1.1.4. KML Cartography
Information
By combining Styles with the various geometric symbols that are available, we can use established cartographic knowledge when creating road
maps, dot maps, and various other familiar forms of cartography.
Let's consider area shaded '
choropleth
' maps by way of example as we are interested in mapping the population and using the results in our visualization.
When portraying aspects of the population using choropleth
maps cartographers would usually consider mapping densities or ratios rather than counts. The absolute numbers returned
in counts tend to be highly dependent upon the areas used to aggregate the population being surveyed and so can be more indicative
of the size and scope of the zones into which people are grouped than the phenomenon under study. In effect, we can consider
a choropleth as a spatial histogram with bins of uneven size. Standardizing the counts or using a denominator can address
the effects of irregularly sized bins (Dykes et al. 1998). For example, it may be considered more appropriate to :
- use the areas of the zones as a denominator when considering population totals and thus map population density rather than
population count
- map secondary variables representing characteristics of population and use the population total as a denominator, thus mapping
ratios
We can also consider many of the phenomena that we study at a number of spatial scales.
For example, in the US, Census 2000 data were released using a number of alternative 'geographies' including national level,
state level, county level and through individual census tracts.
Any geographic effects that are observable or measurable may only be evident at a particular scale.
Under these circumstances it is appropriate to produce and consider a number of cartographically informed maps and use visualization
to help us explore the characteristics of population. These can take advantage of some empirically derived guidelines, such
as those of Mark Harrower and Cindy Brewer (Harrower et al. 2003) for selecting colour schemes as implemented in the online ColorBrewer software (Brewer 2002).
Example
The KML
elements that we have used thus far provide us with plenty of scope for producing maps that take account of such considerations
and for viewing them in Google Earth
so that we can compare them with each other and the various data sets available. Doing so may help us understand the nature
of our data and the geography it represents.
For example, we may want to use Polygon elements within Placemarks to map population counts and compare these with population densities.
The examples provided here show the census tracts of Franklin County.
Data describing the geometries are accessible from a number of online sources and can be processed using a Geographic Information
System or generalized using "MapShaper" (Bloch et al. 2006) or similar services.
Census 2000 population counts (left) and densities (right) for Franklin County, Ohio.We may decide to map ratios rather than absolute numbers, or at least compare the maps of counts with those that show ratios
or relative quantities, such as change over time.
We should do so with an appropriate colour scheme that diverges around a zero value. These examples compare the 2006 population
estimates (U.S. Census Bureau 2008) with the Census 2000 resident population for the counties of Ohio.
Change in population : Census 2000 - 2007 estimates. Absolute change - counts (left), relative change as a percentage of population
- ratios (right).We may vary the colour schemes according to how we expect the map to be viewed.
These examples map the absolute 2006 population counts for the US at state level using alternative ColorBrewer schemes (Harrower et al. 2003).
Census 2000, total population by State - alternative sequential schemes YlOrBr (left) and BuPu (right).
Exercise
The issues and examples introduced here can lead us to consider a number of questions.
See if you can begin to answer those that follow by exploring choropleth
maps showing population counts and associated derivatives at different scales in Google Earth
.
Access the maps through the Places window in Google Earth
.
Use the checkbuttons to toggle Places on and off.
The Places ordered highest in the list will be uppermost and visible in Google Earth
. Places can be re-ordered in Google Earth
by dragging the names in the list.
Questions regarding the national maps of conterminous states ...
- What can we infer from the total population of the US at state level?
- What does the population density of the US at state level tell us about the states (and the data model)?
- What patterns might we infer when considering change in population (2000-2006) at state level?
Questions regarding the maps of counties in Ohio ...
- What can we infer by looking at the total population of Ohio at county level?
- What does the population density of Ohio at county level tell us about the counties?
- Is there a spatial pattern in change in population in Ohio when the 2006 estimates are compared with the Census 2000 population
totals?
Consider the ...
- changes in terms of absolute numbers of people
- relative (percentage) changes in population
Questions regarding the maps of Census 2000 tracts in Franklin County ...
- Which tracts contain the most people?
- Which tracts contain the densest populations?
- What does this information tell us about the population of Franklin County?
Think in particular about what your answers might tell us about the geographies, data models and maps that we are using to
explore population issues.
Use the data provided in Google Earth
, including the air photography and ancillary data sources to help.
Notes
We can generate a variety of choropleths such as those used in this exercise in KML
by using a URL with a series of parameters.
These specify the geography, variable and cartography required in our maps.
Parameters and values are case sensitive and separated by & symbols.
The following data URL was used to generate the examples provided here and is of the form :
http://www.gicentre.org/data/uspop/data.php?d=D&v=V&cbS=S&cbN=N
The four parameters can have a number of values through which the KML
that is returned can be varied.
-
d : geographic data
The value of 'D' given to the 'd' (geographic data) parameter, can have three values.
- 'st' : conterminous states at state level
- 'co' : counties of Ohio
- 'tr' : Census 2000 tracts for Franklin County, Ohio
If the 'd' parameter is omitted the geography defaults to conterminous states at state level - 'd=st'.
-
v : variable
The value of 'V' given to the 'v' (variable) parameter, can have six values.
- 'p00' : resident population 2000 (counts)
- 'p07' : resident population estimate 2007 (counts)
- 'p00d' : population density 2000
- 'p07d' : population density 2007 (from estimate)
- 'pDa' : absolute change in population 2000-2007 (counts)
- 'pDp' : percentage change in population 2000-2007 (ratio)
If the 'v' parameter is omitted the variable defaults to Census 2000 resident population - 'v=p00.'
The 2007 population estimates are not available for Franklin County census tracts - 'd=tr'.
-
cbS : colorBrewer Scheme
The values of 'S' given to the 'cbS' (colorBrewer colour scheme) parameter and 'N' given to the 'cbN' (colorBrewer number
of classes) parameters can have a whole range of values to define schemes defined in ColorBrewer and listed here :
For example ...
- 'cbS=YlOrBr&cbN=5' : YlOrBr sequential scheme with 5 classes
- 'cbS=YlGn&cbN=7' : YlGn sequential scheme with 7 classes
- 'cbS=BuYlRd&cbN=6' : BuYlRd diverging scheme with 6 classes
If the 'cbS' parameter is omitted the colour scheme defaults to YlGn - 'cbS=YlGn'. If the 'cbN' parameter is omitted then the data range is classified into 9 equal intervals - 'cbN=9'. If a diverging scheme is used then the classes are allocated around zero.
You can use the 'data URL' introduced above along with the listed parameters and values to create additional maps for exploring the population data
sets in Google Earth
.
