Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Recurso digital |
| Language: | |
| Published: |
Zenodo
2025
|
| Subjects: | |
| Online Access: | https://doi.org/10.5281/zenodo.15255691 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866902137498238976 |
|---|---|
| author | Wei, Yuxin Liu, Bin Li, Xiaofeng Jiang, Bohui |
| author_facet | Wei, Yuxin Liu, Bin Li, Xiaofeng Jiang, Bohui |
| contents | <ul> <li> <h2><strong>Dataset Overview</strong></h2> <p>This dataset consists of 256×256 pixel SAR and optical image patches, designed to support water-body analysis before and after flooding events. It provides 30-meter resolution data from both pre-flood and post-flood periods, separated into distinct collections for easier comparison. The dataset is intended for training and testing deep learning models focused on water segmentation, specifically targeting the identification of water bodies in flood-affected regions. By including both SAR and optical data, the dataset enables enhanced futher flood mapping and monitoring, which is crucial for effective disaster management and response.</p> </li> <li> <h2><strong>Image Size</strong></h2> <p>All image patches are 256×256 pixels, ensuring consistency and allowing for straightforward application in machine learning models, particularly for convolutional neural networks (CNNs) and other image classification tasks.</p> </li> <li> <h2><strong>File Naming Convention</strong></h2> <p>The dataset follows a consistent naming convention to maintain organization and ease of use. Each file is named according to the pattern:</p> <p><code>{LOCATION_CODE}_{ACQUISITION_START_TIME}_{ACQUISITION_END_TIME}*.tif</code></p> <p>This structured naming facilitates quick identification and processing of the data based on the location and time of acquisition.</p> </li> <li> <h2><strong>Mainly Band / File Types</strong></h2> <p>The dataset includes various file types representing different bands and indices:</p> <ul> <li> <p><code>*_432.tif</code><br>Sentinel-2 true-color composite (bands 4, 3, 2), which is used as a visual reference for manual labeling of water bodies.</p> </li> <li> <p><code>*_843.tif</code><br>Sentinel-2 false-color composite (bands 8, 4, 3), which helps highlight vegetation and water signatures, useful for labeling water bodies.</p> </li> <li> <p><code>*_NDWI.tif</code><br>Normalized Difference Water Index (NDWI), derived from Sentinel-2, which enhances water body boundaries, making it easier to delineate water regions.</p> </li> <li> <p><code>*_VV.tif</code><br>Sentinel-1 SAR VV polarization (linear backscatter), the primary input for flood detection, capturing flood-induced changes in surface structure.</p> </li> <li> <p><code>*_VH.tif</code><br>Sentinel-1 SAR VH polarization (cross-polarized), which provides complementary surface scattering information, enhancing flood detection accuracy.</p> </li> <li> <p><code>*_VV_vis.tif</code><br>VV polarization rescaled to an 8-bit unsigned integer (uint8, 0–255) for easier visualization.</p> </li> <li> <p><code>*_VH_vis.tif</code><br>VH polarization rescaled to an 8-bit unsigned integer (uint8, 0–255) for easier visualization.</p> </li> </ul> </li> </ul> |
| format | Recurso digital |
| id | zenodo_https___doi_org_10_5281_zenodo_15255691 |
| institution | Zenodo |
| language | |
| publishDate | 2025 |
| publisher | Zenodo |
| record_format | zenodo |
| spellingShingle | Dual-Polarization Synthetic Aperture Radar (SAR) for Water Body Segmentation: A Dataset Collected in Bangladesh from 2015 to 2024 Wei, Yuxin Liu, Bin Li, Xiaofeng Jiang, Bohui SAR Flood Segmentation Deep Learning/classification <ul> <li> <h2><strong>Dataset Overview</strong></h2> <p>This dataset consists of 256×256 pixel SAR and optical image patches, designed to support water-body analysis before and after flooding events. It provides 30-meter resolution data from both pre-flood and post-flood periods, separated into distinct collections for easier comparison. The dataset is intended for training and testing deep learning models focused on water segmentation, specifically targeting the identification of water bodies in flood-affected regions. By including both SAR and optical data, the dataset enables enhanced futher flood mapping and monitoring, which is crucial for effective disaster management and response.</p> </li> <li> <h2><strong>Image Size</strong></h2> <p>All image patches are 256×256 pixels, ensuring consistency and allowing for straightforward application in machine learning models, particularly for convolutional neural networks (CNNs) and other image classification tasks.</p> </li> <li> <h2><strong>File Naming Convention</strong></h2> <p>The dataset follows a consistent naming convention to maintain organization and ease of use. Each file is named according to the pattern:</p> <p><code>{LOCATION_CODE}_{ACQUISITION_START_TIME}_{ACQUISITION_END_TIME}*.tif</code></p> <p>This structured naming facilitates quick identification and processing of the data based on the location and time of acquisition.</p> </li> <li> <h2><strong>Mainly Band / File Types</strong></h2> <p>The dataset includes various file types representing different bands and indices:</p> <ul> <li> <p><code>*_432.tif</code><br>Sentinel-2 true-color composite (bands 4, 3, 2), which is used as a visual reference for manual labeling of water bodies.</p> </li> <li> <p><code>*_843.tif</code><br>Sentinel-2 false-color composite (bands 8, 4, 3), which helps highlight vegetation and water signatures, useful for labeling water bodies.</p> </li> <li> <p><code>*_NDWI.tif</code><br>Normalized Difference Water Index (NDWI), derived from Sentinel-2, which enhances water body boundaries, making it easier to delineate water regions.</p> </li> <li> <p><code>*_VV.tif</code><br>Sentinel-1 SAR VV polarization (linear backscatter), the primary input for flood detection, capturing flood-induced changes in surface structure.</p> </li> <li> <p><code>*_VH.tif</code><br>Sentinel-1 SAR VH polarization (cross-polarized), which provides complementary surface scattering information, enhancing flood detection accuracy.</p> </li> <li> <p><code>*_VV_vis.tif</code><br>VV polarization rescaled to an 8-bit unsigned integer (uint8, 0–255) for easier visualization.</p> </li> <li> <p><code>*_VH_vis.tif</code><br>VH polarization rescaled to an 8-bit unsigned integer (uint8, 0–255) for easier visualization.</p> </li> </ul> </li> </ul> |
| title | Dual-Polarization Synthetic Aperture Radar (SAR) for Water Body Segmentation: A Dataset Collected in Bangladesh from 2015 to 2024 |
| topic | SAR Flood Segmentation Deep Learning/classification |
| url | https://doi.org/10.5281/zenodo.15255691 |