I want to share with you the translation of the thematic mapping guide from the guys from axismaps .



I recommend reading to information designers, journalists (data), analysts, novice cartographers, as well as anyone who wants to learn how to read thematic maps and distinguish a good map from a bad one, which misleads the reader. I invite everyone who is interested.

What is thematic maps?

Navigation maps and thematic maps

Most cards fall into one of two broad categories:

')

General geographical (physical) maps display various locations and objects, such as countries, cities, rivers, etc ... These are the maps that you probably use in everyday life, for example, Google maps that help you find where things are. Other common physical maps emphasize topography - these are topographic maps.



Thematic maps do not just show locations, but show various attributes or statistics corresponding to a given area, various spatial patterns and interrelations between locations. For example, if a physical map shows the location of a city, then a thematic map can also show the population of this city. In one case, we map the terrain, in the other data. This guide focuses on thematic maps, their various types and basic creation principles.

Data presentation

Generally speaking, thematic cartography is about a different representation of spatial data. This is done using several visual variables , such as size, color and shape. The specific method depends on the nature of the data presented (whether it is counting data or nominal), as well as on the type of geometry (point or area object). In this guide, we will look at some common ways to create thematic maps, coupled with tips and best practices.

Types of measurements: nominal, ordinal and numerical data

Know your details

The success of many thematic maps due to the correct choice of the method of data presentation. In other words, not all geographic data are the same, so they should be mapped in different ways. For example, an areal cartogram works well for such things as population size or life expectancy (which are numbers), but it is not suitable for nominal data, especially if categories cannot be streamlined. Such data as the predominant religion or soil type is not inherently can be measured quantitatively. For areal cartograms, numbers are needed to scale the region accordingly, without them in any way. The same is true for graded character maps, choroplets, and point density cartograms.



The following is a brief description of the measurement types. In more detail, each type of card will be discussed in the appropriate section below.

Types of measurements

Numeric data is the most common data type of thematic maps. Everything that can be calculated (people, barrels of oil) or measured (temperature, income) is great for thematic maps. Do not forget about the importance of data normalization; this affects which card types you can (or cannot) use for visualization.



Nominal data (also known as categorical or qualitative data) is in no way associated with numbers, and in principle cannot be ordered (ranked) in any way.



Ordinal data is essentially categorical data that can be ordered. For example, T-shirt sizes (s / m / l / xl), risk of flooding (low risk / medium risk / high risk) or age group (young people / mature age / elderly). For mapping ordinal data, a choropleth with a consistent color scale, or a graded character map with a number of classes equal to the number of categories in the data, is best suited.

Data normalization

Normalize or not?

A very important aspect of thematic cartography is the way in which you use the data, in its pure form (for example, the population of each country) or in a normalized form (for example, the population of a country per square kilometer of its territory). In the first case, we will see how many people live in the country, in the second we will see how densely populated the territory is. The main reason for data normalization is the ability to compare very different territories. For example, you can directly compare such a large country as Canada with such a small one as Switzerland. And although much more people live in Canada than in Switzerland, the population density there is much less. Without data normalization, seeing this fact is not so easy.



Note: If you are going to create a choropleth, use ONLY normalized data.



To summarize: If you want your users to see magnitude (quantitative order of magnitude), use the data in its pure form. If you want to show the relative difference (which already takes into account such things as the size of the territory), then use the normalized data.

Are my data already normalized?

It may very well be! If the numerical data contains in the description " x per square kilometer / mile / ... ", or " x per capita ", or " percent ", or " x / y ratio ", then you can skip the data normalization step.

How to normalize your data

The main ways to normalize data are as follows: divide the data by (1) the area corresponding to this data , thus creating data of the form " x per square kilometer / mile "; (2) on the number of people within this territory , creating data of the form " x per capita " or " x as% of the total population ".

Basics of data classification

When to use

If you are going to classify your data, you should determine both the number of classes and the method of splitting into intervals (classes). There are many different ways to systematically classify data ; below we look at their advantages and disadvantages.



