Cartographic Design Data type and symbolization spatial types - point, line, polygon value types - nominal, ordinal, interval, ratio visual variables -- using color, value, shape, size, orientation, and texture to create symbols Cartographic Design principles map scale legibility text size density complexity contrast Using labels and fonts
Cartographic Design Symbolizing Thematic Data Normalization Classification schemes
Cartographic design is the process of using symbols to take some part of the real world and simplify it to make it more comprehensible.
Data Types Point Line Polygon All data is symbolized on a map by point, line, or polygon symbols. Its usually obvious which type of symbol you will use to portray a given object, but this depends on the scale of the map and on how complicated your message is. On this sample from the Montana Highway Map, the forests, lakes, wilderness, and Helena are represented by polygon symbols. The counties have their boundary lines and their names shown, but are not symbolized as polygons. The city of Helena has both a polygon and a point symbol, while the towns are only shown as points. At a large enough scale, everything is really a polygon, such as the streets in the neighborhood base map on the previous slide.
Value Types Nominal Ordinal Interval Ratio Data without values It's possible to contrive an example map where you don't have to worry about the value of the data elements on your map -- all lines could be roads, all points could be towns, and all polygons could be lakes. But in most cases you need to symbolize different things in different ways. In this example, it is difficult to tell which lines are highways and which are streams if they are given the the same symbol. The symbolization method depends on what types of values distinguish your data. The four types of values are nominal, ordinal, interval, and ratio. Data with values
Data values do not represent any quantity or ordered relationship. Nominal Values Data values do not represent any quantity or ordered relationship. With nominal data, different objects have different names, and there is no connotation that one name has a greater or less value than another. You probably won't want to use symbols that imply that Blaine is more important that Broadwater, airport is more important than church, or road is more important than river.
Ordinal Values Data values represent an ordered relationship, but there is no numerical information. Ordinal data does imply that one value is larger or more important than another, but contains no information about how much larger. A paved road might be considered to be better than a gravel road and a highway is better than a paved road, but you do not imply that the difference between Gravel and Highway is twice as much as the difference between Gravel and Paved. Even if you assign numbers to the values, you aren't trying to say that a class 2 river is twice as big as a class 1 river or that the difference between class 1 and class 2 is the same as the difference between class 2 and class 3.
Interval and Ratio Values Interval data are numbers measured on a scale that does not have an absolute zero value and cannot be compared as ratios. Ratio data are numbers that are based on a meaningful zero. Both of these are difficult to map if you do not generalize them back to ordinal or nominal groups. Interval Data – You may add 10 or 20 degrees Celsius to a temperature, but 30 degrees is not three times as warm as 10 degrees. Interval data are numbers that you can meaningfully add and subtract, but which cannot be compared by multiplying and dividing. The change from 10 degrees to 20 degrees is the same as from 20 degrees to 30 degrees, but 30 degrees is not three times as warm as 10 degrees. This type of data is not common, but you can recognize it by thinking about whether the value of 0 is arbitrary - does it really mean that there isn't anything? Ratio data is numeric data where it is meaningful to speak of multiples. A population of 40 really is twice as much as a population of 20.
Interval and Ratio Values It is very difficult to scale map symbols to portray interval and ratio data in an easily readable fashion. Scaling point sizes to the data value results in a 24-point dot that is 16 times larger than the 6-point dot. But scaling the area of the dots to the values results in sizes that are difficult to distinguish, even though the “24” dot really covers 4 times the area of the “6” dot. The third scale shows a compromise, which is difficult to come up with. These are sample legends for a map where each symbol would be a different size depending on the data. A point with a value of 7 would be slightly bigger than a point with a value of 6. But it is almost inconceivable to me that someone would want to sit down to a map and determine data values by measuring symbols with a ruler. It is much easier to generalize the data values into ordinal or nominal classes and select symbols that are clearly distinguishable from each other.
Visual Variables Hue (color) Value (darkness) Shape Size Orientation Spacing (texture) These are the properties of map symbols you can change to make them different from each other.
Hue and Value In the top legend, the only variable changed between the symbols is hue. In the bottom legend, the items all have the same hue, but their values are different. Hue can be used for nominal values, but there are some groups of hues that can imply an ordinal or numeric progression, such as the yellow-orange-red sequence at top right. Differing values of the same color definitely imply a progression. Using different values for different counties, as shown here, is probably a bad idea. There is lots of symbolization with hue and value that is probably done automatically without much thought. Rivers and lakes are blue. Urban land use is red, forests are green, and evergreen forests are darker than deciduous. Red commands more attention than yellow and may imply a higher value. Dark colors imply a higher value than light ones.
