🌊 Deep coastal sea elements forecasting using U-Net based models ========

The official code of the following paper : https://arxiv.org/pdf/2011.03303.pdf

📊 Results -----

Some animation of the actual vs prediction of the AsymmInceptionRes-3DDR-UNet model, using a 48h ahead prediction:

Variable	Actual Vs Prediction
Sea Surface Height
Sea Water Salinity
Eastward Current Velocity
Northward Current Velocity

Note: the dark area is made of pixels that correspond to the land.

💻 Installation -----

The required modules can be installed via:

pip install -r requirements.txt

Quick Start

To launch the training, please run:

python train_selected_model.py 

📜 Scripts -----

• The scripts contain the models, the data preprocessing, as well as the training files.

🔍 Models -----

We show here the schema related to the AsymmInceptionRes-3DDR-UNet model.

📂 Data -----

In order to download the data or any of the trained models, please email to the following address:

s.mehrkanoon@uu.nl

The data must be downloaded and unzipped inside the 'Data/' directory.

🔗 Citation -----

If you use our data and code, please cite the paper using the following bibtex reference:

@article{fernandez2022deep,
  title={Deep coastal sea elements forecasting using UNet-based models},
  author={Fern{\'a}ndez, Jes{\'u}s Garc{\'\i}a and Abdellaoui, Ismail Alaoui and Mehrkanoon, Siamak},
  journal={Knowledge-Based Systems},
  pages={109445},
  year={2022},
  publisher={Elsevier}
}