Cycling In Toronto

Read More »

Median Household Income in the City of Toronto

This map shows the median household income of every Census Dissemination Area in Toronto, based on the 2011 National Household Survey collected by Statistics Canada. This is the smallest unit these data are reported on, and are only available to me as a graduate student at Ryerson. I accessed these data via the excellent University of Toronto CHASS research site.





I realize these data have their limitations including sampling bias and small geographic unit sample size, but I still think it is interesting to see the overall patterns. I was inspired to review these data after a recent graduate class discussion about the "Three Cities of Toronto" project by Dr. David Hulchanski at the University of Toronto. His work involved mapping real income change between 1970 and 2005, not only a single year as shown above.

UPDATE: I have added one more map below showing Median Household Value. These data appear to have much more spatial clustering, with significant division between the central area and out suburbs.

Read More »

Public Park Accessibility in the City of Vancouver

I am very happy that Vision Vancouver Councillor Reimer tabled a notice of motion yesterday called "Protecting Vancouver’s Public Green Space" which included the following statement:

"The Greenest City Action Plan adopted by Council in 2011 sets a target that every person in Vancouver lives within a five-minute walk (400 metres) of a park, greenway or other public green space by 2020"

My first thought when reading this was simple: how will this indicator be measured? Specifically, I recalled a recent diagram in the Oakridge rezoning application slides (a project I fully support, just for the record) that shows 400 metre buffers from surrounding parks:



The trouble with this diagram is the 400 metre buffers appear to originate from the park centroids. Alternatively, I beleive 'accessibility' has actually been measured based on a buffer from the outer edge of parks. In other words, if I leave my house and walk 400 metres and arrive at the edge of a park, I would consider it to be accessible. I decided to explore these two interpretations; the following map shows the area within a 400 metre buffer of park 'centroids' in medium green, and the area within a 400m buffer from park 'areas' in lighter green:



In addition, I also clipped each 2011 Census Dissemination Block to the 400 metre buffer from the edge of parks, and calculated the fraction of each block that was within the threshhold. The fraction of the area was then used to calculate a proportion of the population per block that has no park accessibility. For example if a block had 100 people and it was 75% within the buffer zone, 25 people were considered to have no access to a park. As a result of this analysis, I determined that 79,407 people or 13.2% of the population of the City of Vancouver live greater than 400 metres from the edge of a public park. These people are shown using the yellow proportional symbols in the map below:



I also reviewed the Greenest City progress report and noted the indicator the City has selected is significantly simpler.Their indicator is "Percent of city's land base within a five-minute walk to a green space". In other words their target is based on area instead of population. Which is fully valid, but I still wonder how if they are measuring their covered area based on buffers from the park centroids (as shown on the Oarkridge map) or from outer edge of the parks (which would yield a larger total covered area and I am fairly confident this has been done). At the end of the day, I do not mean to criticize the outstanding work the City is doing, I am simply trying to highlight how an oversimplified diagram can lead to ambiguous results.

I am excited to learn more about this motion as it is developed because I have been working on many projects related to this theme at Golder Associates Sustainable Communities Group over the past few months. For example, I have conducted extensive GIS analysis and visualization to support the City of Vancouver's ongoing Integrated Stormwater Management Plan (ISMP)and the John Hendry (Trout Lake) Park Master Plan.

I understand this Notice of Motion is referencing much more area than just parks, and exactly what should be considered 'greenspace' will likely be a topic for a future post. For example, community gardens and greenways are referenced in the Greenest City action plan as features that satisfy the 'greenspace accessibility' target. Alternatively, given my recent focus on stormwater management I would also suggest it may be equally important for the City to implement a 'no net loss of pervious surfaces' target to maximize onsite infiltration of rainwater.

I hope this post shows that GIS tools are not always as objective as they appear and it is important for policy makers to consider how various calculation methods can impact progress towards targets.

Read More »

How Does Global Forest Area Change Impact CO2 Emissions?

This is the map that literally started my career as a cartographer. It was a final project for my GEO372 - Advanced Cartography course at UBC way back in 2008. After a guest speaker presented to my class, I learned there was a job opening at the Human Early Learning Partnership. I brought this map to my interview, I was hired the same day, and I have been a professional cartographer ever since. What did I learn from this experience? Take pride in every project, you never know where a great map may take you!

