Cool Analysis

Exploring Urban Change Through Population Density

So, after an epic break from blogging, Licker Geospatial Consulting is back at it, with a nice new post on our favorite topic, Urban Change!

Part of this has been spurred on by our recent collaboration with the fine folks at Renewable Cities. With them, we've been looking at changes in population dynamics over time in Metro Vancouver, British Columbia and other Major Canadian Municipalities. This post will show off something similar but will be more related to how we can possibly understand multi-temporal, multi-variate changes through some nifty mapping.   

Now as many of our past and current collaborators know, Licker Geospatial Consulting cut their teeth in urban planning and then did a right angle turn into environmental planning and earth sciences (and then some sort of squiggle into geospatial consulting!). One of the coolest things about the earth sciences (i.e. geotech) is the use of cross sections to comprehend and visualize multi-dimensional data. Now because urban planners need to fancy things up all the time, they took the humble cross section and renamed it: The Transect (Thanks DPZ) The transect is, of course, totally awesome and very useful to help planners systemize urban environments. Very rarely, however, do we see the transect used as a tool for urban analysis and information synthesis.  Even more rarely do we see transects used to understand temporal change. What we want to achieve with this post is to show how a transect, or cross section can be employed to understand urban spatial-temporal change and possibly the influence of mass transit on urban development.

Let's Talk Data

(Non GIS nerds please skip to the next heading) As mentioned, through my collaboration with Renewable Cities, I've been amalgamating an awesome longitudinal set of census data (specifically population) for all of Canada. These data are great because they represent a controlled and assured set of measured data that is mostly comparable across multiple years. For this little post, I've focused my efforts on the lower mainland, but there is nothing stopping us from seeing how things have changed in any part of urban Canada, really. What we are using for inputs are Census Enumeration Points, Census Dissemination Areas, or Census Dissemination Blocks, depending on the year. Comparing these data is not as easy as just overlaying all these polygons into a massive set of longitudinal greatness, indeed what's really annoying (but inevitable) is that all of these blocks or areas are not the same size, shape or even types due to the fact they were created to administer Census data not to allow for cutting edge analysis, alas...

 A big data jumble, 1971 data are light blue polygons, 1981 and 1986 data are light and dark blue points respectively, and 1991, 1996, 2001, 2006, 2011 and 2016 data are purple, blue, red, black and green respectively... Nothing matches up!!

A big data jumble, 1971 data are light blue polygons, 1981 and 1986 data are light and dark blue points respectively, and 1991, 1996, 2001, 2006, 2011 and 2016 data are purple, blue, red, black and green respectively... Nothing matches up!!

Transforming these data into information that we can actually use requires some generalization and interpolation. The quickest and easiest technique out there is to use a kernel density function on the centrepoints of each dataset so at least everything is on the same scale and in area density (which is what we wanted anyway). In plainer  English, we made everything into density surfaces. Once I had these generated, we got even more clever and interpolated between years to generate this mesmerizing beauty below (but that's another blog post)....

ezgif.com-optimize.gif

Where's the analysis already?

So we got the density by year and you'll note we found some Skytrain lines somewhere. It's time to run some sections! Cross sections are pretty cool because they allow us to view overlaid data along a common axis, which greatly enhances our ability to synthesize data at a glance (In English again: lots of data quickly). What we did was: first, measure the change in population density between 2016 and 1976 (40 years of change: superb!) to develop an overall understanding of change, and second we used some GIS wizardry (go to the very end of this post for the method) to develop sections by Skytrain line. The results, with a little bit of interpretation are displayed below for your amusement and edification (click to expand all figures).

The Expo Line

Let's walk through this slowly. First the rainbow of colours on the top figure represent change in population density, as expressed by the average number of people per hectare within an 800m neighborhood. Areas that are red are where the population has actually declined in the past 40 years and areas in brown are where it has increased (sometimes a lot!). The bottom graph shows the population density along the line starting at Waterfront Station to the left and ending at King George Station to the right. Each line represents the population density at that particular year. The Expo line was more or less completely built out by 1995, but change along the route has varied tremendously by station location. First note that between 1976 and 1986 there was almost no growth along the line! (astounding in this day in age!) However, post expo things started to change mostly around Joyce Station and New Westminster. After 1996 there was a ton of development around the Edmonds and Joyce Skytrain stops as well as in Downtown Vancouver (certainly related to de-industrialization and the Skytrain). After that development boomlet, you'll notice that the next big changes were both in Downtown Vancouver and in Surrey which saw the most growth in the 2006-2016 period. There are some other cool things to note: population density has not materially changed much at Nanaimo station, 29th Avenue, 22nd Avenue, and Scott Road (I would love to look at changes to employment though!). 

