1 of 48

Intro to GIS

Lesson 3: Cartography

1

2 of 48

Objectives

Use properties of symbols to differentiate features or rasters in maps
Distinguish among nominal, categorical, ordinal, and ratio/interval data
Create maps from attributes using different map types
Select appropriate classification methods when displaying numeric attributes
Display thematic and image rasters
Explain the relationship between a layer and its source data set
Use graphic design principles to create effective maps
Create text and labels on maps
Create map layouts and printed maps

2

3 of 48

3

4 of 48

Map Design and GIS

Cartography is the art, science, and techniques of making maps. With that said, poorly made maps (and even well made maps) can miscommunicate information.

Each map is just one of all possible maps.
Complex maps can be difficult to understand.

When a GIS map is the result of a complex analytical or modeling process, good design is essential for understanding.
Ultimately, the map is what distinguishes GIS as a different approach to the management and analysis of information.

5 of 48

Differentiating map objects

Changes in category are expressed by varying symbol shape, line type, pattern, color, or font
Changes in quantity are indicated by varying symbol size, thickness, or color

Access the text alternative for slide images.

5

Cartographers may choose from many strategies to symbolize features in a map. A layer may simply be portrayed using one symbol, or different features can be assigned different symbols depending on the value of an attribute field. Point data are shown with marker symbols, line features with linear symbols, and polygon features with shaded area symbols (Fig. 2.1).

Cartographers have many ways to signify differences between features in a map: shape, size, thickness, line type, color, pattern, and font (Fig. 2.1). Traditionally these variations are used to show either changes in category (the type of thing) or changes in quantity. Shape, line type, pattern, and font are typically used to show changes in category, such as different types of wells (points), different classes of roads (lines), or different soil units (polygons). Size and thickness are generally used to indicate increases in quantity, such as the population of a city (points) or the discharge of a river (lines) (Fig. 2.1). Text symbol variations usually indicate categories (towns vs. rivers), although font size can indicate qualitative differences in value, such as town size, as long as not too many different sizes are used.

6 of 48

Visualization

For a long time, visualization was treated as a powerful tool and approach to represent and explore both scientific and nature phenomena. As Buttenfield and Mackaness (1991, p.427)¹ note:�

"Visualization is an important component of any effort to understand, analyze or explain the distribution of phenomena on the surface of the earth, and will become increasingly important as volumes of digital spatial data become more unmanageable.” �

Visualization is broad, including human cognition, analysis of scientific or physical data, computer graphic display and so on. The following definition of visualization by Buttenfield and Mackaness (1991, p.432) is more sophisticated and complete: �

“Visualization is the process of representing information synoptically for the purpose of recognizing, communicating and interpreting pattern and structure.”

6

¹Buttenfield B and W Mackaness, 1991, Visualisation, in: D Maguire, M Goodchild and D Rhind (eds), GIS: Principles and Applications, Longman, London, Vol 1, 427-443.

7 of 48

Principles of Visualization: Color

Let’s explore how the human eye sees colors (and what colors it actually sees).�

The Human Eye has a radiometric resolution that ranges between 3 and 4 bit. That is, the human eye can not differentiate between more than 16 shades of gray.
On the other hand, the human eye is capable of distinguishing about 7 million colors.

7

8 of 48

8

Device that provides guidance in choosing colors.
Use opposite colors to differentiate graphic features.
Three or four colors equally spaced around the wheel are good choices for differentiating graphic features.
Use adjacent colors for harmony, such as blue, blue green, and green or red, red orange, and orange.

Color Wheel

9 of 48

H S V color model

Hue represents the color on a scale of 0-360 (the color wheel)
Saturation represents the intensity of the color (0-100)
Value represents the brightness of the color (0-100)

Source: Esri

9

The HSV method is instructive for discussing the use of color in portraying features on a map. HSV stands for hue, saturation, and value (Fig. 2.3). Hue refers to the shade of color (such as red, blue, or yellow) and is established by the wavelength of the light observed. Typically hue is portrayed as a color wheel so that the color values range in degrees from 0 to 360. Saturation corresponds to the intensity of the color and is measured as a percentage. Imagine mixing a can of paint, starting with a white base and adding a single pigment —a small amount of pigment yields a low saturation, but a large amount of pigment results in high saturation. Value refers to how light or dark the color is, somewhat like putting pigment into a can of base paint that varies from black through gray to white. Value is also measured as a percentage. Any color can be defined using a combination of the three properties.

