Atmospheric rivers dataset for machine learning training

<p>A thorough description of the data and how it was created can be found:&nbsp;<a title="http://climate-cms.org/CNN-Atmospheric-Rivers/" href="http://climate-cms.org/CNN-Atmospheric-Rivers/" target="_blank" rel="noopener">http://climate-cms.org/CNN-Atmospheric-Rivers/</a></p> <p>A Jupyter notebook has also been created where we'll show you how to use this data to train a deep learning model to identify whether an Integrated Vapor Transport map contains an atmosperhic river. it can be found: <a title="CNN_AR_tutorial.ipynb" href="https://github.com/coecms/CNN-Atmospheric-Rivers/blob/main/CNN_AR_tutorial.ipynb" target="_blank" rel="noopener">CNN_AR_tutorial.ipynb</a> and had been published here:&nbsp;</p> <p>Mesto, M., Hobeichi, S., &amp; Green, S. (2024). CNN-Atmospheric-Rivers (v1.0.0). Zenodo. <a href="https://doi.org/10.5281/zenodo.12538779" target="_blank" rel="noopener">https://doi.org/10.5281/zenodo.12538779</a></p> <p>The data is organised in three folders:</p> <ul> <li>IVT_ERA5_2Deg: Contains global IVT data.</li> <li>AR_Global: Contains polygons representing AR objects identified and hand-labelled in each IVT map.</li> <li>Training_Testing_tiles: Contains tiles of IVT data with an annotation file that classifies each tile as one of the following: &lsquo;Atmospheric River&rsquo;, &lsquo;Ambiguous&rsquo;. The &lsquo;Ambiguous&rsquo; class refers to objects that are not clearly identifiable as atmospheric rivers.</li> </ul> <p><strong>Integrated Vapor Transport maps:</strong></p> <p>These maps were computed using the magnitude of the vertical integral of northward and eastward water vapour flux variables from ERA5. The IVT values are expressed in units of kg m^-1 s^-1. All the IVT TIFF files were loaded into ArcGIS software and displayed using a colour scheme that allows for the visual identification of atmospheric rivers. Data details:</p> <ul> <li>Folder: <strong><em>IVT_ERA5_2Deg</em></strong></li> <li>File format: TIFF</li> <li>Spatial resolution: 2 degrees</li> <li>Spatial coverage: Global (longitude: -180 to 180 , latitude: -90 to &nbsp;90)</li> <li>Geographic Coordinate System: GCS_WGS_1984</li> <li>Temporal coverage: 1^st &ndash; 5^th day of January, April, July, October for 2010, 2013, 2015; these years correspond to La Ni&ntilde;a, neutral, and El Ni&ntilde;o year respectively</li> <li>Temporal resolution: Daily</li> <li>Naming of files: ivt_2deg_ddmmyyyy.tif</li> <li>Number of files: 60 (5 days &times; 4 months &times; 3 years)</li> <li>Number of channels in each file: 1</li> </ul> <p>&nbsp;</p> <p><strong>Atmospheric Rivers in IVT maps:</strong></p> <p>The annotation tool &lsquo;Label Objects for Deep Learning&rsquo; was used to draw polygons to cover the shape of atmospheric rivers on each IVT map. Each polygon was assigned one of two labels: 'Atmospheric Rivers' or 'Ambiguous'. The polygons were drawn based on visual identification of the shape of atmospheric rivers, guided by IVT values close to 500kg m^-1 s^-1 as in Reid et al (2020). The 'Ambiguous' label was assigned to objects that were unclear in their classification as ARs. This ambiguity arose from objects that were shorter, wider, had slightly lower IVT values, or it was hard to tell if they were ARs of tropical cyclones during the early stages of their formation. Data details:</p> <ul> <li>Folder: <strong><em>AR_Global</em></strong></li> <li>File format: SHP (shapefile)</li> <li>Spatial coverage: Global</li> <li>Geographic Coordinate System: GCS_WGS_1984</li> <li>Temporal coverage: 1^st &ndash; 5^th day of January, April, July, October for 2010, 2013, 2015 (corresponding to La Ni&ntilde;a, neutral, and El Ni&ntilde;o year respectively)</li> <li>Temporal resolution: Daily</li> <li>Naming of files: ivt_2deg_ddmmyyyy_labelled.shp</li> <li>Number of files: 60 (5 days &times; 4 months &times; 3 years)</li> </ul> <p>&nbsp;</p> <p><strong>Dataset for deep learning training:</strong></p> <p>The tool 'Export Training Data for Deep Learning' uses the IVT maps in the 'AR_Global' folder and the shapefiles in the 'IVT_ERA5_2Deg' folder to create labelled tiles for deep learning training. Each tile in the map is assigned a label: 'Atmospheric River', 'Ambiguous', or no label if it doesn&rsquo;t contain any AR or ambiguous shape. The generated map chips are stored in folder <em>Training_Testing_tiles/ RCNN_Masks_All_Tiles (no 3-10-15)/images</em>, and the labels are provided in the textfile 'map.txt' file in folder Training_Testing_tiles/ RCNN_Masks_All_Tiles (no 3-10-15)/. Data details:</p> <ul> <li>Folder: <strong><em>Training_Testing_tiles/</em> RCNN_Masks_All_Tiles (no 3-10-15)/images</strong></li> <li>File format: TIFF</li> <li>Spatial resolution: 2 degrees</li> <li>Spatial coverage: varies. Width of tile = 40 gridcells. Height of tile = 20grid cells</li> <li>Geographic Coordinate System: GCS_WGS_1984</li> <li>Temporal coverage: 1^st &ndash; 5^th day of January, April, July, October for 2010, 2013, 2015. These years correspond to La Ni&ntilde;a, neutral, and El Ni&ntilde;o year respectively. Please note that data for certain days are missing; these omissions correspond to days with no or only a single atmospheric river detected.</li> <li>Temporal resolution: Daily</li> <li>Number of files: varies</li> <li>Number of channels in each file: 1</li> </ul>