Talk:2010 07 Pakistan Floods/Imagery and data sources

From OpenStreetMap Wiki
Jump to: navigation, search

Dates table for Pre-floods imagery

Could we provide a table which relates the dates of the resources. SPOT is indicated as 2008; I think I saw mention that the Yahoo imagery is from 2006; no idea of the date of the Landsat imagery. The reason why I bring this up is that there is a comment on the page that Yahoo imagery should be used in preference to SPOT in many cases due to resolution being better. I would argue that the most recent should be used for man-made structures and that the highest resolution should be used for natural structures ... in cases where both sources provide sufficient information to trace a feature. However, for water features, it is useful to compare several imagery sources to that variance in the feature can be taken into account in mapping things like wetlands or riverbanks. Thanks for considering this. --Ceyockey 13:00, 20 August 2010 (BST)

So the region where we have a choice between SPOT or Yahoo high-res is here: http://www.openstreetmap.org/browse/way/73305298
Conceivably you might find some new man-made stuff which was built between 2006 and 2008, and which is reasonably visible in SPOT despite the low resolution, and not present in Yahoo. ...not very likely though is it. And for the rest of the time you'd be mapping much more accurately from Yahoo high res. Maybe if people have run out of anything else to do, we could spend some time checking against SPOT after using Yahoo in the region, to see if there's anything new visible. -- Harry Wood 14:25, 21 August 2010 (BST)

Mapped with SPOT 2008 but not other imagery

FYI - some things appear to be traceable from SPOT imagery but not from either Landsat or Yahoo. The reservoir at http://www.openstreetmap.org/browse/way/73383719 is one example. It seems that SPOT is better at showing some water features. --Ceyockey 13:53, 20 August 2010 (BST)

Yes. SPOT is better than low resolution Yahoo imagery/Landsat. (although it appears to be mightily wonky, with overlappy bits e.g. on the south side of that reservoir)
It's not better than hi resolution Yahoo imagery (in the areas where that is available). So in fact there's no point using SPOT in the largest of the three areas where SPOT is available
-- Harry Wood 14:14, 21 August 2010 (BST)

Client restriction

Why would a sane mind limit usage of clients to trace data for a crisis-mapping??? --amai 15:56, 21 August 2010 (BST)

It does seem rather odd, though there is surely a reason, either technical or endpoint licensing-related. --Ceyockey 18:30, 22 August 2010 (BST)

Unlisted resource in use

I have found the url http://osm.arkemie.org/cgi-bin/tiles/1.0.0/pakistan_250k_mosaic/!/!/! (link) as a source on a number of roads and waterways with the author listed as 'andygates'. Material from this source is not matching up well with low-res Yahoo or Landsat imagery, but that might merely be a matter of the material being invisible to those imagery sources (i.e. too low contrast or below resolution). Does anyone have information on this resource? Is it postentially avaialble as a WMS feed? Thanks. --Ceyockey 18:25, 22 August 2010 (BST)

See http://wiki.openstreetmap.org/wiki/2010_07_Pakistan_Floods/Imagery_and_data_sources/DMA_Maps_Mosaic - it's 1:250,000 scale paper maps. I've been trying to match to existing OSM data as best as I can but I suspect the warping's a bit off. Is it better to carry on, given the large amounts of unmapped area ,or is the offness really bad? --andygates 12:31, 27 August 2010 (BST)

Road vs. Canal

It seems sometimes tough to distinguish between a canal and a road using the low-res Yahoo and Landsat imagery, which I presume emerges from canal paths being dry or nearly so. Any thoughts on this matter? Thanks. --Ceyockey 18:32, 22 August 2010 (BST)

I cannot comment specifically to Pakistan (where I've traced only a dozens of nodes from yahoo), but I know that problem in general. Even in high-res I am sometimes struggling! One helper is the analysis of intersections with well-known objects - but in low-res that is rarely an option. Maybe a lame idea, but maybe trace anyway and add a fixme tag?! --amai 18:49, 22 August 2010 (BST)
To increase the complication, many canals seen in highres have ways parallel to them! --Andygates

Proposed - summary table for preflood imagery

I thought I would ask here before adding to the page. "(default)" in the WMS column indicates urls which are included in the WMS plugin (if memory serves, I didn't have to add these urls when first installing the plugin). Resolution - where could we find the resolution for the Yahoo hi-res imagery? Year - I've never been able to find out the year of capture of Yahoo imagery. Links in 'short name' - these are in-page links to sections. --Ceyockey 02:28, 24 August 2010 (BST)

Short name Year Value for source=* coverage resolution WMS
SPOT 2008 SPOT, EULA, 2008 Relation 1136222 7 m copy url
Yahoo hi-res unspecified Yahoo hi-res Relation 353295 unknown (on par with SPOT) (default)
Yahoo unspecified Yahoo complete unknown (on par with Landsat) (default)
Landsat 1999–2003 Landsat complete >>15 m (default)



Category:2010 07 Pakistan Floods ← link to help in seeing aggregate related changes