10 of 48

R G B color model

Based on mixtures of red, green, and blue primary light
Each primary color brightness is indicated on a scale of 0 (black) to 255 (bright)
Each R G B triplet represents a different color

10

Color may be used to indicate either category or quantity. Colors may be designated using one of several common methods. The CMYK model is often used for printing, and it specifies mixtures of inks used in printers or plates (cyan, magenta, yellow, black). The RGB model is based on mixing different proportions of red, green, and blue light on a scale from 0 to 255. If the value of the red light is high (close to 255) and the other two colors have low values near zero, a bright red color will result. If red and green are both high and blue is low, a mixture of red and green will create yellow. (Mixing light is a different process than mixing paint, and different colors will result.) Figure 2.2 shows two possible mixtures and the resulting colors. The RGB model can specify 256 3 different colors (over 16 million)

11 of 48

Using color

a) Use different colors (red, blue, orange) for categories

b) Use different shades of the same color for quantities

c) Use shades of two colors for divergent quantities, such as negative/positive or cold/warm values

11

The set of colors chosen for a map layer should follow certain guidelines. Figure 2.4 shows three different sets of five colors. A set based on variations in hue should be used to depict changes in category, such as different soil units (Fig. 2.4a). For categories, the saturation and value of the colors should be similar. Quantities are generally indicated using differences in saturation or value, with light or unsaturated colors indicating lower quantities and darker or more intense colors indicating higher quantities (Fig. 2.4b). Sometimes a divergent color set is helpful in showing variation around a significant middle value (Fig. 2.4c), such as in a climate change map. Areas with no significant temperature change are shown in the middle neutral color, colder temperatures are shown in blue, and warmer temperatures are shown in orange.

12 of 48

Map Types and Data Types

Different maps are used to display different types of data

Single symbol maps are used for nominal data

Unique values maps are used for categorical and ordinal data

Many types of maps are used for numeric data

Graduated color maps.
Graduated symbol maps.
Dot density maps.
Chart maps.

12

13 of 48

Nominal data

Nominal data names or uniquely identifies objects

County names.
Airport names.
Tax I D numbers.
Parcel I D numbers.

Each feature is likely to have its own value

Usually portrayed on a single symbol map with optional labels

Source: Esri

Access the text alternative for slide images.

13

14 of 48

Categorical data

Features belong to categories

rock type, volcano type, highway class, or land cover class.

Category names may be text or numeric Portrayed with a unique values map

Source: South Dakota Geological Survey

Source: Esri

Access the text alternative for slide images.

14

15 of 48

Ordinal data

Ordinal data is a type of categorical data

Ranks categories along an arbitrary scale

Snail habitat rank:

(0) Unsuitable

(1) Marginal

(2) Acceptable

(3) ideal

Use a unique values map with a single-hue color scheme

Source: Esri

Access the text alternative for slide images.

15

16 of 48

Interval data

Interval data places values along a regular numeric scale

elevation in Oregon.

Supports addition/subtraction

If it can have negative values, it is interval data

Source: USGS

Access the text alternative for slide images.

16

17 of 48

Ratio data

Ratio data places values along a regular scale with a meaningful zero point

population of state capitals.

Supports addition, subtraction, multiplication, division.

Population can’t have negative values, so they are ratio data

Source: Esri

Access the text alternative for slide images.

17

18 of 48

Mapping quantities data

Interval and ratio data must be divided into classes before mapping

Quantities data are mapped using variations in

symbol size (city populations).
thickness (number of highway lanes).
hue (blue shades for average annual precipitation).

Many map types are suitable for numeric data

Source: Esri

Access the text alternative for slide images.

18

19 of 48

Classed maps

Place features into ranges and vary color or symbol size

Graduated color or choropleth maps use different colors.
Graduated symbol maps vary the symbol size.

Source: Esri

Access the text alternative for slide images.

19

20 of 48

Colors for choropleth maps

a) Use change in saturation or value to indicate larger quantities

b) Avoid rainbow color schemes for quantities maps; too many colors make maps harder to interpret

Source: Esri

Access the text alternative for slide images.

20

21 of 48

Modifiable Areal Unit Problem

Arbitrary aggregation units like states or counties may influence values

a) Number of farms in state is affected by size of state

b) Number of vacant houses in state is affected by population of state

Maps reflect the influence rather than the data being mapped

Source: Esri

Access the text alternative for slide images.

