[Coral-List] Fw: Aerial photos of Praslin Island, Seychelles

Serge Andrefouet serge.andrefouet at noumea.ird.nc
Fri Jun 1 23:39:17 EDT 2007

Hi listers,

I complete a bit what is say below by Klaas.
It's true that everything will depend on your mapping goals.

Then, in general, Google is a fair help in many cases, and totally useles 
in many others. At least it's free and easy to check.

Then, there are several sources of free Landsat imagery, including the 
Global Land Cover Facility at U. Maryland mentioned by Klaas, or the 
Millennium Coral Reef Mapping archive of University of South Florida/IRD 
hosted by the SeaWiFS project at Goddard Space Flight Center. Part of the 
image data set is available on both GLCF or GSFC sites. Most recent ETM+ 
images will be principally acquired between 1999-2003. You can access 
through http://oceancolor.gsfc.nasa.gov/cgi/landsat.pl or 
Most Landsat images georectification accuracy is 30-90m, though once in a 
while you have up to 250 m. But generally it's very acceptable, and a 
matter of correcting an offset if there is a problem for a particular reef. 
Of course, again, 30-60m may be useless depending on your application, or 
just fine. All GLCF data are orthorectified, so it means you get 30-60 m 
accuracy since several images have been used to get an average position. I 
have seen programs requesting ASTER or SPOT 5 data for reef mapping, hoping 
that their better spatial resolution (10-15m) will enhance the results, but 
they are frequently useless (except if you look at terrestrial targets like 
post-eathquake uplifted reefs) compared to Landsat (30) coz they lack a 
blue band, and are quite noisy.

An often wrong statement is to say that images are really expensive. If it 
is not airborne, but spaceborne, no, it is not. If your site is already 
imaged and archived, it will cost you about ~15-18US$/km2 and you have to 
purchase 25km2 at least (for Quickbird, who has an extensive archive or 
coral reef sites) of an area that can have any shape you want ...so for 
many reefs and small scale mapping project, this is largely enough and for 
~500$ you have a tremendously useful piece of data, generally in 99% of the 
cases georeferenced at <20 m accuracy. If your site is not in the archive, 
you'll have to make a specific request and  pay at least for 64 km2, and 
gamble a bit of the quality of the image you'll get (may be useless if very 
strong winds, turbidity, small clouds exactly where your transects 
are...  etc...). Of course, for very large scale mapping, you'll need big 
bucks. But for most small projects, and say, for most MPA sizes, it is a 
very good investment, often for a fraction of the cost of the airplane 
ticket you'll have to pay to go there.

I consider Quickbird data at 2.4m data  to be the best around, affordable, 
for reef mapping, though  we still waiting for that 2 to 8 m hyperspectral 
commercial imagery which will provide routinely significantly improved 
capabilities for specific applications. But for many shallow habitat-based 
applications, the current commercial sensors are quite fine. Contrary to 
popular belief, the limitations are generally not the remote sensing data, 
but the amount of in situ ecological/geological data that can be matched 
with the amount of information provided by an image. Again, it all depends 
on your applications. I largely prefer a 2.4 m multispectral to a 18 m 
hyperspectral, but again, application-dependent.
In any case, whatever the future, if you're interested by change detection 
you'll still have to process good old aerial (color) photos that can often 
be resampled to quasi-match Quickbird specifications. For many countries 
worldwide, even so-called developped, their access is a random function of 
talking to the right person, luck, perseverance and significant digging in 
a variety of obscure offices. They may not be cheap, but they are priceless 
since it is often the only way to vist the past.

About the UNESCO handbook by Green et al. 2000, it's a good start indeed 
for general reading but lacks a coverage of the new commercial sensors and 
methods, for which you can try to rely on recent publications which are not 
necessarily  found only in technical remote sensing journals, but more and 
more often in ecology/geology/oceanography litterature. The handbook is 
also based mostly on only one Caribbean site, so you may need other recipes 
for different configurations of reefs, even if theoretically the principles 
should be the same. Most of cost-effectiveness information which were 
perfectly fine 7-8 years ago, should be  checked/updated given the current 
commercial prices, sources of free  data, sharing possibilities and 
scientific gvt programs that buy and offer large amount of data for 
innovative scientific applications.



