Thematic maps - Università degli Studi di Trento
Transcript
Thematic maps - Università degli Studi di Trento
Thematic maps Thematic maps Paolo Zatelli Dipartimento di Ingegneria Civile ed Ambientale Università di Trento Paolo Zatelli | Università di Trento | 1 / 58 Thematic maps Outline 1 Thematic cartography 2 Data, features and representation Data Representation Points, lines and areas 3 Attribute classification Natural breaks - Equal intervals - Standard deviation Quantiles - Equal areas - Progressions User defined classes Paolo Zatelli | Università di Trento | 2 / 58 Thematic maps | Thematic cartography Thematic maps What are they used for? Thematic maps are used to display a “thematism” of data, i.e. to display an attribute of geometric features (points, lines, areas). To create thematic maps we need: graphics to choose line types and colors to display the attribute symbols to display the attribute statistics to choose the classes for the attribute’s values Paolo Zatelli | Università di Trento | 3 / 58 Thematic maps | Thematic cartography Thematic maps An example A thematic map is used to highlight the distribution of an attribute. The example below displays the distribution of the world population. Countries Paolo Zatelli | Università di Trento | Countries by population (n. of inhabitants) 4 / 58 Thematic maps | Thematic cartography Vector and raster Entities or fields Usually thematic maps are created from vector data, however, it is possible to user raster maps, modifying the color table or reclassifying them. DTM of the Spearfish County Paolo Zatelli | Università di Trento | Reclassified DTM (9 classes of 100m) 5 / 58 Thematic maps | Data, features and representation | Data Outline 1 Thematic cartography 2 Data, features and representation Data Representation Points, lines and areas 3 Attribute classification Natural breaks - Equal intervals - Standard deviation Quantiles - Equal areas - Progressions User defined classes Paolo Zatelli | Università di Trento | 6 / 58 Thematic maps | Data, features and representation | Data Input data What is needed to create a thematic map To create thematic maps we need: 1 geometry (features: points, lines, areas) 2 attributes to be displayed on the map (obviously connected to the features) In the vector model an attributes corresponds to a column in the table associated to geometric features (points, lines and areas). Paolo Zatelli | Università di Trento | 7 / 58 Thematic maps | Data, features and representation | Data Input data Geometry Paolo Zatelli | Università di Trento | 8 / 58 Thematic maps | Data, features and representation | Data Input data Attributes FIPS AC AG AJ AL AM AO AQ AR AS BA BB BD BF BG BH .. ISO2 AG DZ AZ AL AM AO AS AR AU BH BB BM BS BD BZ .. ISO3 ATG DZA AZE ALB ARM AGO ASM ARG AUS BHR BRB BMU BHS BGD BLZ .. UN 28 12 31 8 51 24 16 32 36 48 52 60 44 50 84 .. Paolo Zatelli | Università di Trento | NAME Antigua and Barbuda Algeria Azerbaijan Albania Armenia Angola American Samoa Argentina Australia Bahrain Barbados Bermuda Bahamas Bangladesh Belize .. AREA 44 238174 8260 2740 2820 124670 20 273669 768230 71 43 5 1001 13017 2281 .. POP2005 83039 32854159 8352021 3153731 3017661 16095214 64051 38747148 20310208 724788 291933 64174 323295 15328112 275546 .. REGION 19 2 142 150 142 2 9 19 9 142 19 19 19 142 19 .. SUBREGION 29 15 145 39 145 17 61 5 53 145 29 21 29 34 13 .. 9 / 58 Thematic maps | Data, features and representation | Data Population distribution by country frequency World populatione - 30 equal classes data Paolo Zatelli | Università di Trento | 10 / 58 Thematic maps | Data, features and representation | Representation Outline 1 Thematic cartography 2 Data, features and representation Data Representation Points, lines and areas 3 Attribute classification Natural breaks - Equal intervals - Standard deviation Quantiles - Equal areas - Progressions User defined classes Paolo Zatelli | Università di Trento | 11 / 58 Thematic maps | Data, features and representation | Representation Vector layer representation Data distribution The representation of a layer substantially depends on the type and on the distribution of the attribute to be depicted. Examples: the attribute is qualitative (type of soil) or quantitative (pH of the soil)? the distribution (of pH) is uniform or some values have different frequencies? Paolo Zatelli | Università di Trento | 12 / 58 