21

Data are often measured over one set of areal units, such as a census block, but then combined, or aggregated, into a set of larger areal units, such as block groups, tracts, and counties. When the measurement and aggregation units are arbitrary shapes and sizes instead of a regularly spaced grid, the size and shape of units will influence the recorded values. For some measurements, such as the number of farms or lightning strikes, units with larger areas tend to have higher values just because they are larger. In Figure 2.10a, the number of farms is greater in the bigger states like Texas and California. Certain variables, such as the number of home vacancies, are strongly linked to population, and maps made from them reflect population rather than giving insight into underlying patterns. In Figure 2.10b, the largest and most populous states have the most vacancies.

22 of 48

Minimizing M A U P

Normalizing (dividing) data by a suitable field allows data patterns to emerge

Farms per square mile instead of number of farms.
Fraction of housing units that are vacant instead of number of vacancies.

Source: Esri

Access the text alternative for slide images.

22

Another issue that can arise is that larger polygons receive greater prominence in the map simply because of their size. In Figure 2.10, the large western states attract more attention than the eastern states. Both of these issues are examples of a phenomenon known as the modifiable areal unit problem, or MAUP, which occurs when measurements are being aggregated over arbitrarily defined areas. Analyzing spatial or statistical patterns becomes more difficult because the patterns are obscured by the aggregation scheme. However, methods to reduce the effects of MAUP on maps are available.

One approach is to normalize the data, dividing each value by a specified variable. If one is concerned that the size of the aggregation area is influencing the magnitude of values, one can divide by the feature area as shown in Figure 2.11a, where the number of farms is divided by state area (compare to Fig. 2.10a). If the values are being influenced by population, then one can divide by population or an other suitable variable, as in Figure 2.11b, where the vacancies are divided by the number of housing units, giving a far more interesting and informative map than Figure 2.10b.

23 of 48

Normalizing by field percentage

Dividing each value by the total of all the values is another way to normalize data

Divide the number of congressional districts in each state by the total number of districts to view the percentage of Congress controlled by each state.

Source: Esri

23

24 of 48

Reducing visual M A U P

Use a different map type than graduated color

Density of farms portrayed using graduated symbols instead of graduated colors.
Number of farms portrayed with a dot density map.

Source: Esri

Access the text alternative for slide images.

24

25 of 48

Bivariate maps

Source: Esri

Use to portray and compare two different fields, such as median ages for males and females in Oregon counties

Counties with similar median ages for both groups are purple
Counties with higher median ages for females are pink, like Lane County
Counties with higher median ages for males are blue, like Jefferson County

Access the text alternative for slide images.

25

A bivariate choropleth map can be used to compare the spatial patterns of two numeric variables. Consider the map in Figure 2.13 showing the median ages of males and females in Oregon counties. The median age of females is shown with three quantiles (thirds) in light to dark pink, and the median age of males is shown in blue. In most counties, the median ages of both groups behave similarly, and the pink and blue combine to produce purple. When both male and female median ages are low, a light purple color results; when both are high, a dark purple is shown. In Jefferson county, males tend to be older than females and the county appears blue. In Lane county, females tend to be older than males, so the county appears pink. The next step might be to explore the socioeconomic characteristics of these two counties to try to determine the factors that produce this marked difference in median age patterns

26 of 48

Raster types

Source: Black Hills National Forest; (Land use): Source: USGS

Discrete data

Represent objects such as roads or land use polygons.
Take on relatively few values.
Adjacent cells often have same values.
Values may change abruptly at boundaries.

Continuous data

Represent a measurement that occurs everywhere.
Thousands or millions of potential values.
Few adjacent cells have same values.
Values may change rapidly from cell to cell.

Access the text alternative for slide images.

26

27 of 48

Thematic rasters

Contain quantities that represent map data such as geology or elevation
May be continuous or discrete

(a): Source: South Dakota Geological Survey; (b-c): Source: USGS

Categorical/ordinal rasters use(a) unique values or discrete color display
Interval/ratio (quantities) rasters use (b) classified or (c) stretched display methods

27

28 of 48

Stretching

Stretching reallocates a smaller portion of the raster values to the 256 shades of the color scheme
Improves brightness and contrast
There are many types, such as min-max or standard deviation stretches

(b–c): Source: USGS

Access the text alternative for slide images.

28

A thematic raster containing interval or ratio data must be classified into ranges in one of two ways. The stretched display method scales the image values to a color ramp with 256 shades (Fig. 2.15c). The raster is first subjected to a slice, which rescales the elevation values (ranging from 800 to 1600 m) into 256 bins, then matches the bins to the 256 shades in the color ramp. A slice is essentially a classification that creates 256 classes.