The Millennium Line

It's a different story for density along the M line (and the sneaky extension I added in there) than the E line, that is for certain. First, you'll note that residential density increased very little in the City of Vancouver in the past 40 years along this line (likely because most of the existing M line route is still surrounded by active Industrial Land, for the most part, as well as by low density residential lands and the new M Line extension had not been built yet). However, Burnaby is a different (and interesting story). First, Skytrain induced development caused population density at both Gilmore, Brentwood, and Holdom to increase from effectively 0 people per hectare in 1996 to about 75 persons per hectare in 10 years! There's been a similar rate of change in Central Coquitlam as well. Another interesting density spurt is at Loughheed Town Centre which had a huge boom between 1976 and 1986 that didn't happen anywhere else along the line (this was basically due to an attempt at highway oriented multi-family development!). Finally, it looks like Moody Centre and Coquitlam Centre are due for booms given the recent Evergreen line extensions. 

The Canada Line

So, a drawback from similar scales between these three figures means that population density at the Yaletown Station is literally off the chart! To properly understand this change just note that in 20 years, the population density at Yaletown Roundhouse has increased from 100 persons per hectare to 245! (and was 0 in 1986!!!) Nothing else in the lower mainland even comes close. Take, for instance, the stretch between King Edward and 41st Avenue, basically zero change in 40 years. Richmond has seen some pretty big changes in the past 20 years but an order of magnitude less than Vancouver (likely flight path/flood plain induced). Now, more than any other graph we've looked at in this little assessment, this one will change the most in the next 5-10 years as there are big changes underfoot in the Cambie Corridor area, so stay tuned for a follow up blog post in 2022 ;) .

Some Sort of Conclusion

As with all of our posts here, we are trying to highlight an application of geospatial analysis that can help us make better urban planning decisions and understand our urban environments. The next time someone questions the transformative impact of mass transit coupled with de-industrialization then all you need to do is show them these graphs and explain what a cross-section is. (if only it were that easy). That being said, all this work is exploratory in nature, but super fun to do, so if you feel the need to conduct some longitudinal urban analysis then feel free to drop a line to aaron@lgeo.co.  If you want to pick this post or our methods apart, the comment section is below!

Finally, Nerdy method for making the graphs.

Simply put the method is to: first interpolate the vertices of a line against a surface and then plot those those interpolated values as y values on an xy graph where x values are measured and scaled against the length of the line!

 

Metro Vancouver 1968 - An Amazing Window Into Regional Urban Evolution

Because I cut my teeth as young GIS analyst supporting land use planners, I am huge sucker for old land use maps.   Recently, my old colleague Peter Russel from the City of Richmond found an amazing copy of Development in the Western Portion of the Lower Mainland Region, 1968.   How amazing is this map? Well for starters, it is completely packed with information, its super well designed, and it was drawn up almost exactly 50 years ago!

While it's unlikely that the young draftsperson who constructed this map in 1968 actually reads LGeo's blog, if you are out there and reading this, then please know that your work has stood the test of time! Take a look at the gorgeous map below and tell me if you disagree. (full size 20mb version can be found here)

Now the reason why I really love this map, is that it provides an extremely valuable perspective on long range land use planning and the evolution of our urban environments.  That is to say, if we know how much things have changed in the last 50 years, we can start to grapple with the profound challenge of figuring out the next 50.  To do this, we can use a bit of fancy GIS and can run some fairly cool analysis on this map to figure out how things have changed.

The western portion of the lower mainland region was a different world compared to Metro Vancouver

The first thing you notice is how much influence the Fraser River had on the development of our region.  In 1968, the original Port Mann Bridge had only been constructed four years prior in 1964, the Massey Tunnel had only been around since 1957, and only the venerable Pattullo Bridge had been standing since 1937.  These lack of connections were likely one reason why the population south of the Fraser River was so small.   Second, you can notice how agricultural lands are relatively unchanged since 1968, likely that is due to the implementation of the Agricultural Land Reserve which was established in 1974.  Finally, notice the primacy of roads and highways on this figure.  By the time this map was produced, our region was completely ensconced in the world of motordom. Cars were the mode of choice on this figure and it shows.  Rail lines are shown but faintly and only fragments of the interurban lines can be inferred from the very interesting place names scattered throughout Metro's suburban communities.

Contrast this map with Metro Vancouver's Regional Growth Strategy map here.  You'll note that roads and highways are completely de-emphasized with a focus instead on transit and multi-nodal development has supplanted land uses. 

