Oil spill monitoring workflow

• - This is a Deep Learning based oil detection workflow
• - Examples of revealing oil spills over China Seas
• - Check the paper for details:
• - Jiao, J., Lu, Y., and Hu, C., 2024. Characterizing oil spills using deep learning and spectral-spatial-geometrical features of HY-1C/D CZI images. Remote Sensing of Environment. 308, 114205.

Expamples of oil spill detection

Optical image

Oil footprint

Model structure

Model structure

The oil optical detection Convolutional Network (OODCNet) is based on U-Net structure

Training dataset

Training dataset structure

It includes three parts: optical images with four spectral bands, corresponding oil labels, and calculated viewing angles

Here θ_m is calculate as:

cosθ_m = cosθ₀×cosθ - sinθ₀×sinθ×cos(φ₀-φ)

We also share the twin-satellite observations here.

Validation results

Validation results

Oil spills over China Seas

Oil spills over China Seas

The oil spills over China Seas from 2018-2021

Time-series monitoring with SAR and Optical observations

Time-series monitoring with SAR and Optical

Combining SAR and Optical sensors to improve oil spill monitoring