A 06:02 PM 5/31/2007 +0200, Klaas Pauly a écrit :

>----- Original Message -----
>From: "Klaas Pauly" <Klaas.Pauly at ugent.be>
>Sent: Thursday, May 31, 2007 5:45 PM
>Subject: Re: [ALGAE-L] Aerial photos of Praslin Island, Seychelles
> > Google Earth is a good friend when it comes to previewing a site and
> > getting a general impression, or even for planning ground truthing field
> > work, but it is absolutely useless when it comes to mapping - first of all
> > you have no spectral information whatsoever, and more importantly, the
> > coordinates are far from reliable.
> > I wonder what kind of mapping study you would like to do. If you aim only
> > at swimming transects with a gps and then delineating habitats on an
> > aerial picture, google earth pictures might even do, as long as you
> > georeference them yourself.
> > However, if you would like to do satellite of airborne mapping based on
> > supervised classification, there is no simple answer to your questions.
> > Recent satellite or airborne imagery (on request) is always very
> > expensive; airborne even more than spaceborne, but you also need to
> > consider the spatial resolution you want to work at in taking a decision.
> > If moderate resolutions are fine (30m), a good starting point is the free
> > database of Landsat ETM imagery at
> > http://glcf.umiacs.umd.edu/data/landsat/. You might want to search the FTP
> > server; Praslin is located in paths 155 and 156, row 63 in the WRS-2
> > system. Again, the provided georeference is unreliable, but with a bit of
> > fieldwork you can directly geocode them to your own ground control points.
> > You can perform an unsupervised or even supervised classification (if you
> > already got recent, detailed and georeferenced ground information) on
> > these images in order to see if it suits your needs. However, these
> > archived images will likely be quite old.
> > If you want to request new acquisitions, you might consider the
> > CHRIS/PROBA sensor (18 spectral bands at about 18m resolution); if your
> > project proposal gets accepted by ESA, they'll provide you imagery for
> > free for scientific research. This is because the sensor is experimental,
> > and although not free of flaws, the datasets have proven to be very useful
> > in coastal mapping studies.
> > More spatially detailed imagery (about 2m) can be obtained by IKONOS or
> > QUICKBIRD, but those provide only three spectral bands. More spectral
> > bands (hyperspectral datasets) on even higher spatial resolution (10s of
> > cm) can be obtained by the airborne CASI sensor, but these acquisitions
> > are very, very expensive.
> > A methodically very good, although technically out-dated book is "Remote
> > Sensing Handbook for Tropical Coastal Management", published by UNESCO.
> > Decent and free (if you wait for another month) remote sensing software
> > can be obtained at http://www.itc.nl/ilwis/.
> >
> > All the best,
> > Klaas
> >
> > ---
> >
> > Klaas Pauly - Teaching Assistant, PhD student
> > Phycology Research Group, Biology Department, Ghent University
> > Krijgslaan 281 (S8), B-9000 Gent, Belgium
> > Tel +32 9 264 8507 - Fax +32 9 264 8599
> > Klaas.Pauly at UGent.be
> > http://www.phycology.ugent.be
> >
> >
> >
> > ----- Original Message -----
> > From: "Emanuel Mendonca" <emanuel.mendonca at GMAIL.COM>
> > Sent: Thursday, May 31, 2007 5:02 PM
> > Subject: Re: [ALGAE-L] Aerial photos of Praslin Island, Seychelles
> >
> >
> > Google Earth is your friend!
> >
> > http://earth.google.com/
> >
> > :)
> >
> >
> > 2007/5/31, Adrien Cheminee <adriencheminee at yahoo.ca>:
> >>
> >>
> >>
> >>
> >> Hi everyone,
> >>
> >> In order to undertake field work for marine biological monitoring at
> >> Praslin
> >> Island, Seychelles, I am looking for aerial or satellite photographs of
> >> the
> >> north coast of Praslin. I would be very grateful for any photos or advice
> >> in
> >> order to gather them. I am more precisely interested in photos of the
> >> coast
> >> near Anse Takamaka and also Anse Volbert. Field work will aim at mapping
> >> main nearshore bentic habitats, more particularly seagrass beds and
> >> corals.
> >> I am consequently looking for photos with good resolution that could be
> >> used
> >> for preliminary photo interpretation and mapping.
> >>
> >> Thanks in advance for any help,
> >>
> >> Best regards,Adrien CHEMINEE
> >>
> >>
> >> Adrien CHEMINÉE
> >>
> >> Chargé d'étude en biologie marine
> >>
> >> mel : adriencheminee at yahoo.ca
> >> adresse : 10 rue Pastoret 13006 Marseille, France
> >> tel: (+33) 6 81 73 95 19
> >>
> >>
> >>  ________________________________
> >>  Obtenez des réponses à vos questions ! Profitez des connaissances et des
> >> opinions des internautes sur Yahoo! Questions/Réponses.
> >
> >
> > --
> > The Mediterranean - a gateway, not a barrier,
> > The Wind - a driving force, not a obstacle,
> > The Nature - a mother, not an enemy.
> >
>Coral-List mailing list
>Coral-List at coral.aoml.noaa.gov

Serge Andrefouet
UR Coreus - Institut de Recherche pour le Développement (IRD)
BP A5 - 98848 Nouméa cedex - Nouvelle Calédonie
Tél : (+687) 26 08 00 - Fax : (+687) 26 43 26 - andrefou at noumea.ird.nc

More information about the Coral-List mailing list