Dataset Card for SAR-Duty-Cycle project

This is a preliminary training and validation dataset for sub-aperture reconstruction using generative AI.

Dataset Details

Dataset Description

• Curated by: [More Information Needed]
• Funded by [optional]: [More Information Needed]
• Shared by [optional]: [More Information Needed]
• Language(s) (NLP): [More Information Needed]
• License: [More Information Needed]

Dataset Sources [optional]

• Repository: [More Information Needed]
• Paper [optional]: [More Information Needed]
• Demo [optional]: [More Information Needed]

Uses

This dataset is intended for use in training and validating a generative AI model able to reconstruct SAR sub aperture images.

Direct Use

[More Information Needed]

Out-of-Scope Use

Dataset Structure

[More Information Needed]

Dataset Creation

Curation Rationale

[More Information Needed]

Source Data

11 Sentinel-1 SAR strip map images processed to level-1 were used as source imagery. They were selected for the expected presence of anisotropic features (buildings, mountains, etc) and so as to include imagery from different seasons, years and sensors (Sentinel-1 A, B, and C) All source imagery is available through the Copernicus Data program: https://browser.dataspace.copernicus.eu . The Phi-Down module was used for downloading.

Data Collection and Processing

500 random patches were extracted from each source image. Five sets of sub-look images were produced for each patch (3, 5, 7, 9, and 11 sublooks) using the subsubs module. Relevant metadata from the original source image, spatial subsetting, and spectral sublooking are aggregated and included with each sample.

Who are the source data producers?

This dataset was created by Gillian S.L. Rowan with input from Dr. Roberto Del Prete, Dr. Nicolas Longepe, and Dr. Lorenzo Papa at the European Space Agency in Frascati, Italy.

Personal and Sensitive Information

No personal or sensitive data is contained in this dataset.

Bias, Risks, and Limitations

The dataset contains only SAR images from the Sentinel-1 constellation. Input images were not randomly selected and do not cover the full diversity of possible EO scenes. Instead, they were selected to ensure the presence of anisotropic features such as building or mountains. Majority water or flat scenes were explicitly excluded. Further, due to the timing of data retrieval, only one Sentinel 1C image was included as it's data catalogue is substantially smaller than sensors A/B.

Recommendations

This dataset should not be used as a sole data source for models or analyses of global extents as it was not designed to be globally representative.

Citation [optional]

Publication describing the full dataset production is forthcoming. This section will be updated upon publication.

BibTeX:

[More Information Needed]

Glossary [optional]

Sub-look: SAR image of degraded resolution in the azimuth direction created by considering a fraction of the available Doppler frequency spectrum available for each image line. This post-hoc sub-looking creates an image equivalent to what would be produced with a shorter synthetic aperture.

Dataset Card Authors [optional]

Gillian S.L. Rowan

Dataset Card Contact

< ! -- contact information. -->