Shape and Size You almost never see shape used in polygon or line symbols. Size isn’t used very often in polygon symbols. Shape is a good visual variable to use with nominal values, as it does not imply any magnitude, and you can frequently find shapes that people easily associate with different names. Size is good for ordinal or numeric values.
Orientation and Spacing Orientation is very rarely used with line symbols and is not common with points. Spacing is rarely seen as a visual variable in point symbols. Orientation is a good variable to use to indicate nominal values. It doesn’t imply any ordinal ranking. Spacing is sometimes scaled in polygon symbols to be proportional to interval or ratio values, so that you could estimate the value of a polygon’s attribute by measuring the space between the lines. Once again, it was probably a bad idea here to use spacing for county names, because this legend implies that Carter has more of something than Custer.
Map Scale Your map scale will usually be dictated to you by the region you need to show and the page size you are allowed to use. If you need to show a large area in great detail, you may be forced to split your map among several pages. Sometimes you need to fight to get the client to accept that a certain size map cannot show everything they want to see – they must accept a simpler map, a larger map, or a multi-page map. If you need to make a map where there are an exact number of miles per inch, remember that 1 mile = 63,360 inches. So a map with 10 inches per mile is 1:633,360 scale, and a map with 65 inches per mile is 1:4,118,400, which may be a good scale to show Montana on an 8.5x11 inch page. But it can be a bad idea to put a statement like “1 inch = 65 miles” on a map, because someone might enlarge or reduce the map later and not remove the statement. A scale bar doesn’t have this problem.
Legibility Text size Density Complexity Contrast Emphasis Visual Balance
Density/Text Size Legibility is partly determined by the purpose of the map. As a PowerPoint slide or as a huge map pasted to a billboard, this map’s text is too small, some of the data sets shown are too dense, it is too complex because it shows too much data, and no single data set is emphasized. But this is a reduced image of a 60x36-inch reference map. It works fine if you can walk up to it and study it.
Density/Text Size This map is much less complex than the previous one and would probably work for an 8.5x11 inch page, but there is still information here that you can’t see or don’t have time to see in a slide show.
Density/Text Size I think this map works as a slide. I removed all of the small ownership classes that would be barely noticeable, combined all of the different State-owned land types into one, and removed all extraneous information except county lines and major streams.
Complexity Complexity can change depending on the data you want to show. These two maps show the same data, but distributed differently among the counties. A simple distribution lets you use more symbols. In a complex distribution, you can’t see the difference between similar values. On the top map, you can see the difference between almost any two counties, since the similar colors are next to each other. In the lower map, it is very difficult to tell the difference between Powder River, Prairie, Powell, Richland, and Ravalli counties, even though they are all different colors.
Contrast – Color/Value The all-blue map is very difficult to read at a glance, but might be the best choice if you have something else to display on top of it, such as a weather map. The brown-blue map gives a great picture of the shape of the European continent, while the bottom map makes it easier to see the shapes of nations and their relative positions.
Contrast – Line Weight Use a wide variety of line weights to make a map that is both more attractive and easier to read.
Contrast – Subject Emphasis Emphasize your subject by giving it more contrast. In the upper map, you see a round red thing sitting above a chair-shaped blue thing. In the lower map, you see a mostly red county above two blue counties and next to a county that is half blue.
Contrast – Symbol Discrimination If the map reader needs to be able to tell symbols apart from each other, use several of the visual variables rather than just one to make them different. The dots on the left are different colors in additon to being different sizes.
Contrast – Symbol Discrimination For this map, I used 3 visual variables, hue, size, and shape, to make it easy to tell the point symbols apart from each other.
Visual Balance Spread out the stuff around your map to avoid awkward white space and designs that look unstable. The visual center of your page is slightly above the physical center. Avoid putting the most important stuff below the middle of the page. A map like this shouldn’t have a north arrow. North is different directions on different parts of the map.
Labels Water features are blue, italic Never run a label upside-down The biggest thing to avoid is having labels that are upside-down, like part of the Poplar River is here. Most of the river labels were automatically placed by ArcMap. But I took care to fix the West Fork Poplar River so that the label follows its entire course without having the words fall on top of the tributary stream or the highway. I also placed the Missouri River by hand. Notice how much better it looks if you increase the spacing between the letters and between the words.
Labels Avoid angled, straight line text Don’t use too many fonts Use text size, boldness, and italics to distinguish between different features before resorting to font changes. All the text on the right map use the same font. The left map uses three different fonts, which tends to create a more cluttered appearance. If you want text to run across a polygon at an angle, it should be curved.
Labels Use a text mask to cover up lines that can’t be avoided. Use upper- and lower-case text rather than all capitals. In the left map, the county lines and highways are useful for locational reference, but are not the subject of the map. The town names use a mask to give white space between them and the background features, making them legible. The upper map on the right uses 12-point text. The lower one uses 10-point text, but the labels take up more space due to being in all caps.
