Geography 372

Document Sample
Geography 372 Powered By Docstoc
					                                                Geography 372
                                          Introduction to Cartography
                                           Lab 6 – Choropleth Maps

Choropleth maps are the most common type of quantitative thematic map. Data on many topics are commonly
captured as area-level variables, and these data are most commonly mapped in a choropleth style. In particular,
census data is collected for many different types of enumeration areas and is thus generally well suited to
choropleth mapping. In this lab, you will map several census variables for Vancouver dissemination areas, using
ArcMap. ArcMap facilitates linkage of spatial data (e.g. a map (shapefile) showing the dissemination areas of
Vancouver) with associated attribute data (e.g. census variables).

Classification of your data is, arguably, the most important of the many processes involved in producing a
choropleth map. Since a cartographer can greatly alter what is communicated by a choropleth map by changing
the classification of the data, it is very important to develop an understanding both of the data's characteristics
(e.g. is the data normally distributed or uniformly distributed, and does it have any outliers?) and of the purpose
and audience of the map (e.g. a map produced by the Downtown Eastside Resident's Association might be very
different from a map produced for the Chamber of Commerce using the same data since each group would likely
have a very different perspective). There are many options available for classifying data (e.g. the method used to
place the data into classes, the number of classes to be used), and only by exploring the options will you be able to
see for yourself how the message of the map can be manipulated.

In this lab you will choose a census variable and classify your dataset using four of the six classification methods
available in ArcMap (these are only a few of the methods available to cartographers) and then prepare four maps
(black and white) of the variable, classified four ways. Your maps should illustrate how different classification
schemes can present four different views of the same data. You will then select the map you feel best represents
the spatial distribution of the data and prepare a final colour map.

Guidelines for Choropleth Map Design
   • Use data that are uniform throughout the areal unit (e.g. the dissemination area)
   • Since areal units usually vary in size and population, we usually do not map raw totals. Instead, use
       derived values, such as ratios, rates, proportions or percentages
   • The 'best' classification method depends on the data, the purpose and the audience of the map
   • The full range of the data must be included and the class values should not overlap
   • Generally, use no less than four and no more than six or seven classes – five is usually a safe choice
   • Area symbols associated with classes (e.g. the colour saturation or value) must be easily distinguishable

Assignment (30 marks)

1. Add the data
A table of 2006 Census immigration variables was created for this lab and has been joined to a shapefile of
Vancouver dissemination areas (DAs). A shapefile is a spatial file for use in GIS that can show points, lines, or
areas. In this example, our shapefile is a map showing the boundaries of all the DAs in Vancouver. Shapefiles
always have tables of data associated with them, and the immigration variables have been appended to this table.

Use GetData to load the dataset. Start ArcMap, and use the Add Data button to add the shapefile VANda06.shp. A
map of Vancouver DAs will appear in your data frame. To view the attribute table of census variables associated
with each DA, right-click on VANda06 in the Layers list and select Open Attribute Table.

Each row in the table refers to a DA, and each column to a variable describing the DA – ignore the first five
columns, but have a look at the following six. Each variable (column) is labeled in abbreviated form:
       tot_pop                    Total population
       Can_cit                    Canadian citizens
       notCan_cit                 Not Canadian citizens

        tot_immi                  Total immigrants
        immig_eur                 Immigrants from Europe
        immigAsia_ME              Immigrants from Asia and the Middle East

Close the attribute table when you are finished exploring.

2. Standardize the data
Right click on VANda06 again, and select Properties. In the Layer Properties dialog box, select the Symbology
tab. This is where you classify and symbolize data. At the moment, all features (the DAs) are drawn using the
same area symbol. To colour each DA separately, click Categories in the list at the left of the dialog box, choose
‘DAUID’ (abbreviation for DA Unique ID) as the Value Field, and click Add All Values (toward the bottom of the
dialog box). Click Apply, and each DA will be coloured a different colour. If you want, choose a different Colour

Next, try colouring the DAs based on one of their census attributes. Select Quantities in the list at the left of the
dialog box, and in the Value Field select ‘Can_cit’ and click Apply. Can_cit is a raw number variable, and seeing
as you are creating a choropleth map, you should standardize this value to show the percent of population in each
DA that are Canadian citizens. ArcMap uses the term ‘normalize’ to refer to standardization, so, in the
Normalization Field choose tot_pop. ArcMap will automatically divide the numerator (Can_cit) by the
denominator (tot_pop) for each DA, creating a variable that is a proportion instead of a raw total.

Note: It is important that you normalize the variable you are mapping by the appropriate denominator. For
example, for number of persons who are married, you would normalize by the total population 15 and over. We
would not expect anyone under 15 to be married, so it doesn’t make sense to normalize by the total population.
Variables such as unemployment rate and average income should not be normalized since they are derived values
(i.e. averages) already. To normalize total population you could use the area of each DA (square km) as the
denominator, creating a population density map. You could, of course, use area as the normalizing variable for
many of the other variables – the variable you choose to normalize by depends upon the message you are
attempting to communicate.

