Automated Urban Land Use and Land Cover Classification for Mesoscale Atmospheric Modeling over Canadian Cities

[Alexandre Leroux]

Some javascript/style in this post has been disabled (why?)

Hi Geodata list,

The forwarded email below may interest Canadians on the list. The approach described uses only open source software (including the obvious GDAL/OGR) and mostly uses publicly available geospatial data.


Have a nice day,

Alex


On 11/20/08 15:58, Alexandre Leroux wrote:

Dear GeoNET list,

I wanted to inform you of an article recently accepted for publication that could interest several of you. The article is named "Automated Urban Land Use and Land Cover Classification for Mesoscale Atmospheric Modeling over Canadian Cities" and will be on this ftp site for the next two weeks:
ftp://depot.cmc.ec.gc.ca/upload/eer/people/Alex/lerouxa_urban-lulc_geomatica_v081120.pdf

It is important to take into account that the approach described in this article is our 'old approach' developed about three years ago. We plan to update it by using CanVec (replacing the NTDB), SRTM-DEM CSI-CGIAR v4, Census 2006 and newly available 3D urban buildings for several Canadian cities. Our 'old results' are 44 urban classes at a 5-m spatial resolution for all major Canadian cities. The resulting classification is, in our opinion, quite interesting for urban areas (can be automatically generated for anywhere in Canada, but does not provide good results in non-urban areas). The process is fully automated and very flexible (e.g. can easily be modified and augmented to suit other needs than atmospheric modeling). The approach is simple and mostly uses publicly available data at its core and only open source software, so it is rather easy to duplicate the process.

Here's the abstract:

An automated geospatial databases processing approach has been developed in order to characterize urban areas of major Canadian cities for their use in mesoscale atmospheric modeling. Mesoscale atmospheric numerical models, including urban canopy models such as the Town Energy Balance (TEB) model, require surface characteristics to represent surface processes occurring in cities. The developed methodology uses the following pan-Canadian databases: the National Topographic Data Base (NTDB) vector data for land use and land cover (LULC) characterization, the Shuttle Radar Topography Mission (SRTM-DEM) and the Canadian Digital Elevation Data (CDED1) digital elevation models (DEM) for building height assessment, and census data for characteristics of residential districts. These databases are jointly processed to automatically generate a high-resolution urban LULC classification for Canadian cities.

The processing establishes the NTDB vector layer priority while making use of some of the feature attributes. To complete the information derived from the NTDB database, additional processing is done such as a population density measurement and an optional building height assessment. The final classes will be used, by means of an associative process, to specify the inputs for urban meteorological modeling. Results over Montreal and Vancouver are presented, with a discussion of their benefits and limitations. The approach compares favorably to previously available LULC classifications for mesoscale atmospheric modeling in Canada (e.g. Lemonsu et al., 2006 & 2008) and the main added benefits that this approach has are (a) Canada-wide applicability with available continuous databases, and (b) complete automation, with the exception of optional post-processing.

Have a nice day,

Alex

_______________________________________________
Geodata mailing list
[hidden email]
http://lists.osgeo.org/mailman/listinfo/geodata