Normalized Data Think of other variables that influence the data you want to show. Population – some counties have a small population because they cover a small area. Normalize data by dividing the quantity you want to show by the quantity that has an influence on it. Population density is population divided by area. Notice how some large counties got lighter and some small ones got darker. Population Density -- Shows which areas really have a larger concentration of people
Normalized Data The data on this slide are NOT normalized. The distributions all look the same because they all depend on the population. Population Number of people less than 5 years old Number disabled people
Normalized Data The data on this slide are normalized. They look very different than the same data from the previous slide. Population Density Percent of population less than 5 years old Percent of population disabled
Classification Schemes Natural Breaks Quantile – each class has same number of members These maps are of the same data – percent of people over 5 years old. This is a difficult data set to map because there is one county (Powell) that is much different than the rest. The natural breaks method of setting classes (the default in ArcMap) is good for this data set because it looks for areas in the distribution where there are few values to insert the breaks. The quantile method groups Powell County with others that it is not really similar to.
Classification Schemes Standard Deviation Equal Interval These maps are of the same data – percent of people over 5 years old. This is a difficult data set to map because there is one county (Powell) that is much different than the rest. In the equal interval method, the data range is divided into equal intervals. In this data set, there are no values in the second interval, only one in the third, a few in the fourth, and almost the whole state lands in the fifth class. The standard deviation method uses an equal interval where possible, but has a rule saying that all classes must have some members. The mean value of all the data is taken as the center of one of the classes, and the width of each class equals the standard deviation of the data. In the under 5 data shown here, the mean is in the middle of the fourth class. The fifth class is narrower because there are no values high enough to go an entire standard deviation above the fourth class. The second and third classes are the same width as the fourth. The next lower class is much wider, because that is how wide it had to be in order to get one of the counties to join it.
Grid North vs True North North isn’t always straight up on a map, but you can make it so. Most projections result in converging meridians, with north being different directions on different parts of the map. If you are interested in a small area away from the center of a the projection, you will end up with north being tilted by about the same amount all across your map. Many GIS and mapping packages let you show your data at an angle, so you can make north go straight up and down in the center of your map.
North is Which Direction?
North is Which Direction?
North is Which Direction?
Python Script for True North # Rotate Data Frame so that True North is straight up. import arcpy, os, math # Find the current map view and get its extent mxd = arcpy.mapping.MapDocument("current") df = mxd.activeDataFrame # Figure the latitude and longitude of the top center of the map mapRef = df.spatialReference geoRef = arcpy.SpatialReference(4269) ptTopY = df.extent.YMax ptTopX = (df.extent.XMin + df.extent.XMax)/2 ptTop = arcpy.PointGeometry(arcpy.Point(ptTopX, ptTopY), mapRef) # Project to geographic coordinates ptTop = ptTop.projectAs(geoRef) # Find the projected map coordinate 0.5 degrees of latitude south of the top center ptBtmX = ptTop.centroid.X ptBtmY = ptTop.centroid.Y - 0.5 ptBtm = arcpy.PointGeometry(arcpy.Point(ptBtmX, ptBtmY), geoRef) # Project back to map coordinates ptTop = ptTop.projectAs(mapRef) ptBtm = ptBtm.projectAs(mapRef) # Figure the angle between the top of the map and the point 0.5 degrees south of it dX = ptBtm.centroid.X - ptTop.centroid.X dY = ptBtm.centroid.Y - ptTop.centroid.Y rAngle = math.atan(dX / dY) * 180 / 3.1415926 print rAngle # Rotate the map to make north straight up and down df.rotation = rAngle # Refresh arcpy.RefreshActiveView()
North is Which Direction?
Map Projections When you look at maps with an application like Google Earth and can see the world as a globe, it doesn’t bother you that a place starts to look strange when you spin it away from the center of the map. Your brain interprets the two-dimensional image you are seeing as a three-dimensional object, and it knows that, on the right, Alaska is farther away from us and seen at an angle, and that it should look smaller and distorted. If you think about it, the image of Alaska on the left is also distorted. The middle of the state is closer to us than the Aleutian Islands or the panhandle, so the middle is bulging out towards us slightly. No matter how close you zoom in on one of these globes, the effect still exists, but it becomes more and more difficult to observer or measure. As soon as you stop seeing a globe in front of you, you forget that you are looking at a three dimensional object.
Map Projections
Web Mercator Projection
Web Mercator Projection
Web Mercator Projection Web Mercator – North end of Montana looks pulled apart to me. State Plane – Meridians correctly converge towards the North Pole