Notice that ArcMap classified your data for you. The default classification method is Natural Breaks (Jenks) with
five classes. Click OK to see your map. Notice that a legend is now displayed in the Layers list. It shows that in
the DAs with the highest proportion of Canadian citizens, 93-101% of the population are citizens, and in the DAs
with the lowest proportion of citizens, only 43-67% of the population are citizens. The 101% value seems a little
high (well, impossible) – it is due to the rounding rules that Statistics Canada follows. You can safely refer to it as

3. Classify the data
To classify the data using another classification method, open the Symbology tab again (right click on VANda06
and select Properties), and click Classify (to the right, toward the top).

In the Classification dialog box you can choose from six classification methods: Equal Interval, Defined Interval,
Quantile, Natural Breaks (Jenks), Geometrical Interval and Standard Deviation. Spend some time trying out
different classification methods and different numbers of classes. Pay close attention to the Classification
Statistics at the top right and to the graph below. This graph is a histogram, and it shows the distribution of your
dataset and where your class boundaries fall. The x-axis shows the range of your variable (0.4342, or 43%, to
1.0099, or 101%) and the y-axis shows the number of DAs that fall on each data value in the range. The blue lines
show the class breaks in your current classification.

1. Describe the distribution of the data. What is the range? The mean? Is the data normally or uniformly
distributed, or is it skewed? Are there outliers? How does this histogram help you in choosing an appropriate
classification scheme? Include an image of the histogram in your answer (use the Print Screen key to take a
screenshot of the histogram, then paste the image into Word). (2 marks)

You may want to use the ArcMap Desktop Help to learn more about these classification methods. In the Help
Contents, expand Mapping and visualization, then Symbolizing data, then Applying symbology, then Standard
classification schemes.

4. Choose your own variable
Repeat the process you just went through with the proportion of population that are Canadian citizens variable to
make four maps of a variable of your choice, using the Equal Interval, Quantile, Natural Breaks (Jenks) and
Standard Deviation classification schemes, each with five classes. Ensure that you normalize your variable
appropriately. Try to choose a variable where the different classification schemes create maps with different
messages. Notice that in the Symbology tab you can choose an appropriate colour ramp for your data. These maps
can be in grayscale (choose the gray ramp from the selection of colour ramps). Pay close attention to the
histogram as you classify your data – make notes about how (un)evenly the DAs are distributed between the
classes – these will come in handy for your lab write-up.

For each map you will make a quick layout to print your results. Once your map with your first classification
scheme is ready in the Data Frame, select Layout View from the View Menu. To change your layout to
Landscape, choose Page and Print Setup from the File Menu. Resize your map if necessary, and add a title, a
legend, and your name using the options in the Insert Menu. Make sure you note on the map what variable you
have mapped (including how it was standardized) and what classification scheme you used. Print your map, and
repeat the process for the other three classification methods. (8 marks)

5. Create a final map
Now that you have explored the distribution of your variable and seen how the four classification methods
classify your variable differently, you are ready to select what you feel is the best classification option and make a
presentation-quality colour map.

Return to the Data View (View Menu) and follow the same steps as above to symbolize and classify your data
(using the Symbology tab), but this time select an appropriate colour ramp instead of grayscale. In the Symbology
tab you should change the symbol labels from proportions (0.4342) to rounded percentages (43%) to make your
legend easier to read. You can round to the nearest whole percent, or to one or two decimal places. Think about
what would most help your audience. You can also click the checkbox beside Show class ranges using feature
values to create a legend that is gapped instead of continuous.

Use the Add Data button to add background.shp to your map, so that you can see the outline of all of Vancouver
and UBC and the DAs that are blank due to there being no immigration data for them. (Move background.shp
below VANda06 in the Layers list and turn it pale grey or an appropriate colour.)

Return again to Layout View and create a final layout with all the necessary cartographic elements – informative
title, legend, source statement (the source is 2006 Census, Statistics Canada), north arrow, scale, neatline, and
your name and lab section. All these elements can be added using the Insert Menu. Make sure you note the
classification method you used in your legend. If you would like to edit your legend once it has been added, right-
click on it and select Convert To Graphics, then right-click again and select Ungroup. Now it is editable. You can
do the same to your scale and change it to kilometers. When your final layout is ready, print it out in colour (5

2. Print out the histogram of your chosen variable, and in one page discuss (11 marks):
    • The variable you chose and why you chose it (1)
    • How and why the variable was normalized (1)
    • How the four methods of classification changed the number of data points (DAs) within each class (2)
    • How the four methods of classification changed the spatial representation of the data (2)
    • The pros and cons of each classification method (2)
    • How different classifications might be appropriate for different purposes and/or audiences (2)
    • Why you selected the classification method you did (1)

3. Why is a dichromatic colour scheme appropriate for a standard deviation classification scheme but, generally,
not for the other classification schemes? (2 marks)

4. How valid is the assumption that data are uniform throughout the areal unit, and how could you alleviate the
problem? (2 marks)

Make sure you hand in:
  • Four grayscale classification maps (8 marks)
  • One final colour map (5 marks)
  • Answers to questions 1-4, with histograms (17 marks)