Public Lab Research note


This is an attempt to replicate an activity.
  • 2

Hurricane Harvey Industrial Impact Image Sorting

by gilbert |

image description

On Thursday, October 12, 2017, a Workshop, jointly sponsored by the Gulf Restoration Network (GRN) and the Public Laboratory for Open Technology and Science (Public Lab), was convened in New Orleans, LA at The Warehouse (3014 Dauphine St.) for the purpose of assessing NOAA Airborne sensor data post Hurricane Harvey. The objective was to visually search for evidence of oil spills and gas flares from industrial facilities that may have been impacted by flooding consequent to Hurricane Harvey. Scott Eustis, GRN Coastal Wetland Specialist, enabled the participants to differentiate individual NOAA flight paths by accessing an icon in the upper right quadrant of the image, as well as to discern evidence of oil spills.

According to NOAA's website https://oceanservice.noaa.gov/news/sep17/hurricane-harvey.html the data was captured, as part of the National Geodetic Survey, during the period August/September 2017, for specific sites designated by FEMA and the National Weather Service. The online data incorporates latitude/longitude coordinates and allows for "before and after" comparisons. "A team of NOAA aviators captured the images using specialized remote-sensing cameras aboard NOAA's Hawker Beechcraft King Air aircraft flying above the area at an altitude between 500-1500 meters. The aircraft can support digital cameras, multispectral and hyperspectral sensors, as well as topographic and bathymetric LIDAR systems. "Each photo covers an area of approximately two square miles of the earth's surface." The aerial camera is a Wild RC-8 Lens and Shutter Assembly.

The Workshop's volunteer participants successfully identified and logged 55 images of areas of probable concern, with their associated geo-coordinates. Given the large volume of data to be analyzed, in addition to GRN's continuing the quest for online volunteers to assist in data analysis, Public Lab has reached out to volunteer remote sensing image analysts at Purdue University and at Tuskegee University to determine the utility of Multispec freeware for automated classification of the NOAA data, based upon feature extraction and spectral signatures.

Participants voiced some suggestions for future improvement of the activity:

-Designate one person with more experience to serve as quality control, thereby allowing everyone else to work faster identifying probable sightings and having such sightings reviewed by the designee for quality control.

-At the beginning of the session, have two examples presented for each of the features of interest (e.g. spills; flares).

-In searching for spills within the imagery, note that following the course of rivers can be a useful strategy.

-Sharing optimal strategies for panning across the map would minimize frustration and offer assurances that the task is being completed successfully.



gulf-coast image-sorting hurricane nas disaster-response harvey hurricanes featured-request

replication:14916


4 Comments

Yes! We divided up and looked at images from the August 30th and 31st and the first flight of September 1st. Click the button in the top right hand corner of the data set to see all the dates. Lots of the flights days still need to be sorted through!

Screen_Shot_2017-10-13_at_10.52.47_AM.png

Check out some of the spills and flooded industrial sites we found:

Screen_Shot_2017-10-13_at_11.03.40_AM.png

Screen_Shot_2017-10-13_at_11.01.41_AM.png

Screen_Shot_2017-10-13_at_11.00.26_AM.png

Screen_Shot_2017-10-13_at_10.59.29_AM.png


Wow, these are scary pics and I'm so glad to hear about the success of this meeting. I hear you on the "using an expert" part. Zooniverse makes up for the expert by having many people review same images. Somehow things even out in the end for accuracy. They also mention the issue of panning, which is why I have to learn more about distance and tiling. :). Oh hey, what was the level of expertise in the room? Did you have any mapping experts? Thanks for doing this--it just made my week to see something concrete happen to help with identifying pollution from storms.

Is this a question? Click here to post it to the Questions page.


I guess i was the expert, having spotted sheens for 5 years.

scott


Hey @eustatic I didn't know you were there--of course you would be the expert! So glad this was a success. I've been away for a few days but I'm saving tonight to try out your methods. May be possible to do an event in Philly. Glad you are getting some reports in. #stormhero


You must be logged in to comment.