Thematic maps | Data, features and representation | Representation Vector features and layers How an attribute value is represented on the map In a GIS three types of vector features are used, points, lines and areas, which are represented by: points lines areas symbols hatching symbols colors colors colors dimensions widths dithering An internal point (e.g. centroid) can be used to represent an attribute of an area. Paolo Zatelli | Università di Trento | 13 / 58 Thematic maps | Data, features and representation | Representation Types of attributes - I to create thematic maps Representation of attributes depends on their properties: 1 Quantitative nominal [or categorical] groups have names (“labels”) but not values (e.g. land use, country, etc.). fuzzy sets like nominal attributes, but each element belongs to a category to a given “degree”. Paolo Zatelli | Università di Trento | 14 / 58 Thematic maps | Data, features and representation | Representation Types of attributes - II to create thematic maps 2 Qualitative ordinal quantities do not have a numeric meaning, still it possible to sort them, therefore to assign a number. This is often the case of categories expressing an assessment (e.g. a forest in “good”, “fairly good”, ..., “very bad” condition). In general arithmetic operations make no sense, logical operations do (e.g. reclassification). interval attributes are numeric and distributed on an interval, however the origin and the interval are arbitrary (e.g. years: different calendars use different years as “year zero”). Some arithmetic operations make sense (e.g. difference of years), other do not (e.g. ratio between years). Paolo Zatelli | Università di Trento | 15 / 58 Thematic maps | Data, features and representation | Representation Types of attributes - III to create thematic maps ratio attributes are quantitative, the origin is not arbitrary but the interval is (e.g. age, field value). Arithmetic operations are possible. count such as ratio, but measurement units are not arbitrary, therefore it is not always possible to rescale the interval (e.g. dwellers in an area, number of municipalities in a region). Paolo Zatelli | Università di Trento | 16 / 58 Thematic maps | Data, features and representation | Representation Choice of colors and dithering General criteria Some general rules: how attributes are classified is fundamental colors must be chosen to cater for color blind people colors and dithering must be used in intuitive ways: light colors correspond to lower values, darker ones to higher values thick dithering corresponds to higher values continuous lines are used for certain boundaries, dashed lines for uncertain ones Paolo Zatelli | Università di Trento | 17 / 58 Thematic maps | Data, features and representation | Points, lines and areas Outline 1 Thematic cartography 2 Data, features and representation Data Representation Points, lines and areas 3 Attribute classification Natural breaks - Equal intervals - Standard deviation Quantiles - Equal areas - Progressions User defined classes Paolo Zatelli | Università di Trento | 18 / 58 Thematic maps | Data, features and representation | Points, lines and areas Points and lines IGMI symbols IGMI: points - edifici di culto (churchs) Paolo Zatelli | Università di Trento | IGMI: lines - strade e ferrovie (roads and railroads) 19 / 58 Thematic maps | Data, features and representation | Points, lines and areas Areas - I Types of thematic maps For area features, the most used types of maps are: area class areas’ boundaries depend on an attribute because all areas with the same value are represented in the same way, therefore common boundaries are dissolved (e.g. land use). coropleth boundaries are defined when data are gathered/processed (e.g. census units). dot the number of dots (randomly placed inside the area to simulate a density) represents the attribute’s value. proportional symbols symbols (usually placed in the center for the area) proportional to the value of the attribute are used (see lines and points). Paolo Zatelli | Università di Trento | 20 / 58 Thematic maps | Data, features and representation | Points, lines and areas Areas - II Types of thematic maps isarithmic are used to represent scalar fields (e.g. height, pressure), using contour levels. 