Jingwei: Hybrid Graph Learning Reconstructs Global Ocean Oxygen Spatiotemporal Changes

Requirements

• python 3.9
• pytorch 2.1+
• torch-geometric 2.4.0
• torch-scatter 2.1.2
• torch-sparse 0.6.18
• numpy 1.24.1

How to Train and Test Jingwei

To run the code, simply execute the following command:

python main.py

This will run the code with the default parameters. If you want to customize the parameters, you can pass them via the command line. Below are the available command-line arguments you can set.

Argument	Type	Default	Description
--batch_size	int	32	Batch size for training.
--max_patience	int	10	Maximum patience for early stopping.
--lr	float	0.1	Learning rate for training.
--num_epochs	int	1000	Number of epochs for training.
--gpu	int	0	GPU index to use for training (0 for the first GPU).
--geo_dim	int	5	Dimension of the DO geo factor.
--hidden_dim	int	64	Dimension of the hidden layer.
--time_length	int	11	Length of the time series.

Visualization of Reconstruction Result @YouTube

Multi-Source Observation Dataset

In our work, we collect comprehensive dissolved oxygen observation data from five publicly available databases as follows.

Database	Time	Institution	Source	Access Date
World Ocean Database (WOD18)	1900-2023	National Centers for Environmental Information	https://www.ncei.noaa.gov/	2023-05
CLIVAR and Carbon Hydrographic Database (CCHDO)	1922-2023	CLIVAR and Carbon Hydrographic Data Office	https://cchdo.ucsd.edu/	2023-05
Argo	2001-2023	Argo Global Data Assembly Center	https://argo.ucsd.edu/	2023-05
Global Ocean Data Analysis Project version2.2022 (GLODAPV2_2022)	1972-2021	NOAA’s National Centers for Environmental Information (NCEI)	https://glodap.info/	2023-05
Geotraces IDP	2007-2018	GEOTRACES International Data Assembly Centre (GDAC)	https://www.geotraces.org	2023-10

The Name of "Jingwei"

Jingwei is a classic figure in Chinese mythology, featured in the "Shan Hai Jing". The story tells of Jingwei, the daughter of Emperor Yan, who drowned in the East Sea. She was reborn as a bird and decided to fill the sea with pebbles and twigs, endeavoring to prevent similar tragedies. Today, Jingwei symbolizes perseverance and determination, embodying the spirit of never giving up despite difficult challenges.

In our work, we thank all marine scientists, researchers, and technicians who tirelessly venture into the field to measure oceanic dissolved oxygen data. Although their collected data might seem as extremely sparse as the pebbles and twigs Jingwei used, their persistent efforts aim to reveal the patterns of global ocean deoxygenation. To honor the spirit of these ocean explorers, we have named the AI-driven algorithm for reconstructing ocean deoxygenation proposed in this paper "Jingwei" and designed its logo as follows.

Meet the Team

Our project is supported by a diverse and talented group of experts from various fields. The team is divided into two main groups: the Information Science Team and the Oceanography Team. Below, you will find an overview of the members in each group, along with their academic backgrounds, research interests, and current roles.

Information Science Team

Name	Position	Research Focus	Affiliation
Xinbing Wang	Distinguished Professor	Big Data, Knowledge Graph	Shanghai Jiao Tong University
Xiaoying Gan	Professor	Data Mining, Crowd Computing	Shanghai Jiao Tong University
Luoyi Fu	Associate Professor	Data-driven IoT, Graph Network	Shanghai Jiao Tong University
Meng Jin	Associate Professor	IoT, AI for Science	Shanghai Jiao Tong University
Bin Lu	PhD Student	Graph Neural Network, GeoAI	Shanghai Jiao Tong University
Ze Zhao	Master Student	Knowledge Graph	Shanghai Jiao Tong University
Haonan Qi	Master Student	AI for Ocean	Shanghai Jiao Tong University

Oceanography Team

Name	Position	Research Focus	Affiliation
Jing Zhang	Academician of CAS, Professor	Biogeochemistry and Chemical Oceanography	East China Normal University
Chenghu Zhou	Academician of CAS, Professor	Geographic Information Systems, Spatio-temporal Data Mining	The Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences
Lei Zhou	Distinguished Professor	Ocean and atmosphere dynamics	Shanghai Jiao Tong University
Lixin Qu	Tenure-track Associate Professor	Ocean submesoscale processes, Artificial intelligence in oceanography	Shanghai Jiao Tong University
Yuntao Zhou	Associate Professor	Oceanic Oxygen, Climate Change, Statistics and Geostatistics	Shanghai Jiao Tong University
Luyu Han	Master Student	Climate Change, Artificial intelligence in oceanography	Shanghai Jiao Tong University, University of California, San Diego
Jingjing Shen	Master Student	Climate Change, Artificial intelligence in oceanography	Shanghai Jiao Tong University

Contact

Bin Lu ([email protected])