Skip to content

Latest commit

 

History

History
187 lines (105 loc) · 7.32 KB

README_old.md

File metadata and controls

187 lines (105 loc) · 7.32 KB
Raw

Introduction

This is a short course for the HexWatershed model.

HexWatershed: a mesh independent flow direction model for hydrologic models.

Spatial discretization is the cornerstone of all spatially-distributed numerical simulations including watershed hydrology. Traditional square grid spatial discretization has several limitations:

  1. It cannot represent adjacency uniformly;

  2. It leads to the “island effect” and the diagonal travel path issue in D8 scheme;

  3. It cannot provide a spherical coverage without significant spatial distortion;

  4. It cannot be coupled with other unstructured mesh-based models such as the oceanic models.

Therefore, we developed a watershed delineation model (HexWatershed) based on the hexagon mesh spatial discretization.

We further improve HexWatershed to fully unstructured mesh-based to support variable-resolution meshes such as the MPAS mesh.

For more details, please refer to the HexWatershed documentation (https://hexwatershed.readthedocs.io/).

Requirements

You need internet connection and several tools to run the examples in the tutorial.

To download the model and the tutorial repository, you need:

  • git, or download as a zip file.

The whole HexWatershed package includes both the C++ backend and Python frontend. To compile and install the C++ backend, you need:

  • C++ compiler, i.e., g++ 8.1.0 and above
  • cmake 3.10 and above

To install the Python frontend, you need:

  • conda 4.10 and above (anaconda or miniconda)

To run the Python examples in this tutorial, you need

  • Visual Studio Code with the Python extension.

You need addition tools (e.g., QGIS) to visualize some of the model results. Depending on your system, these tools can be obtained from these resources:

MacOS Ubuntu HPC
git sudo apt install git-all
Homebrew https://brew.sh/
g++ brew install gcc sudo apt-get install g++ module load gcc
cmake brew install cmake https://cmake.org/download/ module load cmake
conda https://docs.conda.io/en/latest/miniconda.html https://docs.conda.io/en/latest/miniconda.html module load anaconda3
VS Code https://code.visualstudio.com/ https://code.visualstudio.com/ https://code.visualstudio.com/

Step-by-step instruction

  1. Download additional data files using an internet browser (Chrome recommended)

Download the dem1.tif and lnd_cull_mesh.nc files from the following url:

https://rcdemo.pnnl.gov/workshop/

  1. Install the HexWatershed backend C++ component

  • git clone https://github.com/changliao1025/hexwatershed.git

  • cd hexwatershed/build

  • If you are on MacOS, it is recommended to use the Homebrew to setup the g++ and cmake.

    cmake CMakeLists.txt -DCMAKE_CXX_COMPILER=g++-11

    Your homebrew installed g++ may have different versions, check it using brew info gcc. If your g++ is not in the system path, you may need to update/fix using brew install gcc.

    If you are on Linux, and the correct g++ is already in the system path

    cmake CMakeLists.txt

  • make install

  1. Install the HexWatershed frontend Python package

    Create/activate a conda environment and install Python packages

  • conda config --set channel_priority strict

  • conda create --name hexwatershed python=3.8

  • conda activate hexwatershed

  • conda install -c conda-forge gdal=3.2

    you can test whether gdal is working using from osgeo import gdal in a Python session

  • conda install -c conda-forge hexwatershed

Because the GDAL library is used by this project and the proj library is often not configured correctly automatically. On Linux or Mac, you can set it up using the .bash_profile such as:

Anaconda:

export PROJ_LIB=/people/user/.conda/envs/hexwatershed/share/proj

Miniconda:

export PROJ_LIB=/opt/miniconda3/envs/hexwatershed/share/proj

  1. Download this tutorial

git clone https://github.com/changliao1025/hexwatershed_tutorial.git

You need to copy the compiled hexwatershed binary file into the bin folder.

You need to copy the downloaded data files into the input folder.

  1. Run the examples within the example folder

  • You need to edit the template configuration json file to match with your data set paths.

  • Depending where you downloaded the data and the example, different configurations are required.

  1. Output files are stored within both pyflowline (conceptual river network) and hexwatershed (flow direction, etc.) folders.

  • Visualize the output geojson files using QGIS.

  • The hexwatershed.json file contains all the flow routing parameters.

Behind the scene

In general, HexWatershed run the following the algorithms step-by-step.

  1. Flowline simplication
  2. Mesh generation
  3. Topology reconstruction
  4. Elevation resampling
  5. Stream burning
  6. Depression filling
  7. Slope calculation
  8. Flow direction
  9. Flow accumulation
  10. River networks
  11. Export outputs

Miscellaneous

  1. Why a hybrid Python and C++ approach?

    Answer: HexWatershed can be run at both regional and global scale, so performance is a factor. Data I/O is much easier in Python so users won't have to build NetCDF or GDAL from the source code.

  2. What if my model doesn't produce the correct or expected answer?

    Answer: There are several hidden assumptions within the workflow. For example, if you provide the DEM and river network for two different regions, the program won't be able to tell you that. A visual inspection of your data is important.

    Optionally, you can turn on the iFlag_debug option in the configuration file to output the intermediate files.

  3. Most common issues:

    • conda cannot create environment, turn off the VPN or bypass it.

    • GDAL not found, consider using the conda-forge channel or use an earlier version such as 3.2.

    • proj related issue OSGeo/gdal#1546, make sure you correctly set up the PROJ_LIB

Learn more

  1. JIGSAW is the mesh generator that is used to generator the variable resolution meshes.

  2. Other meshes such as DGGrid will be supported.

  3. The depression filling algorithm is modified based on the RichDEM priority-flood depression filling method.

  4. HexWatershed can be run at both regional and global scale, see this Youtube clip for example: Global scale HexWatershed simulation

References