3D views can be used to represent an attribute (assuming continuous or discrete values) simulating a “height”. Paolo Zatelli | Università di Trento | 21 / 58 Thematic maps | Data, features and representation | Points, lines and areas Area class - I All areas with the same value are represented in the same way, therefore boundaries change depending on the attribute’s value. Whenever possible, colors and dithering appealing to the phenomenon are used (e.g. green -> forest, blue -> hydrography, etc.). This kind of map can be used to evaluate the dispersion/concentration of phenomena (e.g. same land use in different areas). Paolo Zatelli | Università di Trento | 22 / 58 Thematic maps | Data, features and representation | Points, lines and areas Area class - II Countries by region (continent) Paolo Zatelli | Università di Trento | 23 / 58 Thematic maps | Data, features and representation | Points, lines and areas Dots map - I A dot represents one unit (or multiple) of the attribute’s value (e.g. 1 airport, 100000 inhabitants). Dots are randomly placed in each area, to simulate a density. There is no need to split the attribute’s range into classes. Paolo Zatelli | Università di Trento | 24 / 58 Thematic maps | Data, features and representation | Points, lines and areas Dots map - II World population in 2005 - 1 dot = 500 000 inhabitants Paolo Zatelli | Università di Trento | 25 / 58 Thematic maps | Data, features and representation | Points, lines and areas Proportional symbols map - I Symbol’s dimension is indicative of the value of the attribute. Symbols with simple shapes are used (circles, squares and triangles); their surface is scaled as a function of the attribute’s value. 3D symbols can be used to represent more than one attribute at a time; however, these are usually less readable than 2D symbols. Paolo Zatelli | Università di Trento | 26 / 58 Thematic maps | Data, features and representation | Points, lines and areas Proportional symbols map - II User perception of surface usually leads to underestimate them, with a greater effect on large areas. Possible solutions are: to use the apparent scale method, where symbols’ surfaces grows faster than direct proportionality (exponentials are used for circles, usually r 1.14 instead of r ) to use symbols with discrete values of surface to optimize the legend to remove this effect Paolo Zatelli | Università di Trento | 27 / 58 Thematic maps | Data, features and representation | Points, lines and areas Proportional symbols map - III World population in 2005 Paolo Zatelli | Università di Trento | 28 / 58 Thematic maps | Data, features and representation | Points, lines and areas Isarithmic - I to represent scalar fields The distribution of an attribute is represented using lines passing through points with the same attribute’s value. Usually lines correspond to fixed increment of the attribute’ value (e.g. a contour level each 10 meters), in this way it is possible to mimic a 3D view. Lines’ density must take into account the accuracy of the attribute’s values (e.g. contour levels on maps with different scales have different distances). Paolo Zatelli | Università di Trento | 29 / 58 Thematic maps | Data, features and representation | Points, lines and areas Isarithmic - II to represent scalar fields 3D contour levels - Trento Paolo Zatelli | Università di Trento | 30 / 58 Thematic maps | Data, features and representation | Points, lines and areas Isarithmic - III to represent scalar fields Contour levels - Trento Paolo Zatelli | Università di Trento | 31 / 58 Thematic maps | Data, features and representation | Points, lines and areas 3D views- I It is possible to use a 3D view (usually axonometric) to represent an attribute using a “height” associated to the geometric features. This approach can be used for scalar fields (e.g. DTM) and for entities (e.g. areas representing countries). The “height” can be a real height (e.g. DTM, buildings’ height) or any attribute (even not an extent). Paolo Zatelli | Università di Trento | 32 / 58 Thematic maps | Data, features and representation | Points, lines and areas 3D views- II Scalar field - DTM 3D - Trento Paolo Zatelli | Università di Trento | 33 / 58 Thematic maps | Data, features and representation | Points, lines and areas 3D views- III Entities - World population in 2005 Paolo Zatelli | Università di Trento | 34 / 58 Thematic maps | Attribute classification Attribute classification Most of the times an attribute has a large number of