Use the simple application below to explore the figure. I've found that by fiddling around with opacity of the figure (or the extracted data), it becomes really easy to see what has changed.  Once you are done panning zooming and overlaying, check out the discussion below for some insights into the evolution of our region.

adjust overlay opacity:
  • Swap Overlay

Spatial analysis & Discussion

As mentioned, this figure is full of information.  The clear colors, well positioned labels and highly accurate mapping render this map well positioned to produce informational champagne.  However, old figures like this one do not yield their information without a digital fight! I had to spend a fair bit of time, extracting, digitizing and correcting the data to transform the information from paper to silicon.  All that being said and done, I've managed to create a reasonable dataset from this figure with the following standard (and very important caveats):

  • Extracting digital data from old figures is a profession in of itself. I'm not an expert in this area and I am sure that others couple likely do a better job. That being said I used a simple interactive supervised classification system coupled with some aggregation and manual correction to try and extract all the land use types. If you can't read GIS, in English what I did was pull the colours of the map and make some assumptions about things I couldn't see.
  • There is no way for me to easily verify the veracity of this figure.  That is to say, some of the information could just be plain wrong, out of place or simply ambiguous.  I just don't know because I was neither alive nor present in  the Lower Mainland in 1968. So, calling all baby boomers: I insist that you check to see if Sullivan Station and Essondale were places.  Also take a look at Sea Island, note the mystery community there?
  • All area totals are super approximate and based on my digitization. Therefore, please do not live or die on the fact that I've noticed there was approximately  427.5 sq km of agricultural land depicted on this figure in 1968.  By the same token, please enjoy the fact that there was apparently 487.4 sq km of agricultural land in 2011 but do not quote the 546.3 sq. km total I've used that corrected for right of ways (more on that below).

So, what do we know about the Lower Mainland Land use in 1968 (note, this is chopped  down to MV boundaries, and does not include the Northern Parts of West and North Vancouver as well as Bowen Island):

Designation Total Area (Sq. Km.) Percent of Area
Agricultural 427.6 20.1%
Civic and Institutional 43.0 2.0%
Commercial 21.8 1.0%
Industrial 35.3 1.7%
Residential 262.5 12.4%
No Designation 1,201.3 56.6%
Parks and Recreation 131.3 6.2%

Somewhat interesting... But the temporal analysis is even more interesting.  For instance, how much has the residential footprint grown?  Have we really deindustrialized?  what about outdoor space?  To answer this, as mentioned I've done some work to get the 2011 MV Land Use data in line with the 1968 version.  Some notes.

  • For both 1968 and 2011, I've conflated parks and recreation with no designation as MV 2011 has created this unfortunate category: Recreation, Open Space and Protected Natural Areas
  • Because the 1968 dataset does not have road right-of-ways (ROWS) as a designation, I've had to roll ROWs into all other categories proportionately in the 2011 dataset and inflate land use numbers by approximately 200 sq. km.
  • 2011 mixed use res/commercial has been reclassified to residential; and
  • 2011 extractive industries has been reclassified into no designation because that's how they were treated in 1968...

With those additional caveats in mind. treat yourself to 50 (...err 45 ) years of land use change!

Designation 1968 Area % Land Base Modified 2011 Area % Land Base % change in area (1968-2011) Change in area
Agricultural 427.6 20.1% 546.3 25.6% 27.8% 118.7
Civic and Institutional 43.0 2.0% 53.0 2.5% 23.4% 10.1
Commercial 21.8 1.0% 33.2 1.6% 52.3% 11.4
Industrial 35.3 1.7% 82.3 3.9% 133.3% 47.0
Residential 262.5 12.4% 473.7 22.2% 80.5% 211.2
No Designation 1,332.6 62.8% 945.4 44.3% -29.1% -387.2

So here are some fun takeaways for you to debate with me:

  • Residential land has increased by 211sq. km. or 80% Population increased by 1.5m (about  900,000 to 2.43m) or 170%, ergo we've sprawled out but we are also more dense than 1968 - despite an addiction to motordom!
  • Industrial land area has increased considerably, Notably, Vancouver's industrial areas have shrunk while Surrey's and NE Sector(especially along the Fraser) have grown dramatically. 
  • The ratio of commercial to residential lands has slipped somewhat from 8% to 7% (this may be offset by an increase in industrial and mixed land uses), suggesting that commercial land use demand is linked to residential growth.
  • Lands considered as agricultural (who knows what the Non-designated lands were in 1968!) appears to have increased; suggesting that the ALR has been effective in preserving agricultural lands... (I would love to see more data on this to see if this is the case!)

