[Coral-List] Clarifications re: Allen Coral Atlas

[David Blakeway]

I think we should recognize the inherent difficulty of automated global
reef mapping. There's no way a single algorithm can do it. If the Allen
maps are only 75% correct, or even 50% correct, that's a significant
achievement to build on. Machine learning will do it, given enough ground
truth points.
Although I understand Vassil's disappointment about the missing Golfo de
Guacanayabo reef - possibly the world's largest Holocene reef?? (by volume)



On Wed, Dec 16, 2020 at 1:46 AM Helen Fox via Coral-List <
coral-list at coral.aoml.noaa.gov> wrote:

> Dear Coral-List,
>
> We wanted to response to the comments from Doug Fenner and Vassil
> Zlatarski on
> 12 Dec 2020 re: errors in the Allen Coral Atlas (in the thread "Source for
> "coral occupy <1% of the seafloor but house more than 25% of species"):
>
> - The Atlas maps (depth, geomorphic and benthic composition) will have
> errors in them - we are very up front about this on our page that explains
> how we make the maps and that errors are likely to result
> https://allencoralatlas.org/methods/, and in the FAQ’s “What do I do if I
> find an error or problem in the map?” https://allencoralatlas.org/faq/.
>
> - Using satellite image data for reef mapping has limitations; we are very
> open with what they are (see “USER Guide Part 1
> <
> https://storage.cloud.google.com/coral-atlas-field-data/training-materials/ACA%20user%20guide%20(part%201)%20general%20use.pdf?_ga=2.30240639.-1988623561.1568131422
> >”
> at the bottom of the Field Engagement section of
> https://allencoralatlas.org/methods/; this also explains why we expect
> errors and where to send corrections (corrections at allencoralatlas.org),
> since we cannot know where the errors will occur and cannot check them all.
> It is always a trade-off to map and monitor millions of km2 of any
> environment, so the more we link with field data and knowledge the more we
> improve the process!
>
> - Feedback is an important part of improving the process and we’re grateful
> to everyone that has helped provide feedback on the maps to our team.
> Ideally we can fix systemic errors by improving the semi-automated mapping
> approach, but in other cases we can’t, due to the constraints of this
> approach (for example, we cannot map turbid areas). We appreciate all forms
> of feedback, but we cannot, and have never, guaranteed that any specific,
> localized errors can be corrected.
>
> Our feedback request has been on the website since we launched it, and this
> is what you can provide that will help us:
>
> · Reef name(s) or location, either:
>
>  · coordinates and/or a georeferenced polygon (in kml, shapefile, or JSON
> format)
>
>  · screen shots from the Atlas of relevant area (annotated with details)
>
>  · copy Atlas URL of zoomed in location (this will enable us to see the
> area you are talking about)
>
> · Explanation of the error and your suggestion for correcting it/them
>
> · Send us any relevant field data
>
> · Your contact details
>
> The Allen Coral Atlas is a large international collaborative effort
> building on decades of reef mapping knowledge from this community. It only
> works based on contributions from many reef ecology, management and
> community groups around the world working with the mapping and engagement
> teams. We’re really grateful to everyone on this list who has volunteered
> their time, expertise and datasets to help improve our maps. Our products
> will continue to evolve and improve as we refine it over time and improve
> based on where people find errors.
>
> We will continue to do our very best to provide a state-of-the-art resource
> to support research and conservation by the coral reef community,
> especially in places where no previous maps exist. Local manually-created
> reef maps of well-studied areas will often be more reliable and accurate
> than the Atlas, but many coral reef regions lack detailed or accessible
> spatial data; the Atlas seeks to fill these gaps, and provide
> globally-consistent products.
>
> Many thanks to everybody in the coral reef community (Coral-List and other
> fora) who have acted constructively and supportively to help us build and
> share this resource.
>
> Sincerely,
>
> Helen & the Allen Coral Atlas team
>
> Helen E Fox, PhD Allen Coral Atlas, Field Engagement Advisor (hfox at ngs.org
> )
>
> Conservation Science Director, Coral Reef Alliance (hfox at coral.org)
>
>