The meaning of classification is to reduce a large number of observations by grouping them to several intervals or classes . What for? Because it is much easier for users to perceive several clearly defined classes than “raw” data. If the classification is done correctly, then it helps much easier and clearer to convey the message embedded in the map. However, the classification process is not easy, it is very often to choose the "ideal" method for a specific set of data is not possible the first time. It is always important to understand the data with which you work, and not just to apply the “favorite” classification method. An unsuitable classification method can create false patterns on a map that have little to do with the actual geographic phenomenon that you are trying to visualize. Maps using questionable classification methods are not just ineffective, they are misleading.



The classification is important because grouping data is one of the most fundamental aspects of map generalization — the process of simplifying the real world to the frame of a map canvas. Therefore, even small differences in this process can drastically change the appearance of the card and its sending . Despite all this, users rarely attach importance to this and do not question the classes they offer, but this is one of the simplest ways to " cheat with cards " intentionally or out of ignorance. However, the classification is very useful and is a basic skill when creating thematic maps.





This map uses a scheme with 5 classes with equal intervals (1-10, 11-20, ...).



Note: All of the above is also true for choroplets, graded character maps, and cartograms, since you can create versions of these maps that are divided into classes.

Purpose of data classification

Generally speaking, the main purpose of the classification is to combine similar observations and share significantly different ones. From a mathematical point of view, the goal is to find the optimal number of classes and determine their boundaries in such a way as to minimize variation within the classes and maximize the differences between the classes. For example, there is a data set of 4 observations 1.3, 1.6, 3.5 and 3.9, it is logical to divide it into two groups: 1.3 and 1.6 into the first group and 3.5, 3.9 into the second, because there is an obvious numerical gap between them. Such a campaign is very common and is called "maximum breaks".



However, things are not so simple, and maximizing the difference between groups is not always appropriate. Suppose that in the example above, the value 1.5 is critical , and it is important to differentiate the values ​​relative to this critical point. For example, if the location corresponds to a value below 1.5, then emergency financial assistance is available. In this case, external constraints outweigh the arguments that are logical from the point of view of mathematics, and, although 1.3 and 1.6 are close in meaning, they will be assigned to different classes.

Number of classes

If you are not sure, then create a map with 3–7 data classes . Of course, your goals and the data themselves should influence decision making, for example, the US political map usually has only 2 classes (the well-known maps of red-blue states). Maps showing deviations from the average will also have only 2 classes (below average and above average).



The more classes you use, the more details will be visible on the map (which is good), but this will increase the difficulty of perception of the map and, as a result, the risk of incorrect interpretation of data, since more colors are more difficult to distinguish (and even harder to print such card). The key question is how many details do you want to show? A map with 3 classes / colors will be very easy to read, but it can hide some important aspects of the data from the reader, and at the same time it can create artificial geographic patterns due to the fact that different territories are combined. The only true number of classes for the map does not exist, so experiment.



Not sure how many classes to use? Look at the distribution of your data in the histogram: are there any obvious clusters within your data, are there any large gaps that form natural groups? If so, select the number of classes accordingly.

Classification method

Just as there is no only right amount of classes, there is no only right way to split the data into intervals. Look at the histogram (or scatter diagram) to determine the “shape” of your data. Try to determine values ​​with similar frequencies in one class, and values ​​with strongly differing frequencies should be separated by different classes.





The shape of these histograms can be assumed that 3 or 4 classes would be a good choice.

In the absence of other conclusions, natural “drops / breaks” are a good basis for the formation of intervals.

EQUAL INTERVALS break up data into classes of equal size (for example, 0-10, 10-20, 20-30, etc.) and work best on evenly distributed data. ATTENTION: Avoid using the split into equal intervals, if the histogram shows a clear skew (asymmetry), or there are large outliers. Emissions will produce empty classes, and distortions will lead to a large variation within the classes. Since there are no obvious outliers in hotel data, the use of equal intervals is permissible here.



Quantiles will help create a map with an equal number of observations in each class: if you have 30 regions and 6 data classes, then there will be 5 regions in each class. The lack of quantiles is that they can lead to very different intervals for different classes (for example, 1-4, 4-9, 9-250 ... the last class is huge). Emissions can also divide areas with very close frequencies and lead to the merging of areas with different frequencies, which is highly undesirable, so always look at the splitting on the histogram. ATTENTION: In the example with the data on hotels, the use of quantiles leads to the fact that part of the third cluster falls into the second class, although much closer to observations from the third class.



