Thematic Maps

1
Thematic Maps

• Dr. Baqer Al-Ramadan
• Feb 2004

2
Thematic Maps

• One of the most frequently used GIS functions is
  the creation or use of thematic maps. These
  are statistical maps that display a particular
  theme of data.
• Thematic maps can be made for geospatial features
  of points, lines, or polygons.
• They display the spatial pattern of the value of
  a geographic features attribute - and thus
  visualize a particular spatial theme.
• To create well-designed thematic maps we need to
  think about good statistical practice, graphics
  and symbolization.

3
B
BOUNDARYFILE
E
D
A
C
DATAFILE
Thematic Maps
4
Thematic Maps Cartographic Generalization

• Generalization Tools
• Map Type
• Symbolization
• Scale
• Classification

5
Thematic Maps Map Types - Many Different Kinds,
a few examples

• AREA CLASS boundaries based on attribute, also
  called constant attribute maps - such as a soil
  map.
• CHOROPLETH boundaries based on preexisting
  reporting zones - such as a census population
  map.
• DOT DENSITY count data, number of points
  (randomly placed) in an area represents the value
  of the attribute being mapped.
• PROPORTIONAL SYMBOL point or line data, size of
  each point or line represents the value of the
  attribute being mapped.

6
Thematic Maps
Map Types - examples
Choropleth Map - Median house value, 1990, census
tracts, classified by standard deviation
Area Class Map - land use/land cover, 1986
7
Thematic Maps
Dot Density Map - Population 1990
Proportional Symbol Map - Population 1990
8
Thematic Maps Cartographic Generalization

• Maps ALWAYS involve cartographic generalization.
  Some good rules to follow
• design to maintain relevant characteristics,
• design to send a single, clear message ,
• design for your audience AND your goal,
• emphasize or deemphasize features,
• use multiple map representations, when
  appropriate.

9
Symbolization
USGS Point Symbols

• Point symbols and icons can show
• churches
• schools
• geodetic monuments
• stream gauging stations
• mining activities (mine, tunnel entrance)
• hydrographic features (spring, well)

10
Symbolization
USGS Line Symbols

• By varying line weight, color, and fill, USGS
  cartographers can represent many different linear
  features such as
• railroads
• rivers streams
• streets highways
• power lines
• state, county, municipal, various other
  boundaries

11
Symbolization
USGS Area Symbols

• The use of shading and patterns can show
• water
• forested areas
• orchards
• urban/developed areas
• areas that have changed since the region was last
  mapped

12
Symbolization Area Symbolization

• Chose Map Type wisely classify wisely.
• Avoid area patterns that use lines at different
  orientations, or bright red shades with bright
  blue. There is an extensive literature on the
  color theory of such maps, including helping
  those that are color blind.
• Assign shades or pattern densities to follow
  values (e.g. lowest value lightest shade,
  highest value darkest shade). This is what
  readers expect.
• Use boundary line type of indicate confidence
  (e.g. solid line confident of line placement,
  dashed line not so confident). Use a careful
  choice of shading to represent polygons with
  inadequate data or data with disclosure issues.

13
Map Scale

• The only true one-to-one representation of the
  world is the world itself. All maps MUST
  generalize.
• Generalization means omission, simplification,
  displacement, and aggregation.
• ALWAYS remember that when you use maps or make
  maps!

14
Map Scale

• In order to be useful, the map must be smaller
  than the land that it represents no 11 maps!
• A good map will always tell you what the size
  relationship is between the map features and the
  real-world features they represent.
• This relationship is the scale of the map.

15
Map Scale
Level Of Detail
Photogrammetric Maps - features shown
1600 map (large scale) detail
12,400 map (smaller scale) detail
Manholes Catchbasins Street signs Fire
hydrants Curbs
Center lines of roads Railroads Rivers,
Lakes Large buildings
16
Map Scale
Three ways to describe the scale of a map

• Verbal - 1 inch 2,000 feet or 2,000 scale
• Graphic - scale bar
• Representative fraction - ratio of map distance
  to real world distance

17
Map Scale
things look LARGE at large scales
Large
124,000
1500,000
13,000,000
Small
things look small at small scales
18
Map Scale
Map distance Real-world distance
Representative Fraction
1250,000

• One way of conveying the scale relationship of
  map to the real-world is by using a
  representative fraction (RF).
• An RF of 124,000 tells you that one unit of map
  distance is equal to 24,000 of those same units
  in the real world.
• For example if two points are 1 inch apart on a
  map, they should be 24,000 inches (or 2000 feet)
  apart in the real world.
• RF is nice because it is unitless, i.e., 1 cm in
  map distance 24,000 cm (240 m), bad because if
  map is resized (or copied) it is incorrect.

19
Levels of Measurement
Data Classification

• In addition to data types the GIS user needs to
  be aware of the levels of measurement of the
  attribute data
• nominal
• ordinal
• interval
• ratio

20
Nominal
Data Classification

• Objects are classified into groups. The groups
  have names, not numeric values. There is no
  ordering implied. Also called categorical data.
• Examples gender, ethnicity.
• Examples soil type, land use, zoning.

