Hurricane Florence Tagged Data.csv (11.18 kB)

Labels for Hurricane Florence (2018) Emergency Response Imagery from NOAA

Download (11.18 kB)
posted on 19.01.2020 by Evan Goldstein, Eli Lazarus, Tomas Beuzen, Hannah Williams, Patrick Limber, Nicholas Cohn, Luke Taylor, Daniel Buscombe, Shah Nafis Rafique, Matthew Moretz, John Weber, Rinty Chowdhury, Daniel Foster, Somya Mohanty

This csv file contains labels for >300 images obtained by the US National Oceanic and Atmospheric Administration (NOAA) in response to Hurricane Florence. Images were taken on Sept. 17th 2018, and are downloadable from the NOAA Emergency Response Imagery website (see references below). All labeled images are from the ‘September 17 A 2018’ file. The 'image_id' column in the csv corresponds to the name of the JPEGs.

Images were tagged by the first 8 authors of this dataset. Each image was tagged 2-5 times by different people. Authorship order was derived by the number of images tagged. Image tagging was done using a bespoke dashboard built by five authors (author 9 through 13), under the supervision of the final author (author 14).

There are 10 columns in the csv:

1) image_id — reports the name of the corresponding NOAA JPEG.

2) ocean - 1 if the image is all water.

3) development - 1 if the image has human development.

4) washover - 1 if a washover deposit can be seen in the image.

5) impact - based on the Sallenger (2000) ‘Storm Impact Scale for barrier islands’ (see references below). 0 is no visible impact, 1 for swash, 2 for collision, 3 for overwash, 4 for inundation.

6) terrain_inland - 1 if the image is an inland scene.

7) terrain_marsh - 1 if the image is a marsh scene.

8) terrain_river - 1 if image is of riverine scene.

9) terrrain_sandy_coastline - 1 if image is a sandy coastline scene.

10) terrain_undefined - 1 if no terrain type applies.

multiple terrain types can be specified for a given image — i.e., there can be multiple '1's for a given image in columns 6-10.


The Leverhulme Trust RPG-2018-282 (Lazarus and Goldstein)

NAS GRFP ECRF (Goldstein)

CoPe EAGER: Collaborative Research: COMET: the Coastlines and people Open data and MachinE learning sprinT

Directorate for Geosciences

Find out more...