NATURAL Gaps are in some sense an “optimal” solution because they initially minimize the variation within the classes and maximize the differences between them. One of the drawbacks of this method is that each data set is unique and, accordingly, the partitioning too. This makes it impossible to compare similar maps of different data sets, for example, in map atlases or series of maps showing the dynamics over time. In such cases, it is better to use a different breakdown scheme.



MANUALLY have to set the boundaries of classes in many cases. The reasons may be different: it is necessary to take into account the critical point in the data, make one of the boundaries an average value, make the map a part of a series / atlas (so that the continuity of colors and ranges in the series is preserved). If the splitting of other methods can be improved with minor edits, then do not be afraid to correct them manually.

Signatures and Hierarchy in Cartography

Preliminary knowledge

It is reasonable to assume that signatures and text on maps are used to name objects and locations, but their role is much more important. Signatures not only indicate the location of objects, but also display their type and shape, the connections between them, symbolize the data associated with them. Text in the form of short blocks and descriptions is used to indicate important map elements, such as headings, data sources, projection type, scale, legend, and map purpose. The predominance of signatures on the card (especially physical) leads to competition for attention with the rest of the map symbols, while the signatures and the text greatly influence the overall impression created by the card. They help to deeper the reader into the topic being studied and to facilitate the understanding of the area, like no other graphical tools (for example, color).



Task 1) selection of fonts and styles (outlines) and task 2) placement (positioning) of signatures and text on the card is called “card typography”. Both of them influence how well the card works and is perceived as a whole. Due to the iterative nature of the process (moving a signature, or changing the font often leads to a domino effect and a cascading change in the typography of the map), typography has always been one of the most labor-intensive processes in cartography, and remains so to this day, despite modern advances in automating this process.

What is important to think about before typing a card

1) What is the semantic hierarchy of the objects that I want to sign? The sense hierarchy allows you to rank the map objects in order of importance. For example, on some maps of the capital may be more important than other large cities, which, in turn, may be more important than small cities. Regions and countries can be higher than any of the cities in the hierarchy. Such a conceptual ranking method will help to further create a visual hierarchy of map signatures.



2) Why do I need a visual hierarchy of signatures and text maps? The visual hierarchy is the most important aspect of the card design, it helps the reader to organize graphic information in such a way that it is the most accessible and understandable. When used properly, the visual hierarchy allows the reader to easily and quickly perform basic tasks such as categorizing, grouping, searching and scanning information. On the first map, below is an example of a visual hierarchy of signatures. Without a visual hierarchy, as in the second example, reading the map becomes very difficult, since all signatures are equivalent in importance.









All other things being equal, larger font size and a bold face increase the level of the visual hierarchy of the signature. Capitalization and the use of "heavy" colors, such as black, red or pink, have the same effect. Reducing the font size and letter spacing (tracking) lowers the signature level, and muted colors, like gray, do the same. Of course, the task becomes more difficult as the number of signatures and their styles on the map increases, so creating a good visual hierarchy will require you to repeatedly revise decisions, a series of experiments and gradual improvements.



3) What are the main conventions for printing cards? Convention is a good starting point, but should not be taken as rules that cannot be violated. Here are some of the most common card printing conventions:

• The priority of placing the signature of a point object: 1) above and to the right, then 2) below and to the right, then 3) above and to the left, then 4) below and to the left. Placing directly on top, bottom or sideways is undesirable.
• Visually align in the center and increase the letter spacing between the signatures of polygon objects to indicate their size and shape.
• Use uppercase letters for area object signatures.
• Separate cultural and physical features with sans serif (serif) and serif (serif) font families.
• Signed water objects in blue italics.
• Make the difference between the signatures of different levels in at least 2 points.
• Do not turn signatures upside down.
• Signatures should not be less than 6-7pts for paper cards and 9-10pts for digital cards.
• If necessary, use one serif font and one without, but do not use more than one sans serif font on the card.

4) How to create a suitable overall impression of the card and the corresponding mood? Knowing your audience and the main goal of the card play a key role in shaping the right “feel” of the card. If the card is dedicated to a narrow theme, you need to reflect this in its style. All fonts have their own “character”, which can create a subjective perception of the reader. Do not use very catchy (intricate) fonts, while choosing a font that emphasizes the purpose and theme of the card. Font selection can give the map a formal look, informal, historical, modern, etc.



