Background
The goal of the Geobase - National Imagery Coverage (Landsat 7) Project, being conducted as part of the Geomatics for Sustainable Development of Natural Resources program (GSDNR) in Earth Sciences Sector (ESS), is to produce a national coverage of orthorectified Landsat 7 satellite images. It will also deliver the geometric correction of non-precise digital files of the National Topographic Data Base (NTDB) to ensure the accurate overlaying of geospatial information.

The project, co-funded by GeoConnections, is being undertaken in partnership with more than 20 federal and provincial organizations. It is a unique initiative in the Canadian geomatics community in two aspects: the complete Canadian coverage, and the extensive collaboration at the federal and provincial levels necessary to achieve this national coverage. These products are now available to the public through the Geobase Web portal.

The Ontario Land Cover Database

Larger image [GIF, 184.7 kb, 1224 X 944, notice]

An important partner in this effort is the Ontario Ministry of Natural Resources (OMNR) Inventory, Monitoring & Assessment Unit. Since May 2001, the Unit has been using the orthorectified images for the production of a new, on-line provincial land cover map for Ontario. The Provincial Land cover database was the first land cover classification in Canada to be completed for an entire province wholly from earth observation satellite data. The original Ontario land cover database was produced in nine segments (termed "tiles") under three separate programs of the OMNR between 1991 and 1998 This database is now being up-dated using the orthorectified image data being made available through ESS by the GSDNR program and GeoBase.

In 1996, the land cover data generated by the three Ontario programs were integrated into a single provincial database. In 1998/99, OMNR launched a project to rationalize the provincial land cover data; that is, to create one single, consistent land cover classification at the provincial level by merging certain classes and sub-classifying others. The outcome of this initiative consists of 28 consistent land cover classes mapped across the entire province. The original Ontario Land Cover data was derived from digital, multispectral Landsat-5 Thematic Mapper data acquired between 1986 and 1997, with the majority of the satellite data frames imaged in the early 1990s. As more recent orthorectified Landsat 7 imagery becomes available through GSDNR program, features such as forest cutovers and burns are being up-dated for key areas in the Great Lakes forest region and for most of the Boreal forest region of the province.

Definition of Land Cover
The land cover classes consist of vegetation types (such as forest, wetlands, and agricultural crops or pasture) and categories of non-vegetated surface (such as waterbodies, bedrock outcrops, or settlements). The data reflects the nature of the land surface cover rather than the land use. For example, provincial parks are not discriminated as areas of recreational land use, but are mapped as part of the provincial mosaic of waterbodies, forest types, wetlands, and other cover classes.

How did this project evolve?
The project evolved out of a recognized need in Ontario to have reliable, up-to date land cover information to support decision-making and policy planning processes with respect to sustainable development. The availability of improved orthorectified Landsat 7 imagery from ESS represents a uniquely valuable new source of Earth Observation (EO) data to support Ontario's land cover mapping needs for the strategic management of natural resources in the Province.

When was the project established?
The partnership with Centre for Topographic Information - Sherbrooke (CTI-S) was established in 2001 and renewed with the implementation of the ESS/GSDNR program in 2003.

How does Ontario receive data from the GeoBase - National Imagery Coverage (Landsat 7) Project?
OMNR receives the Landsat 7 images available from ESS as an output of this project and supplements this data with additional purchases of imagery identified through the GeoConnections Discovery Portal (Canadian Geospatial Data Infrastructure - CGDI).

How is the data being used?
The revised Ontario Land Cover database is based on orthorectified Landsat 7 images. These data are classified by OMNR into 28 consistent land classes. The Ministry provides provincial partners with an on-line image mosaic, as well as a historical archive, for use by various agencies within OMNR (e.g., wildlife and conservation, forestry). In addition OMNR responds to requests for landcover data from academia, communities, and the private sector on a regular basis.

How are Canadians benefiting from the use of the data?
The new version of the on-line Ontario Land Cover database enables the Province to provide reliable, up-to-date information on land cover for the entire Province. Regular clients outside OMNR include the regional Conservation Authorities, the Ministry of the Environment, Agri-Foods Canada, Hydro One (corridor studies) and the Ontario Ministry of Health (The Muskoka and Perry Sound Health Unit, for example, accessed the database this summer to identify wetlands that could be of concern for the West Nile virus).

OMNR has also provided sixteen university libraries across the province (Brock, Guelph, Lakehead, Laurentian, McMaster, Nippissing, Ottawa, Ryerson, Sir Wilfrid Laurier, Queens, Toronto, Trent, Waterloo, Western, Windsor, York) with the Ontario Land Cover image mosaic. Faculty and students now have access to the mosaic for enhanced curriculum content, student projects (e.g., environmental studies) and research theses. Without the GSDNR/OMNR partnership initiative these Ontario user communities would not have such ready access to this geospatial dataset.

What has been the benefit of the partnership?
OMNR has access to the entire landmass of Ontario in orthorectified Landsat 7 imagery with an almost complete cloud-free coverage from GSDNR/GeoBase. The value-added associated with the Landsat mosaic from ESS for mapping and classification purposes are that the images are orthorectified. This accurate positioning is especially important for Northern Ontario, where traditional mapping approaches cannot meet current accuracy and content requirements for land cover mapping.

OMNR's experience has demonstrated that the accuracy of their original, high-resolution Ontario Land Cover database (Landsat-5) is 95 percent or better for waterbodies; approximately 90 percent for the forest classes, with exception of some degree of unavoidable confusion between treed wetlands and sparse forest classes; and approximately 85 percent for agricultural land cover, taking into account the presence of widespread, unavoidable confusion with small towns and roads. Mine tailings, quarries, bedrock outcrops, and coastal mudflats along Hudson Bay and James Bay were mapped as a single class of minimal vegetation cover with an overall accuracy of approximately 90 percent, although some quarries may have been indistinguishable amid agricultural land cover.

The orthorectified Landsat 7 images from GSDNR and derived Ontario Land Cover mapping represent a significant improvement in positional accuracy over the earlier Landsat-5 derived product. The best Landsat 5 resolution was 30m. With the Landsat 7 data the accuracy is approximately doubled. This improvement is due in part to the inclusion of the 15m resolution panchromatic band in production of the Landsat 7 orthorectified images available through GeoBase.

For more information contact: Christian Prevost

Other Success Stories

GSDNR Communicator