Public Geodata Hosting: Large-scale Raster Hosting

[Christopher Schmidt]

I'm currently in a conversation with a representative of MassGIS about
hosting of their 2005 orthophotos for the state of Massachusetts.
Currently, these files are available from the MassGIS website in lossy
MrSID format.

Among the concerns with putting up the newer data in the larger JPEG2000
format is the bandwidth hit that MassGIS will suffer once these images
go up. Any organization hosting gigabytes of interesting data probably 
recognizes the problems inherent in the hosting -- namely, that you can
quickly become a product of your own success.

As a result, MassGIS is currently seeking an organization to help mirror
these files on a fast link for public consumption, at least for the
first 6 months when interest in the photos will be at its peak.

Among the stated goals of the Public Geodata committee are:

 # Run a repository of open geodata
 # Promote public access to state-collected geodata
 # Promote the use of open geospatial formats

I think it's obvious how offering to help MassGIS with hosting of such a
dataset would be an excellent demo and test case of any repository of
open geodata. Granted, this information is not the typical form of
'geodata' that is lobbied for in this situation, as vector data can in
many cases be used more effectively. However, large-scale sets of raster
data are clearly not to be ignored.

Additionally, MassGIS licensing is very permissive -- only attribution
is required, which makes it a case where it's high quality geodata that
definitely falls under the umbrella of 'open'.

By working with MassGIS to host such a dataset, we can practice what we
preach by helping to make access to the state-collected orthophotos more
publicly visible.

Additionally, by helping alleviate the bandwidth concerns associated
with more 'open' image format that is JPEG2000 over MrSID, the public
geodata committee can demonstrate by example their ability to help
promote more open formats. (Note that MrSID is still more broadly
supported than JPEG2000, so publishing those files is absolutely
appropriate for MassGIS to reach the most users -- but publishing MrSID
is not, it seems, the more difficult part of the problem.)

Some ideas were bandied about about searching for hosting at local
universities, etc., but I think this could be an interesting technical
project for the public geodata committee to take on: a large-scale, well
defined set of data with an enthusiastic governmental organization
looking to help users. (Governments looking to help their constituents!
excuse me while I faint!) However, the technical problems may be hard to
solve: the data is 15GB in lossy MrSID, so storing lossless JPEG2000 may
be quite a chore. The bandwidth might be much even for a non-MassGIS
entity, etc. 

I'm looking to get a feel for other people's feeelings on this topic:
how feasable is it? Is it a worthwhile project? And most importantly
because I'm extemely impatient, what's the time frame on being able to
set up resources for this type of thing?

Note: I do not speak for MassGIS -- I'm just acting as a go-between
since I feel that it's easier to hear this from a member of the group
than outside to start. :) 

-- 
Christopher Schmidt
Web Developer