5) In what form and under what circumstances will the card be used? Reading conditions are a very important factor in determining how readable and successful a card will be. It is important how the map will be played (printer, large screen, projector, mobile device), from what distance the light will be viewed. In the case of low resolution, poor playback quality, poor lighting and long reading distances, it is necessary to use clearer and more contrasting fonts with good readability.



6) How many signatures should be on the card? There is no simple answer to this question.Generally speaking, there should be as many signatures as necessary to maintain the goal of the map and provide contextual information. Physical maps usually contain a wide range of densely arranged signatures. Thematic maps, on the contrary, usually contain not so many signatures, because they use different graphic symbols, colors and explanations to convey the main message. Also pay attention to the distribution of signatures on the map, you should avoid too empty areas and too "dense", it can give the map an unbalanced look. After all, adding signatures to a card is simple, it is much more difficult to determine when to stop.



7) Do I need to know anything about typography? Yes. Knowledge of fonts, their metrics and components can greatly help to make the right choice of fonts for your card.

. , , . , , . .







(x-height) ( ). , . , , . , «».



, 'T' . , . . , , , «» ( ), Georgia, Droid Serif Bitstream Vera Serif.







. , , . , , , . , Helvetica Neue, .



. . ! , , , , . , , Google Fonts .

Below are some of the fonts for a quick start. Some of them are better for web-card interfaces, others are better for signatures, some are suitable for both tasks. Pay attention to narrow fonts (condensed and narrow), they can be useful for signatures. Also, some types of fonts come with or without serifs, it can be convenient to create a more unified map view. Most of them have at least four basic styles (regular, bold, italic, bold italic), some more. Most are free on Google Fonts , but some are on Typekit or paid.

Sans serif:

• Open Sans and Open Sans Condensed (Google Fonts)
• Source Sans Pro (Google Fonts)
• PT Sans PT Sans Narrow (Google Fonts)
• Myriad Pro Myriad Web Pro (, Typekit)
• Ubuntu Ubuntu Condensed (Google Fonts)
• Arial Arial Narrow (, )
• Proxima Nova Proxima Nova Condensed (, Typekit)
• Oswald (Google Fonts)
• Merriweather Sans (Google Fonts)
• Fira Sans (Google Fonts)

:

• Source Serif Pro (Google Fonts)
• PT Serif (Google Fonts)
• Merriweather (Google Fonts)
• Aver ()
• Lora (Google Fonts)
• Latienne Pro (Paid, Typekit)
• IM Fell (Google Fonts)

Monospace:

• Fira Mono (Google Fonts)

Symbols:

• Font Awesome (Free)

Use of color on thematic maps

Three color measurement control

The success of many thematic maps, such as horoplets, depends on understanding how to use color to encode geodata using three dimensions of the HSL model: hue, saturation, and lightness. Unlike choosing the color of your living room, the choice of color in cartography is far less subjective than most people assume, cartographers do not make the choice on the “looks cute” principle. On the contrary, there are important rules that determine how color schemes work, and how colors should relate to data. In addition, there are a number of restrictions imposed by the characteristics of human perception, for example, a violation of color perception (about 8% of men). These restrictions must be considered when choosing colors, and not only rely on personal opinion. The good news is that these limitations are quite well studied, and there are ready-made recommendations for their accounting and leveling.

The nature of your data determines the choice of color scheme.

Nominal data are essentially unordered categories, and they should be rendered using a nominal color scheme . If you have categories or numerical data to be ordered, then you need a continuous color scheme . Continuous color schemes can be monochromatic or multi-ton, but they should be ordered by the difference in lightness and saturation. Divergent patterns should be used if there is a natural center point in the data, such as zero (separation into positive and negative values), or if it is necessary to compare values ​​relative to some critical point, for example, the national average (data for regions will be higher or below the average). There is a great tool for selecting the color scheme:ColorBrewer , there you can find more detailed information on the characteristics of color schemes.

Schemes taking into account features of perception

The schemes on ColorBrewer take into account the peculiarities of perception ( pdf ) so that the color changes at each step of the scale look consistent for our vision. This needs to be done because human vision perceives the same change in different tones differently. Because of this, creating a good color scheme becomes a very difficult task. You do not have to be limited to options for ColorBrewer schemes , but there you will always find proven, reliable schemes for your project.

