Skip to main content

University of Guelph HomeHome

Search Type
Menu Search

Secondary menu

  • CSAHS
  • Awards
  • Academics
  • Alumni & Friends
  • Research

CSAHS

Navigation
Geography - Applied Geomatics (GEOG*4480*w18)
  • GEOG*4480
  • Geography
  • Login

You are here

Home

Research Findings: Objective 4

Evaluate the Strengths, Weaknesses, and Accuracy of the Model

Evaluation of Model Accuracy:

Ground truthing is employed to evaluate the efficacy of the outlined research methods. Neighbourhoods identified as hotspots and cool spots of vulnerability are inspected to determine if tree cover and heat vulnerability were correctly predicted. Square kilometre tiles of identified neighbourhoods were inspected side by side both through orthophotos and with the tree footprint overlay (Figure 18). The model was run successfully as ground truthing proved that not only did the model identify areas of known vulnerabilities such as Montreal-Nord (high vulnerability) and Westmount (low vulnerability), but tree canopy cover was also accurate and showed significant disparities between high- and low-priority areas. The canopy cover produced by the model is further assessed by comparing it to one that has been produced by the City of Montreal Infrastructure, Roads, and Transportation Service, as can be seen in Figure 19. Derived tree polygons were only slightly larger than the trees they represented due to the smoothing filter applied to the canopy cover raster. This bias is minimal and is constant throughout the study site, so while it may affect absolute differences in tree canopy cover, it would not affect relative differences. Despite some disparities the shapefile produced by the model does seem to accurately represent the differences in coverage between the different census tracts.

Four panel image showing a visual comparison between tree canopy cover in high and low vulnerability areas in Montreal. a) Satellite image of Montreal-Nord neighborhood. b) Satellite image of Westmount neighborhood. c) Tree canopy cover polygon layer overlaying satellite imagery of the Montreal-Nord neighborhood. d) Tree canopy cover polygon overlaying satellite imagery of Westmount neighborhood. The image comparison shows a clear difference in tree canopy cover between the high and low vulnerability neighbourhoods. The low vulnerability neighbourhood has a much greater area of tree canopy cover than the high vulnerability neighbourhood.

                   

 

 

 

 

 

 

 

 

 

 

 

 

 

 

Figure 18.  Visual comparison between tree canopy cover in high and low vulnerability areas in Montreal. a) Satellite image of Montreal-Nord neighborhood. b) Satellite image of Westmount neighborhood. c) Tree canopy cover polygon layer overlaying satellite imagery of the Montreal-Nord neighborhood. d) Tree canopy cover polygon overlaying satellite imagery of Westmount neighborhood. The images above are 1 square kilometre tiles.                  

Image showing the overlay of the canopy cover polygon produced by the City of Montreal atop the canopy cover polygon output by the model. The model produced a canopy cover polygon that is slightly blockier and has slightly larger surface area than the polygon produced by the city.

Figure 19. Sample comparison of the tree canopy shapefile created by the City of Montreal overlayed atop the canopy cover layer output by the model. 
 

Model Limitations:

The greatest limitation of this analysis is the scale at which it is applied. Due to limitations in data availability, the census tract level is the finest scale of data outputs. For tree planting applications, a finer scale output such as at the dissemination area level would be preferable. The use of LiDAR is also a limitation as LiDAR is still an emerging technology and datasets are not widely available.

 

Model Strengths:

The model is able to produce statistically significant results at the 99% confidence level. Confidence in resource allocation decision-making is critical in municipal systems with limited budgets. The model produces data-driven outputs that can be easily interpreted by scientists and planners alike; it also produces a visual representation of vulnerability that can be interpreted by professionals and the general public.

Previous Page: 
Research Findings: Objective 3
Home Page: 
Home
Next Page: 
Conclusion
Powered by CCS