Louisiana Floods - 2016

Heavy rainfall began Friday, August 12, 2016 in Louisiana, with areas reaching almost 3 feet of rain water, causing local rivers to reach historic flood levels. Thousands were forced to evacuate, there are at least 13 dead, and many reported missing. Areas in South Louisiana in and around Lafayette and Baton Rouge were affected most.

Twenty parishes in Louisiana—Acadia, Ascension, East Baton Rouge, Livingston, Avoyelles, Evangeline, East Feliciana, West Feliciana, Iberia, Iberville, Jefferson Davis, Lafayette, Pointe Coupee, St. Helena, St. Landry, St. Martin, St. Tammany, Washington, Tangipahoa, and Vermillion—were declared major federal disaster areas.

Watson, LA—about 20 miles northeast of Baton Rouge—experienced 31.39 inches of rain, White Bayou, LA saw 26.14 inches. Livingston ended up with 25.52 inches. Baton Rouge received over 19 inches.

Content

Files

• train.csv - the training set.
• test.csv - the test set, including the labels.
• train/ - contains the training satellite images of before/after the flood, and during the flood.
• test/ - contains the test satellite images of before/after the flood, and during the flood.

Columns train.csv and test.csv

• Image_ID - a unique id for each image. Note: for each before/after the flood image there is a corresponding during the flood image, eg: 3005.png is an image taken before/after the flood and corresponding to that image there is 3005_0.png image which was taken during the flood and the *_0.png implies that the area shown in this image was not flooded.
• Normal - Indicate whether the image is before/after the flood or during the flood. 1 -> the image was taken before/after the flood. 0 -> the image was taken during the flood.
• Flooded - Indicate whether the image contains flooded regions. 1 -> Flooded 0 -> Not flooded.

Acknowledgement

https://disasterresponse.maps.arcgis.com/apps/StorytellingSwipe/index.html?appid=2e499ec7eb784237bd70fb16ae0f5dcf# http://louisianaview.org/2016/08/historic-louisiana-floods-august-2016/ https://geodesy.noaa.gov/storm_archive/storms/aug2016_lafloods/index.html#

Inspiration

Kerala, a coastal state in India was flooded in 2018 and 2019. More than 500 people died, and almost a million people had to be evacuated from their homes mainly from low lying areas. The people in Kerala and the administration find it difficult to get the geographic areas that were flooded, thus it affected the proper rescue operations. Also people were unaware of the increase in water level around their region, that they didn't leave their homes which eventually lead to more death. So we started developing solutions to help the people/administration to broadcast information, identify flooded regions, alert people and assist them during such events. At some point in time we came across Louisiana flood in 2016 and the websites in the acknowledgement, and we thought of doing some experiments with the data.