One-dimensional maps and multidimensional

One data topic or several?

If you are going to make a thematic map, then you will work with geographic data that represents a certain set of thematic attributes with geolocation. "Thematic attributes" can be any data associated with a specific location / location. For example, life expectancy , political preferences , land use , crime rate , real estate prices , foci of diseases and so on.



If your data has only one thematic level (theme), then of course you will visualize only one attribute.. If the data covers several topics, you can choose between a one-dimensional map (one attribute) and a multidimensional thematic map , that is, a map on which several attributes encoded by a hybrid symbol are visualized at once . Such multidimensional thematic maps encode multiple geographic facts about each location using complex composite characters. Multidimensional maps are not always better than one-dimensional, their disadvantages and advantages will be discussed below.



Most subject maps show one attribute, such as per capita income.. Very often, a simple one-dimensional map is all that is needed, since there is only one attribute to render. However, the most interesting and informative maps intentionally combine several data sets. This allows readers to compare different data directly and often helps to identify important dependencies. For example, a two-dimensional map showing per capita income and life expectancy can reveal a strong correlation between these two factors. Multidimensional maps also help save space because we can fit more data onto one map, rather than spreading them across a series of one-dimensional maps. When multidimensional maps are done well, they show much more than the sum of the data layers, they tell a complex spatial story.

Important points and limitations

Before making a choice in favor of multidimensional maps, it is important to know that creating the design of such maps is quite difficult. They can be easily overwhelmed with a multitude of overlapping symbols and colors, each of which fights for a place on the map. You should also consider your audience and how much time they are likely to spend exploring your map. In the example below, a two-dimensional choropleth uses a rather sophisticated color scheme, which requires frequent reference to the legend in order to correctly evaluate the map. Simple one-dimensional maps where more accessible to a wide audience in this sense.







Despite the risk of oversimplification, one-dimensional thematic maps are easy to read and quickly reach their goal. Multi-dimensional maps are richer, but require more effort to understand.

Making a good multidimensional map

: , , . , , . « : , ».



: , . ( ), . , , , , .



: , () . , «FIPS» «ID» . , , , . : « ».

We display the Earth on the plane

Expand the Earth on a plane without distortion of one or another species is impossible. Imagine an orange peel: if you try to lay it on a plane, then you will have to stretch, squeeze and tear it. A similar situation with the Earth — if we want to create a flat map, then distortion is inevitable. The good news is that cartographic projections allow you to do this systematically. In other words, we will know exactly how exactly any place on the map is stretched / compressed (at any point). There are many projections, and each has its own distortion model, because there is more than one way to turn the orange peel on a plane. Some projections can save individual properties of the Earth without distortion, but to save everything without distortion is impossible.

Projection properties

We usually speak of projections in terms of how they distort or retain certain properties of the Earth, which we call projection properties . There are four main properties:

Area - some projections distort areas (for example, the Mercator projection)

Pay attention to Greenland, on the Mercator projection it is almost the size of South America. In fact, South America is 8 times the size of Greenland. The Mercator projection does not preserve the area, and the closer to the poles, the greater the distortion. On the other hand, there are projections that do not distort the area, for example, Equal cylindrical projection.







Notice, here Greenland is the right size compared to South America. The projections that preserve the area are called equal . A cartographic projection either saves squares everywhere or distorts them everywhere. This property is all or nothing.

Form - some projections distort the shape of objects (for example, Azimuthal projection)

On the projection above, look at Australia (it is on the right), it is impossible to recognize it, and New Zealand is stretched in an arc along the left edge of the map. This projection does not preserve the form or shape of locations, it either stretches them, or twists, or flattens. Compare it with the Lambert Conformal Conic Projection (below), which preserves the common shapes of the continents.







Projections like this are called equilateral , or conformal , they preserve local angles. In practice, this means that things on the map will be more like themselves. The example below shows how Greenland looks on three equilateral projections (top row) and on three non-equiangular projections (bottom row).







Please note that Greenland on all equal-angle projections looks Greenland. The shape changes a little, some parts of the island change size, but on the whole the shape remains. Like a rectangle and a square, they have a similar shape, although they are different shapes, but a square and a circle are not.