To learn more about this project, here is my 'Design Process' statement I submitted along with the final map:

First I stepped back, I conducted a broad survey of published literature to understand how national forest area change is related to climate change. There was surprisingly little evidence linking these two variables. Therefore, I located UN data describing total forest cover and percentage change of forest cover between 1990-2005. In addition, I located IPCCC data on total CO2 emissions due to forestry over the same 15-year period. The CO2 data were 'based on the removal of forest products which are counted as an immediate emission' – essentially the more wood removed the higher the Mt of CO2. I hypothesized there would be a proportionate relationship between forest area change and CO2 emissions. Therefore, I decided to map forest area change as a base choropleth map with an overlay of divergent, proportional rings to illustrate total Mt of CO2 emissions per country. I also decided to include an reference map showing the overall forest area per country in 2005 to contextualize the percentage change data.

I reviewed many possible projections to display the world on a flat page. I settled on a simple Mercator projection because it provides a good compromise of distortion and area accuracy. This is also a very commonly published view of the world and I wanted to avoid unnecessary complexity.

Next, I dug into the data. Joining the data tables with the world countries map layer was difficult – at first only a handful of countries had identical names in the shapefile and my two data sources. This was resolved by copying both source data into a new DBF file and manually realigning the columns to match up the countries I could. I decided to put the carbon data on top of forest change layer because they are both diverging sets of data – instead of my original plan to make forest change the inset. Finally, I experimented with the online MapShaper tool to simplify my shapefile and later deleted Antarctica.

Another significant challenge was the classification of my data into classes. Analyzing the forest area change distribution viewer in ArcMap I concluded 7 classes with 0% change in the middle would provide appropriate representation of this centrally weighted normal distribution. For my inset map I concluded there was a weighted distribution towards the low values, so I chose 5 classes, with the lower three tighter to the left while the darkest two green values captured very forested countries (>25% land area = forest). My proportional symbols were divided into 2 classes: the 20 largest positive (red, emitters) and 20 largest negative (blue, sinks) to show the contrasting size and location.

Now that I had the GIS mechanics over with, I decided to transfer my work to Coral Draw to create my print layout. I had to redraw the proportional symbols manually, by integrating my carbon emissions data into the spreadsheet I created for China’s largest cities in lab 7. My maximum dimension was tuned to make the Indonesia and Brazil stand out clearly from the other 20 top emitters. It is interesting to note how a majority of the circles seem to perfectly encompass the country boundary. I believe the effect I have created acts as a highlighter to each country with large positive and negative emissions, while forcing other countries with missing data to recede in the visual hierarchy.

In terms of base colour, this map started with CMYK values from ColorBrewer. However, after several test prints I increased the saturation and darkened my yellow and orange tones, and increased the contrast between dark greens. There is a clear impression that red countries and circles are ‘bad’ and this is the message of my map: red areas and circles contribute the most to climate change, while green areas have either a neutral to positive impact.

I am very happy with my decision to use white border lines and graticule. These force the coloured countries to ‘pop’ out, without any rough black lines. Additionally, using the fade effect on the background focuses attention to each map and adds depth to the visual hierarchy.

Perhaps the most interesting part of this project has been the blocking of space on the limited page. I began by turning on snap to grid – which I set at 4mm. I spent a great deal of time deciding how to scale each map outline. My decision to overlap the frames created more usable space, but also effectively cropped the jagged Russian north to improve the maps focus. I then began to fill the empty spaces with visually pleasing boxes, with no intention of what to put in each – then I fit all required elements including title, text legends and references into the place they fit best. I worked to create neat lines that were 4 or 8 mm apart to unify the proportions.

It was not until my final 20% of this project that I considered labelling each country on this map or just select few, and if that, which few? By pure genius, I doubled the country legend key with the ranking out of top 20 emitters and sinks. Amazingly, this also allowed me to include the actual Mt/year data which I visualized with the circles. Interestingly, the list inset also efficiently fills in this corner of the page which had been wasted before.

In the end, I became obsessed with picky details. I spent a good deal of the last week adding masking and shadow effects, tweaking colours and proportions, tinkering with the fade effects, formatting text, deciding what to write in my inset and call out box, what kind of graph to include and how to construct it. It has been a constant motivation to know about the cartography contest and the potential for great success in a cartography career. But more tangible results were also gained through this experience such as technical design and planning skills.

The story this map tells helped me answer several questions and provokes more questions. Since forest areas are both increasing and decreasing, carbon is being released and sequestered. Within this period, there was more forest lost than gained, so the impact is negative. However, there were several surprising outliers like USA and China that increased their forest areas and were credited with negative CO2 emissions from forestry. Future monitoring should be conducted to determine how forest carbon balances the global climate change equation.

Read More »