I could go on, and there are probably many more facts and nuggets that you can glean from the data.  So put on your analysis hat because I want to hear them! Please send me your comments, corrections and considerations; because information like what I've exposed above starts conversations that will inevitably allow for better planning and decision making.

-Aaron

 

 

Ride the Ales! A Brewery Map for Modern Times

Taking a break from romance, I thought I would blog this week about my other favorite geographic topic: Beer!

As many of you know, I have two forthcoming apps that will use GIS and geospatial technology to make the world a more fun and interesting place.  In case you couldn't guess, one of the apps will involve dating.  The other is going to help everyone get a bit tipsier.  Indeed, I will be creating an application that helps people to get to sunny patios faster and in the nick of time to catch some rays.

While my PatioFinder app is in progress, I'll share with you folks a homage to one of my favorite cartographic styles: the Beck's London Tube Map of 1931.  I simply love this style and have seen it applied to so many maps and diagrams so excellently. I also love touring around Vancouver by bike sampling our diverse array of craft breweries. So....

The set up:

  • I sourced the most current list of breweries that I could get my hands on from beermebc here;
  • I devised some logical groupings of breweries based on geographic location (and cut out anything south of Dageraad);
  • I plotted and connected the dots using a shortest path route finding algorithm to determine my routing combinations;
  • I then plotted the breweries and the routes and then transferred them to a hexagonal grid to start the schematic process;
  • With a bit of artistic license, I ended up with the top figure;
  • I then used some fancy GIS to figure out timing along each route and used something called linear referencing to build the timetable for the bottom figure; and
  • You can see the results from the real world data and my transfer to "creative space" from left-to-right in the slider image below. (note: creative space can be easily accessed after only a few pints)

The results:

Well this worked out well, I would say.  I didn't have any overarching analytical goal except to use some skills I obtained building strip maps for pipeline alignments and applying them to something enjoyable like trip planning.  There are some notes and caveats that go with this map:

  • Schematic diagrams in GIS are hard to do.  Everything in GIS-land is made to be accurate and precise.  Building a schematic is neither.  Thus, I had to "force" my software to conform to my hexagonal wishes and requirements for reference lines... no easy task
  • I didn't include breweries to the south of Dageraad.  I'm only a little sorry about this, because my page size didn't work with the kind of scaling and I wasn't up to schematizing all of the arms of the Fraser. maybe next week....
  • If this has been done before, please let me know how you did it, as I am sure there are better ways to create a schematic diagram than the method I outlined above.

And with all of those out of the way, please check out and enjoy the final product.  If you want to suggest alternate names for the routes please feel free to do so in the comments below, and I will update the map. A downloadable version is available here.

Finally, I can't help but do some more analysis of population change and things that are cool and interesting in Vancouver.  Take a look at the interposition of breweries and population change over the last five years in Vancouver... notice anything?

 Breweries and population change hmm....

Breweries and population change hmm....

Hello World!

Hello world and welcome to Licker Geospatial Consulting Co.'s Blog.

On this blog, you will be able to view and interact with a weekly selection of intriguing GIS, plus random posts related to the business side of LGeo.  Feel free to share, tweet, email, pin, pass and otherwise disseminate anything I post here.  Simply credit myself, Aaron Licker, or my company, LGeo, and everything should be cool. Comments are welcome. Constructive feedback is welcome as well, as I am new to all of this and you folks know much better than myself about how to communicate socially.  To whet your appetites, here is a list of some of the blog topics i'll hit in the next 52 weeks.  More to come and stay excited!  

  • Better mapping the cultural crawl
  • The beard index (hipster businesses, services and urban markings)
  • Optimization of brewery tour routes
  • Density Transects
  • Mapping the effects of moving the hospital
  • Mapping historic streetcar routes and their relation to small scale commercial
  • Machine space calculations
  • Mapping where the worst beer lists are
  • Building permit analysis (i.e. bigger homes or more homes)
  • Transit density analysis based on google transit feed
  • Mapping historic rail lines that are now roads or alleyways
  • Mapping suburban intrusions into the urban fabric (chains versus independents), unique versus homogenized landscapes.
  • Open data monthly – deriving insights from random community’s open data stacks
  • Mapping cultures and sub-cultures
  • Random geography questions - how far do I have to walk to get to a:
    • Bus stop in Mississauga
    • Depaneur in Montreal
    • Donair joint in Halifax
    • BBQ place in Austin
    • Coffee shop in Portland

-Aaron