyKasso_FloPy_I.ipynb
• Files: CFPy_EX02_RUN
• Description:
  • complex example coupling pyKasso (network generation), CFPy (MODFLOW-CFP input-file generation), and FloPy (MODFLOW input-file generation)
  • multiple node networks are generated with pyKasso, given a single set of inlet and outlet locations
  • generating CFP input files from the validated networks generated by pyKasso
  • model computation with cfpv2
  • evaluation / initial assessment of simulation uncertainty, taking multiple network realizations into account

Example 3

• Notebook: EX03_pyKasso_CFPy_coupling_I.ipynb
• MODFLOW files: none
• Description:
  • demonstration of the coupling process between pyKasso and CFPy
  • a network is generated with pyKasso and validated with the CFPy.preprocessing module
  • the structure of the CFPy input file (the .nbr-file) is described

Example 4

• Notebook: EX04_CFPy_FloPy_PreExisting_FloPyModel.ipynb
• Files: CFPy_EX04_RUN
• Description:
  • EX04 is very similar to EX01
  • it is shown how one can use a pre-existing FloPy model in CFPy
  • first, a simple MODFLOW model is created with FloPy
  • passing the FloPy model instance to the CFPy preprocessor automatically extracts all the usable information from the FloPy model (such as layer elevations, discretization etc.)
  • this greatly improves the usability of CFPy for pre-existing MODFLOW models (which do not yet include CFP)

Example 5

• Notebook: EX05_CFPy_FloPy_ConduitWell.ipynb
• Files: CFPy_EX05_RUN
• Description:
  • like example EX01 but with a conduit network flow boundary condition (FBC)

Example 6

• Notebook: EX06_CFPy_FloPy_II.ipynb
• Files: CFPy_EX06_RUN
• Description:
  • like example EX01 but with an already existing .nbr file (two node planes)

Example 7

• Notebook: EX07_pyKasso_CFPy_coupling_II.ipynb
• Files: none
• Description:
  • like example EX03 but with some additions (i.e., making the pyKasso-generated network quasi-3D and validating a network with two node planes)

Example 8

• Notebook: EX08_stochastic_pyKasso_FloPy_II.ipynb
• Files: CFPy_EX08_RUN
• Description:
  • like example EX02 but more streamlined with less additional functions (more straightforward application of CFPy)

NOTE

To run the examples, make sure that the CFPy package directory (the CFPy folder) is inside the example directory or that CFPy is installed as a package in the active environment.

Workflow Description

Workflow Description for Coupling Stochastic Network Generation (e.g., pyKasso), CFPy and FloPy to Create a Script-Based Framework for Spatially Distributed Karst Simulation with MODFLOW-CFP

1. Definition of Model Domain

   • define a rectangular (in the 2D case) or cuboid (in the 3D case) model domain together with a spatial discretization (i.e., number and width of rows, number and width of columns, number of layers together with layer elevation information)
   • if a non-rectangular model domain should be used, still start with a rectangular domain; later, after all input data are generated, the model domain can be altered in FloPy by setting cells inactive

2. pyKasso Network Generation

   • include the domain definition from step (1) in the settings.yaml-file for pyKasso

   • define other stochastic network generation options in the settings.yaml

   • generate one (or multiple) karst networks with pyKasso

   • NOTE: Right now, pyKasso can only generate 2D karst networks. However, multiple (vertically connected) node planes can be given to CFPy. To give the 2D network some 3D structure nonetheless, node elevations need to be given to CFPy. Those elevations can be uniform or non- uniform. Also, multiple 2D node planes (with each different uniform or non-uniform node elevations) can be connected vertically. See the pyKasso_CFPy_coupling.ipynb notebook for more information.

3. Coupling pyKasso and CFPy

   • NOTE: CFPy generally relies on one single input-file, where all neccessary information is summarized (domain discretization, number of node planes, MODFLOW layer elevations, node network and elevations). This file can either be generated automatically with the generate_nbr method of the CFPy.preprocessing module (preferred option) or can also be generated manually
   • use the pyKassoValidator object in the CFPy.preprocessing module to validate the node network generated with pyKasso in step (2); provide node elevation information in this step
   • option 1 (preferred option): generate the CFPy input information (the .nbr-file) automatically via the generate_nbr method in the CFPy.preprocessing module (providing the pyKasso.SKS catchment as well as domain discretization and layer elevation information)
   • option 2: or generate the .nbr-file manually (also see the pyKasso_CFPy_coupling.ipynb notebook for more information) by exporting the validated network and copying to the .nbr-file

4. Set Up the MODFLOW-CFP Model with CFPy and FloPy

   • create the model with FloPy (using a MODFLOW-CFP distribution, preferrably the cfpv2.exe version distributed by TU Dresden)
   • create all input-files for CFP with CFPy based on the .nbr-file generated in step (3)
   • create the remaining input-files for MODFLOW with FloPy
   • simulate the model with FloPy (model.run_model method)

5. Post-Processing of Results

   • use FloPy post-processing methods to process classical MODFLOW results such as matrix head information etc.
   • use the methods in the CFPy.postprocessing module to obtain and process CFP-specific results (node- and tube- related data)

NOTE: all steps are shown in the example notebooks!