-
Notifications
You must be signed in to change notification settings - Fork 122
Home
OpenDrift is a software for modeling the trajectories and fate of objects or substances drifting in the ocean, or even in the atmosphere.
OpenDrift is open source, and is programmed in Python. As the software is very generic, it is rather a "framework" than a "trajectory model" in the traditional sense. Trajectory models for specific purposes (e.g. oil drift, search and rescue, larvae drift etc) may reuse all common functionality from the core model, and need only implement a Python Class describing the purpose-specific processes (physics/biology etc). See Requirements and Data model for more detailed information.
A journal paper about OpenDrift is published in Geoscientific Model Development. A paper describing the technical details of the oil spill module OpenOil is published in Ocean Science.
- Open source, GPL v2 license, providing full transparency - no black boxes
- Platform independent, runs on Linux, Mac/OS X, Windows
- Python2 and Python3 compliant
- Fast - typical simulation time is ~30 seconds for a 66 hour simulation with 1000 particles
- Modular, may simulate transport and fate of any kind of of particles (oil, ships, persons, icebergs etc.)
- Simple to make new modules, based on existing modules or blank template
- May use input forcing data (e.g. current, wind and waves) from any model, in any file format and any map projection
- avoids need to preprocess driver data
- map reprojection and vector rotation performed on-the-fly, using PyProj library
- may use input forcing from remote datasets (e.g. Thredds)
- May use backup driver models (e.g. current, wind, waves) for robustness if first choice is not available
- May run backwards in time by simply specifying a negative time step
- Output saved to CF-compliant netCDF files (but export modules may be written for other formats).
- Basic graphical user interface (presently supports only OpenOil and Leeway)
- Input from ensemble models
OpenDrift is under active development, and issues may occur. Please submit a bug report if you encounter a problem.
There are currently two active branches of OpenDrift, stable
which supports Python 2, and master
which will soon loose Python 2 support.
OpenDrift can be installed through PyPA, conda or directly from the source repository. The latter is preferred if you expect to make changes to the OpenDrift models or code base.
Consider setting up an exclusive virtual environment for OpenDrift.
$ pip install opendrift
- Install miniconda3
- Set up environment and install OpenDrift:
$ conda env create -n opendrift -c opendrift -c conda-forge -c noaa-orr-erd python=3 opendrift
- Check out source
$ git clone https://github.com/OpenDrift/opendrift.git
-
Use either conda or pip to install dependencies: a)
sh conda env create --file conda_python3.yml
b)sh pip install -r requirements.txt
-
Install OpenDrift locally:
$ pip install -e . # from source root
The latest stable version of OpenDrift can be downloaded from here or with Git:
git clone --branch stable https://github.com/OpenDrift/opendrift.git
The easiest and recommended way to install the dependencies is to use Miniconda. Note that the oil drift module (OpenOil) depends on NOAA ADIOS OilLibrary, which is not yet compliant with Python3. Therefore Python2 (and Miniconda2) has to be used for oil drift simulations with OpenDrift.
The dependencies may then be installed with a single command:
conda env create -f <env-file>
with:
<env-file> |
Usage |
---|---|
conda_python3.yml | For Python3, without NOAA ADIOS oil library (see above) |
conda_python2_oil.yml | For Python2, including NOAA ADIOS oil library |
conda_python2.yml | Python2 without NOAA ADIOS oil library |
The installed environment must then be activated with (see output from previous command):
conda activate <env_name>
To permanently activate this environment, the above line may e.g. be added to your .bahrc file (on Linux).
Finally, OpenDrift may be installed with the command:
python setup.py develop --user
The --user
flag can be omitted if you want OpenDrift to be available for other users on the same machine.
By including the keyword develop
it is not necessary to re-install OpenDrift after each update (git pull
).
To be able to run the commandline scripts, add the folder scrips to your PATH, e.g in .bashrc:
export PATH=$PATH:$OPENDRIFT_FOLDER/opendrift/scripts/
OpenDrift has been tested on various Linux, Mac and Windows operating systems.
To test that OpenDrift works as expected, run the script testall
in the root folder. On Windows, tests can be run from dos prompt in the main opendrift-folder with the following command:
python -m unittest discover tests -p test_*.py
The last line of the output should report that all tests have been executed successfully.
For a demonstration of various capabilities, run some of the executable example scripts located in the subfolder examples. Some of the scripts use sample netCDF files provided in the test_data folder, and others obtain data from online resources (Thredds). If this works well, you may play with the model by editing the example scripts.
Here is a detailed explanation of the steps involved when running the model.
OpenDrift is licensed under GPL v2.0
If you have found OpenDrift useful for your study, please cite it as:
For the oil spill module OpenOil, please cite in addition to above:
Please create an issue or contact knutfd@met.no for any questions.
Click here if you want to receive emails about important changes or releases.