21
Ordinal
Data Classification

• Has the concept of an ordering.
• Example Opinion poll response of strongly
  agree, agree, disagree.
• Example Soils can be ordered from poorly
  drained through somewhat poorly drained
  through well-drained through excessively
  drained.
• We can assign numbers to these categories, but it
  doesnt automatically imply we can use arithmetic
  relationships.

22
Interval
Data Classification

• Moves into the quantitative realm.
• Places an object on a number line with an
  arbitrary zero point and an arbitrary interval
  (choice of distance to be called one).
• Example years (on Gregorian or Islamic
  calendar) - 2000 is not twice 1000 in any
  significant sense.

23
Ratio
Data Classification

• Quantitative attribute that has a true origin
  (zero value) and an arbitrary interval.
• These attributes support the arithmetic
  operations.
• -Examples age of structure, assessed value of
  a parcel of land.

24
Some other useful attribute types in GIS
Data Classification

• Counts Close to ratio level, but unit of count
  not arbitrary, so we have to be careful in
  rescaling.
• -Example Population count in a census tract.
• Fuzzy sets Nominal categories arent always
  simple - an object may have a degree of
  membership.

25
Data Classification Things to be aware of when
creating thematic maps...

• We need to be clear on the type of attribute we
  are mapping. Is the attribute nominal, ordinal,
  interval, or ratio?
• The map of a nominal attribute field is often
  called a unique value map. Why does this make
  sense?
• Ex soils, land use, land cover, zoning,
  forest types.
• The map of an ordinal attribute field should
  typically be shaded using graduated symbols or
  shade patterns. Why does this make sense?
• Ex soil suitability for development
  (unsuitable to highly suitable), building
  condition (poor to excellent)

26
Data Classification Things to be aware of when
creating thematic maps

• When we have ratio or interval attribute fields,
  we will
• need to divide the data up into classes.
  There are
• various ways to do this. (e.g., equal intervals,
  equal areas, quantiles, standard deviations,
  natural breaks, user defined)
• When we have ratio, interval and count data, we
• will often need to normalize the data.
• Ex We will map population density, rather
  than population count (normalize by area). We
  will map minority population, rather than count
  (normalize by total).

27
Data Classification Classification Methods

• 1) Examine the range and statistical distribution
  of data (use the histogram).
• 2) Ascertain how many class are appropriate, (7
  classes 2 classes provides a rule-of-thumb).
• 3) Some classification procedures use algorithms
  to minimize within group variation maximize
  between group variation.

28
Data Classification Classification Methods

• Natural Breaks appropriate for non-normal
  distributions. (default)
• Constant Interval appropriate for an even
  (uniform) or normal distribution.
• Standard Deviation appropriate for a normal
  distribution.
• Quantile appropriate for an even or normal
  distribution.
• Constant Area used scientifically, for
  statistical analysis of spatial distribution
  within the probability distribution.

29
Data Classification Classification Methods

• Natural Breaks bimodal, multi-modal or other
  non-normal distribution.
• Sometimes distributions are very skewed (way to
  the left or right). Sometimes distributions are
  arithmetic or geometric progressions.
• In these cases use another software package to
  transform the data so it approximates a normal
  distribution.

30
Natural Breaks

• This is ArcView 8.1s default classification
  method.
• Identifies break points by looking for groupings
  and patterns inherent in the data. Algorithm
  minimizes the variance within the groups while
  maximizing the variance between the groups. Uses
  Jenks optimization algorithm. Extreme values are
  obvious.

31
Natural Breaks
32
Quantile

• Each class is assigned the same number of
  features. For example, when we divide into 5
  classes (quantiles) the features in the first
  class are the 20 of the features with the lowest
  values, the second class is the 20 with the next
  lowest values, etc. This works like the
  percentiles of reported standardized tests. If
  your score is in the top quantile, you are in the
  top 20 of the test scores.
• Best suited for a data set that does not have a
  large number of features with similar values.

33
Quantile
34
Equal Area

• Classifies polygon features by finding
  breakpoints in the attribute values so that the
  total area of the polygons in each class is
  approximately the same.
• Classes determined with the equal area method are
  typically very similar to Quantile classes when
  the sizes of all the features are roughly the
  same.
• Polygons with the largest values tend to hide
  variation in population between geographically
  smaller areas.

35
Equal Area
36
Equal Interval

• The range of attribute values is divided into
  equal sized sub-ranges. Dividing into 5 equal
  intervals means taking the range of values (max
  min) and then dividing by 5 to calculate the size
  of each sub-range (e.g., if the range was 300
  (min 0, max 300), each sub-range size is 60
  for 5 classes).
• Useful when you want to emphasize the amount of
  an attribute value relative to another value. Not
  good if you want to reveal subtle differences
  between features with similar values.

37
Equal Interval
38
Standard Deviation

• Shows you the extent to which a features
  attribute value differs from the mean of all the
  values.
• ArcView first finds the mean value, calculates
  the standard deviation, and then places the class
  breaks above and below the mean at 1, .5, or .25
  standard deviations.
• ArcView will aggregate any values beyond three
  standard deviations from the mean into two
  classes gt3 Std Dev and lt3 Std Dev. In
  statistics we call these values outliers.

39
Standard Deviation
40
User-defined Breaks can mislead