Gordon Stokes Transport
2011 census journey to work data mapping
Last updated - Sept 2017
More on ...
Warning - slow map loading and other potential hassles!
Most of these maps use Scalable Vector Graphics (SVG) which behaves
differently in different browsers
and different computers. The maps are created within the web page, rather than slotted
in as image files. The web page writes the svg files from data.
The maps are research tools rather than light
entertainment, and take time to load.
For most maps you will need a browser window of about 1100 by 900 pixels to show properly.
They're not phone friendly.
On a fast computer they're not too slow, but on a
small laptop that I have they can take 30 seconds to load, and 20 seconds to change map,
with blank screens and "browser not responding" type messages in between.
My advice is to be patient, but give up if you get really impatient!
I've made a 'lightweight' version for some which may help show if it's my coding
not being compatible with your browser,
or your slow computer that's causing problems!
Introduction
The 2011 census in England and Wales asked for people's method of travel to work, and the location of workplace. The data doesn't cover everybody who works, as many don't regularly go to the same workplace. Also, working from home complicates the issue, and the home address is the permanent one - some people live closer to work during the working week. These maps show various combinations of:-
- The percentage who travel by each mode
- The numbers of journeys between census "Middle level Super Output Areas" (MSOAs) by different modes
- The percentages whose journeys are of different lengths, and
- The percentage within built up areas (BUAs) whose home and workplace are in the same BUA.
Maps of journeys to work by different modes
These maps show journeys to and from work for selected single modes,
and also one for all major modes. They shows flows in each direction,
with a curved line 'homing in' on the destination by a sharper curve -
imagine travelling somewhere fast then searching for a parking place.
They should work with a browser window of any reasonable size.
The single mode maps generally load quicker,
if you're only interested in one mode at a time.
- They allow zooming and panning, but only using the control buttons around the edges, not a mouse or touchscreen.
- They start centred on Oxford, but you can move to other areas using the orange direction indicators (N,S.E,W and diagonal)
- The bigger the browser area (viewport) the larger map you'll see, but if you increase the window size while running it, you have to reload the map
- What happens in local areas will become clearer if you zoom in
- Some maps show implausibly long journeys. I think this is imperfections in the data. ONS know that there are various errors in it. I have filtered out driver journeys over 50kms
Depending on your browser some parts of the
map links below may appear before others. Placenames are the last layer to appear.
Please wait till then (and maybe a little bit longer) before using menus.
The one for all modes and the driver one have a 'dashboard' towards the bottom right,
which I haven't refined yet, but you can place some modes in front of others,
or make them invisible (holding the mouse over 'guide' may help explain).
Figures after each link show seconds to load on an oldish 64 bit computer
Map showing busiest links (good for checking it works)
(3 seconds to load on my computer - could be less or more)
Map showing bus journeys only (4 seconds)
Map showing cycle journeys only (2s)
Map showing driver journeys (10s)
Map showing all main modes -
only shows heavily used driver flows (7s)
Pie charts of journey to work characteristics for settlements
This map is designed to give an indication of
the degree of self-sufficiency of settlements,
by showing the percentage of workers who live in a settlement who also work in it.
It also allows you to look at the percentages using different modes,
for walk, cycle, public transport modes, driver and 'other',
both grouped together and separately, for ONS defined settlements.
In addition it can show the proportion who make short and long journeys,
under 2 and 5kms, and over 30 and 60kms (straight line distances).
- It allows zooming and panning, but only using the control buttons around the edges
- The very small pies will become clearer if you zoom in.
- For most people the lightweight version will be fine as it includes settlement down to about 10000 population. But if you want details of the modes used in small towns and villages you'll need the heavyweight version.
Link to "lightweight" pie chart map - 1000 settlements (4 seconds)
Link to "heavyweight" pie chart map - 6621 settlements (17s)
The size of pie chart is related to the population of the settlement,
but not in a linear way.
Larger settlements are shown relatively smaller than lower population ones
so large ones don't cover smaller ones very often - except for insignificant
place like London :-), while villages aren't so small as to be invisible.
You may wonder why, for example, Greater London is not shown as one settlement.
The ONS have 'built up areas' and 'built-up-area sub-divisions'.
I explain how I handled this in the 'techie' section below.