As in the case of conservation areas, the preservation of forms is carried out either everywhere or nowhere.

Distance - most projections distort lengths (for example, Equivalent projection)

The distance from Madison to Buenos Aires is much greater than from Madison to Madrid. But in the Equivalent projection, the lengths of these segments are equal, because it does not maintain the distance. But on the azimuthal projection of the distance displayed in the correct proportion.







With the preservation of distances there is one feature. We have projections that can save areas and shapes everywhere on the map, but there is not a single projection that maintains distances everywhere. There are only projections that maintain distances relative to one or two points on the map. Distances from the center and to the center of the Azimuthal projection are displayed correctly, and between any other points with distortion. When a projection saves a distance, we call it equidistant , or equidistant .



Area, distance and shape are mutually exclusive properties of the projection; if the projection preserves one property, then the remaining two will be distorted.

Direction - sometimes a straight line is not the shortest path!

New York and Istanbul are almost at the same latitude, at about 41ºN. This means that if you are heading from New York exactly to the east, you will reach Istanbul. But this does not mean that this is the shortest path between the two cities.







In this image, one of the lines is the direct, easiest way between New York and Istanbul, you can simply head east and fly. But if you prefer to travel the shortest path, then you should choose a curved line above. Since the surface of the Earth is curved, then the shortest paths around it are also curved. This may be a little strange at first glance, but everything becomes clearer when you try to make the route yourself. Find a globe, stick a pin into Istanbul and New York, then pull a thread between them. You will notice that the thread covers exactly the path that is shown by the arc on the map above. Such a curved shortest path is called the arc of the great circle , or orthodrome . And the path in the form of a straight line where you stick to one direction is called a rumba line, orRhumb Line .



Some projections, such as the Mercator, show loxodromes with straight lines. Loxodrome make air and sea navigation easier, as you only need to draw a straight line and follow in a given direction. Other projections show large circle arcs as straight lines, making it easy to determine the shortest path between two points. One of these projections is Stereographic Projection.







Now, on the contrary, the orthodrome is straight, and the loxodrome is curved. The lines are the same as those of Mercator, just stereographic projection has changed their appearance.



When the projection displays the arcs of the large circle as straight lines, we call it azimuthalprojection. Unfortunately, like equidistant projections, azimuthal work for only one point. On the Stereographic Above, the projection is centered on New York; therefore, only straight lines leaving or entering this point will be orthodroms, and there will be no straight line between Madrid and Casablanca.

Compromises - do not do it perfectly, do it well

If you go over the examples again, you may notice that the distortion usually increases as you approach the edges of the map. There is usually one area that looks normal and not too distorted, and then things get worse as you move away from that area. As an example, we take the Azimuthal projection considered earlier; it very strongly distorts the shape of Australia, while the British Isles look normal. The basic rule is: the larger the area covered by the map, the greater the distortion, especially at a distance from the center. This means that distortions need to be considered primarily on the maps of the world, and on the maps of the surroundings (city or district) they can be neglected.



To combat the strong distortions on the world maps, special projections have been developed. These special projections are a compromise; they distribute distortions throughout the map, reducing their extent to an acceptable level. One of these projections is the Robinson projection:







Compromise projections distribute distortion approximately evenly. This approach avoids the absurdly large distortions, so they are good for maps of the world. The downside of this is that we are deprived of special areas on the map, where there are practically no distortions. For this reason, trade-off projections should not be used for maps of continents, countries and anything less than the Earth. If you do not display the whole world, then there is no point in making areas with weak distortions worse, at the expense of areas with medium distortions (which are far from the edge).



Compromise projections do not preserve areas, shapes, or distances, but they display them fairly close to normal. They have a low level of distortion in general. If it is important to keep a specific property on your map, for example, an area, then a compromise projection will not work for you.

Choosing a projection

Since there are a lot of various projections, a reasonable question arises: which one should be used? As you might guess, the existence of such a large number of projections means that there is no "best" projection. Each has its own advantages and disadvantages and is better suited for a particular case. When choosing a suitable projection, answer the following questions.