The stretched display method can improve the display of normally distributed values by ignoring the tails of the distribution. One method assigns the lowest and highest image values to 0 and 255, respectively, and stretches the remaining colors along the ramp. This is called a minimum-maximum stretch. Even greater contrast can be developed using a standard-deviation stretch, which uses only the values within two standard deviations of the mean (Fig. 2.17c). Stretches can be applied to thematic continuous rasters as well as images and can be helpful when the image values are not evenly distributed

29 of 48

Image rasters

Image rasters contain satellite or air photo data and generally represent brightness
They are displayed using the (a) stretched method for single band rasters or the (b) R G B composite method for multiband rasters

Source: USGS

Access the text alternative for slide images.

29

Image rasters include aerial photography and satellite data. The pixels represent degrees of brightness caused by light reflecting from materials on the surface. The brightness values are usually placed on a scale of 0 to 255 DN (digital numbers). If one brightness value is given to each pixel, the image is usually displayed with a grayscale ramp; a dark shadow would have a low brightness and a low DN, and a white cement road would have a high brightness and a high DN (Fig. 2.16a). Such images are displayed using the same stretched method that is applied to continuous thematic rasters. Images based on the 0 to 255 scale do not need to be sliced, but some images contain larger values and do require slicing. To display a color image, each pixel is given three DN values representing red, green, and blue light, thereby specifying a unique color combination, as discussed earlier. This display method is called an RGB composite (Fig. 2.16b).

30 of 48

Indexed color raster

An indexed color raster stores a single band of integers
Each value is associated with a specific R G B color combination (the colormap)
The method is simple, consistent, and saves storage space

Source: USGS

Access the text alternative for slide images.

30

The RGB color model is capable of storing millions of different colors and is best for fine rendering of images. However, it uses large amounts of memory and storage space. An alternate method for storing color identifies a restricted set of colors that appear in the image. It assigns an integer from 0 to 255 to each color and stores the RGB proportions needed to make it. This list of colors is called the colormap and is stored with the image. Each value in the raster is portrayed using its assigned color. Figure 2.18 shows part of a digital raster graphic (DRG). A DRG is a scanned US Geological Survey topographic map. The contours are represented by pixels with the value 4 and displayed using the assigned brown color. Every pixel value and defined color used in the map is shown in the accompanying colormap: green for vegetation, black for text, and so on.

31 of 48

Map design process

Map design is an iterative five step process: select, arrange, symbolize, review, and edit
Each step refers back to the objective for the map

Access the text alternative for slide images.

31

Symbolizing spatial data is a powerful analysis technique, but sooner or later the information will need to be shared with others as a formal map. Every map has a purpose established by the cartographer, and the success of a map is judged by how well the map fulfills that purpose. This chapter introduces some basic design principles for maps, but it cannot cover the full range of concepts included in the art and science of cartography. Taking a cartography class or reading map or graphic design texts is a good way to improve such skills. Map design can be envisioned as a six-step process (Fig. 3.1), but it is not linear. It begins by developing a vision of the map’s objective, and this vision must inform each step: selecting the data, arranging the page, symbolizing the information, reviewing the map, and editing to improve it. Each step must constantly refer back to the main objective ▸ Use graphic design principles to create effective maps ▸ Create text and labels on maps ▸ Create map layouts and printed maps to ensure that the final map supports the objective in every possible way

32 of 48

The map objective

Determining the objective is the first important design step

Four key questions must be answered

What is the purpose?
Who is the audience?
What is the medium?
Under what conditions will it be viewed?

32

33 of 48

Selecting the data layers

A good map tells a story

Who are the lead players?

volcanoes.

Which layers play a supporting role?

county population, highways?

Do some layers distract from the story?

rivers?

Source: Esri

Access the text alternative for slide images.

33

34 of 48

Cartographic generalization

Simplify a data set for better performance on a map
A cartographer might: exaggerate, displace, typify, reclassify, collapse, aggregate, simplify, or refine

Access the text alternative for slide images.

34

The scale of the map is an important consideration, and the data layers selected should be appropriate. Coarse-scale data should not be placed on large-scale maps. Occasionally large-scale data may be needed on a small-scale map, although often some type of simplification must be applied to individual layers. This process is called cartographic generalization. Eight different techniques are commonly employed (Fig. 3.2). With the exception of refinement, which can be performed using queries and selection layers, most techniques will involve creating a new simplified feature class to be used in place of the original.

Refine a feature class by omitting certain features, especially smaller or more detailed ones, such as removing higher-order streams and keeping the main trunk rivers.

Simplify lines or polygons by removing vertices or smoothing to give them less detailed shapes.

Aggregate many small features into larger ones, such as turning many close-lying small wetland polygons into a generalized marsh area.