For individual modes, besides the size of the settlement and the size of the
pie slice I've used graduated colours to help identify the places that have
relatively high or low use of different modes.
This is helpful to me in terms of seeing patterns.
I think the two driving ones work best in this respect.
I hope to work on this idea to improve it.
The reason that pies for workplace based are different sizes to residence based ones,
is because settlements don't have the same number of workers as residents who travel to work.
In general, the larger the settlement, the relatively more jobs it has.
Journeys from MSOAs to larger settlements
As larger settlements have their workplaces spread around
rather than in a well defined 'centre' they can give confused pictures
using the MSOA maps. The map below may be useful, showing trps from MSOAs to
ONS defined settlements.
Map showing journeys by different modes to larger settlements (10 seconds)
The busiest links for car users (down to the not so busy)
The map for car driver journeys, nearer the top of this page, categorised journeys by journey length. In this map the layers show fequency of journeys recorded between MSOAs. It start by showing all car journeys with more than 25 journeys between MSOAs. You can filter out less busy ones using the dashboard.
All driver trips between MSOAs with over 25 trips (9 seconds)
Some technical points for techie people
Data sources
Data on journeys was all taken from the 'Nomis' web site and from the public area
of the UK data archive "FlowData downloads page" ("WICID").
The data on modes of travel and distance travelled (non flow data) are
from tables QS701EW (mode) and QS702EW (distance) which are found in 2011 Quick Statistics.
All data used is public access.
The data from journeys between Medium level Super Output Areas (MSOAs)
is from table WU03EW. (The data for self containment of settlements
is also from WU03EW, but used more complex processing which I'll explain later).
There are some errors in the data. Some are deliberate to ensure
that individuals can't be identified, but there seem to be others
(which ONS are aware of). For some maps I have deleted flows of over 70kms
on the assumption that 25 people travelling that distance from one MSOA to
one other is more likely to be an error than the truth. For rail there are cases
where high flows over long distances are likely to be correct,
so I didn't alter rail ones.
Settlements - Built-up-areas
ONS built-up-areas have two levels. Built-Up-Areas (BUAs) and Sub-Divisions of
these (BUASDs). Built up means what it looks like, and any area that is
contiguously built up is part of one area. This means, for example, that Woking,
Hemel Hempstead and Harlow are part of the Greater London built up area, though few
would think of them that way. But at the same time each London Borough is a separate BUASD.
Short of looking at each of the 6000 plus areas in detail it's difficult to know
how to handle this best. I chose to use separate BUASDs where they exist,
so conurbations have separate 'settlements' even when they're joined to others.
Processing
Processing was initially done using the freely available Quantum GIS (QGIS)
program using
Middle level Super Output Areas
(MSOA) population weighted zone centroids from ONS, while settlement
centroids were created using QGIS from ONS boundary shapefiles, so are geographical
rather than population centroids.
The svg files that I use in the fully scaleable maps were made by extracting
grid reference data from the shapefiles, and attaching to the flow,
or settlement journey to work data. For the svg files written within the
web pages this data was put in an array. A curved line is drawn for each
MSOA to MSOA grid references, with the width proportional to the number of journeys made.
For settlement self containment I initially used MSOA level data of origins and
destinations, but had to decide which MSOAs should be in which settlement because the
settlements were defined at much more detailed 'Output area' (OA) level.
I wrote a small program such that any MSOA in which 70% of its OA were in one
BUA was assigned to that settlement. I recently switched to using a file showing
flows between OAs and Workplace Zones (WZs) which can be matched directly to BUAs.
That is what is used in the current maps (from June 2017).
Ackowledgements
I got the idea for curved lines from origins to destinations from Roger Beecham's
mapping of London bike hire journeys, and I owe him thanks.
I also thank ONS for making the census and locational data freely available to all,
QGIS for providing an excellent free GIS program, W3schools and Jakob Jenkov
for their tutorials on SVG etc,
as well as Stackoverflow for their queries which let me know what problems
other people faced in embedding javascript written svg code into html,
and whether there was a solution, or whether other readers felt the questioner
had asked the wrong question or shouldn't have started from there,
and whether someone else thought they were asking a silly question.
I never had the nerve to post a question myself, but then again,
I did learn a lot from these queries.
Gordon Stokes, 2016