Is there a need to keep one of the special properties? Remember that some projections retain without distortion of area, shape, distance or direction. Sometimes the theme of your map requires you to save one of the properties, here are some examples:

• Area - Dot density cartograms require equal projections. If you work with population density data per square kilometer, then it is necessary that each square kilometer look the same size. If the areas are distorted, then some areas will look thinner or denser than in reality.
• Form — Equal-angle projections are usually good for physical maps when we want to keep map locations recognizable and familiar. They are also often used for navigation maps. Preserving the local angles, they do not distort the path - a turn of 45º on Earth looks like a turn of 45º on the map, which is convenient.
• Distance - If you want to show visually how far one place is from another, then you need to use an equidistant projection. Sometimes airports use them to display reachable cities.
• Direction - Also useful for navigation. Sometimes it is useful to show loxodromes with straight lines: the Mercator projection, for example, was invented for Renaissance sailors, to simplify the construction of courses. They could just draw a straight line on the map and keep the desired direction on the compass. On the other hand, in air navigation it is better to show the arcs of a large circle with straight lines, this will allow you to fly the shortest way and save fuel.

There are many other reasons for the preservation of each of these properties, the examples given only indicate the direction of further reflection. What else needs to be considered:

• How large is the area covered by the map? If this is a world map (and there is no need to preserve specific properties), then it is better to use trade-off projections.
• Does your map reach from the north to the south or from the west to the east? Different projections have different distortion patterns. Some, like the equal-sized conic projection of Albers, distort more as they move to the poles, and much less as they move west or east. Therefore, they are suitable for maps of countries such as the United States, and are not suitable for Chile and other countries, stretching from north to south. The traverse Mercator projection (different from the simple Mercator) strongly distorts in the west-east direction, but less in the north-south direction, so that it is suitable for Chile.
• What will your users think? For example, many people are familiar with the projection of the Mercator and are used to it, so the kind of distortion on the Azimuthal map can introduce them into a stupor or, conversely, arouse interest in your map. People are used to seeing the United States in Albers equal conic projection, which gives them a curved look, so the Equal Cylindrical Projection map for them may look "wrong", although everything is in order.

Projection Parameters

After you have decided on the projection, the last step remains. As we have already discussed, each projection has areas with varying degrees of distortion. Fortunately, we can choose an area with minimal distortion. This means that we can always provide minimal distortion for the most important area of ​​the map. This is achieved by selecting projection parameters. Look at these two maps using Azimuth Projection:







Both cards use the same projection, but with different parameters. The map on the left is centered on the Great Lakes region, and the map on the right to the south-east of Australia, that's why there are areas with minimal distortion. Both of them, on the contrary, use Azimuthal projection, that is, they show distances without distortion if measured from the center of projection. Thus, changing this parameter (center), we can customize the projection for our goals.



Different projections have different parameters that must be set. The projection parameters derive from the mathematical model used (which we leave out of the box for now). If the projection is necessary to specify the longitude of the center and / or latitude of the center, specify the coordinates of the center of your map. As in the example above, you define in this way an area with minimal distortion.



Some projections, such as the equal-sized cone of Albers, require the specification of the main parallels in addition to or in place of the center coordinates. These are the lines along which there will be no distortion. For example, the main parallel of 30ºN means that there will be no distortions at this latitude, but they will be slightly distorted as they move north and south - 31ºN will have small distortions, 32ºN will be worse than 31ºN and so on ... others need two. If you need only one, then specify for it the latitude of the center of your map; again we are doing so that the area with minimal distortion coincides with the area of ​​interest on our map, and not somewhere far away. If you need to specify two main parallels, then make sure that they divide the map into approximately three equal parts, as in the example below:







Thus, you will achieve that the distortions are distributed on the map in the minimum possible way.

Our recommendations

In the choice of projection there is no one correct answer, because the choice depends on the weight of all the factors listed above. However, there are a few rules that can narrow the choices. If you work with web maps, then this is usually Mercator. Note that this projection is considered to be unsuitable for most thematic maps, for all that there are more neighborhood maps, so avoid the Mercator in these cases. If you are creating a choropleth or dot density cartogram , choose an equal projection. As a crib for the types of projections and their use, use a table created by Bill Rankin.



The original work was done by Axis Maps . License: Creative Commons Attribution-NonCommercial 4.0 International License .

Author translation: KoGor. The translation license is also a Creative Commons Attribution-NonCommercial 4.0 International License .