Collapse features to simpler forms, such as turning detailed building footprints into simple rectangles.

Classify a detailed set of attributes into a simpler set, such as reducing geologic formations to a classification of igneous, metamorphic, or sedimentary rocks.

Typify features using representative rather than actual features. For example, replace the point markers for actual homes with a smaller number of “fake” points representing scattered houses.

Displace features slightly so they do not run over each other. For example, move a road to avoid the appearance of an intersection with another road.

Exaggerate the importance of features to give them greater prominence, such as enlarging the width of a water body to prevent it from appearing as a line.

35 of 48

Map grids

Graticule grids put longitude-latitude marks around the map frame
Measurement grids show x-y coordinates
Index grids show lettered and numbered squares

(a) Source: Esri; (b) Source: Black Hills National Forest; (c) Source: South Dakota Geological Survey

Access the text alternative for slide images.

35

36 of 48

Map elements

Source: Esri

Access the text alternative for slide images.

36

Map layouts traditionally contain certain map elements (sometimes called map surrounds) in addition to the map itself. A location map is used to show where the main map extends, especially when the main map is not easily recognizable to the readers. A legend explains the symbols used. A scale bar indicates distances, and a north arrow shows direction. A graticule grid shows latitude-longitude lines or tics. A neatline can be used to enclose the entire map or groups of elements within it. Titles and text explain what the map is about and cite the data used to create it. Images or graphic shapes can be added to support the story. Common practice asserts that all maps must have a title, legend, scale bar, and north arrow, but this rule is not necessarily true. Some maps are so simple and well symbolized that a legend is not needed.

37 of 48

Establishing a visual hierarchy

The order in which a reader perceives the elements of a map is affected by the cartographer’s choice of

balance, arrangement, negative space, symbols, …

Source: Esri

Access the text alternative for slide images.

37

38 of 48

Visual center

The visual center of a page is about 5% higher than the geometric center
Maps centered on the geometric center may feel a little “heavy”

38

39 of 48

Rule of thirds

The guideline is used to compose and crop photographs but also works for maps
Place important items at the intersections of lines dividing the page into thirds

39

40 of 48

Alignment

Place elements in ordered columns rather than a haphazard arrangement
Align edges of boxes and frames exactly using guides and snapping

40

41 of 48

Balancing elements

Balance elements to avoid crowded or open areas
Use negative space as a design tool to separate or combine elements more subtly than neatlines

Access the text alternative for slide images.

41

The treatment of negative space, or the blank areas on the page, is more important than it might at first appear. Large irregular blank areas make a map look unbalanced, yet too little space can make a design appear crowded. Negative space can actually be used as another design element that can separate or group elements more subtly than neatlines. Maps often look better when the designer resists the temptation to put a box around every element. Give each design a swift glance. Which one seems the most balanced?

Figure 3.5c is the best arrangement so far. In this case, the placeholders for the elements have been reshaped to echo the state’s shape, which leaves uninterrupted negative space around California and draws attention to it. The north and south sections of the state fall roughly in the target areas advocated by the rule of thirds and encourage the eye to scan the state left to right. The exterior edges of all elements fall on the boundary of the data frame, which subtly encloses the map and allows the neatline to be removed.

42 of 48

Foreground and background 1

Dark colors and high contrast bring symbols to the foreground and focus attention on them
Vary symbols to highlight the important and de-emphasize the less important through size, color, style, or contrast

Access the text alternative for slide images.

42

43 of 48

Map Design: Contrast

The greater the difference in value between an object and its background, the greater the contrast.
Keep the background light and use lots of contrast for important features!

43

44 of 48

Bad Map: Not Enough Contrast

Contrast is needed to distinguish features

From GIS TUTORIAL 1 - Basic Workbook, by Gorr and Kurland

45 of 48

Good Map: Better Contrast

From GIS TUTORIAL 1 - Basic Workbook, by Gorr and Kurland

46 of 48

Bad and Good Maps

From GIS TUTORIAL 1 - Basic Workbook, by Gorr and Kurland

47 of 48

Symbols and patterns

The brain doesn’t just observe, it seeks patterns

Experiment with different ways to portray the relationships between features

Color is more effective than shape to highlight categories.
Changes in size usually simply changes in quantity.

Access the text alternative for slide images.

47

48 of 48

Foreground and background 2

Choose bold symbols that bring important features to the foreground (countries)
Use lighter, lower-contrast symbols to put less important features in the background (ocean, graticules)

Source: Esri

Access the text alternative for slide images.

48