different values: if each value is represented in a different way (by color, symbol, etc.) the resulting map is impossible to read. For this reason, attribute’s values are grouped together in classes. Usually, no more than 5 - 7 classes are used at the same time. A method to set the intervals identifying the classes must be chosen. Such method must minimize the differences between elements of the same class, while maximizing the differences with elements of the other classes. In some cases classes are chosen according to legal or de facto standards. Paolo Zatelli | Università di Trento | 35 / 58 Thematic maps | Attribute classification Classification methods Different methods can be used to classify data, i.e. to chose the intervals for grouping data. The choice of the intervals depend on the attribute’s distribution: natural breaks for non normal distributions equal intervals for uniform or normal distributions std deviation for normal distributions quantile for uniform or normal distributions constant area for maps with areas with similar values progressions for distributions with many values on the tails user defined if the possibilities above are not satisfying Paolo Zatelli | Università di Trento | 36 / 58 Thematic maps | Attribute classification | Natural breaks - Equal intervals - Standard deviation Outline 1 Thematic cartography 2 Data, features and representation Data Representation Points, lines and areas 3 Attribute classification Natural breaks - Equal intervals - Standard deviation Quantiles - Equal areas - Progressions User defined classes Paolo Zatelli | Università di Trento | 37 / 58 Thematic maps | Attribute classification | Natural breaks - Equal intervals - Standard deviation Natural breaks - I This approach is used for non normal nor uniform distributions. Classes’ boundaries are placed at discontinuities of the distribution. Given the number of classes, the difference between the sum of square differences inside each class and the sum of the square difference with respect to the global mean is maximized (Jenks’ algorithm). Advantages: classes have the maximum possible internal homogeneity this is a robust classification method it is easy to implement an automatic procedure Paolo Zatelli | Università di Trento | 38 / 58 Thematic maps | Attribute classification | Natural breaks - Equal intervals - Standard deviation Natural breaks - II Drawbacks: results are good only if discontinuities do actually exist in the distribution (at least a number of discontinuities equal to the number of classes-1) the outcome depends on the number of classes the procedure is iterative and can be computationally intensive classes depend on the distribution, therefore the comparison of maps is usually difficult (because in general the same attribute in different regions has different distribution and therefore different classes) Paolo Zatelli | Università di Trento | 39 / 58 Thematic maps | Attribute classification | Natural breaks - Equal intervals - Standard deviation Natural breaks - III frequency break break break break Natural breaks - 5 classes class 1 class 2 class 3 class 4 cl. 5 data Paolo Zatelli | Università di Trento | 40 / 58 Thematic maps | Attribute classification | Natural breaks - Equal intervals - Standard deviation Natural breaks - IV World population in 2005 - natural breaks Paolo Zatelli | Università di Trento | 41 / 58 Thematic maps | Attribute classification | Natural breaks - Equal intervals - Standard deviation Equal intervals - I Classes with equal width This method is used for uniform or quasi uniform distributions. Interval (constant) width is given by max − min n. of classes Advantages: it is easy to understand if the absolute maximum and minimum are used, classes are independent from the mapped region, therefore maps are comparable Drawbacks: it is not suitable if the distribution is not uniform or quasi uniform (empty classes/overcrowded classes) it “hides” the differences between values Paolo Zatelli | Università di Trento | 42 / 58 Thematic maps | Attribute classification | Natural breaks - Equal intervals - Standard deviation Equal intervals - II Classes with equal width frequency Equal intervals - 5 classes class 1 class 2 class 3 class 4 class 5 data Paolo Zatelli | Università di Trento | 43 / 58 Thematic maps | Attribute classification | Natural breaks - Equal intervals - Standard deviation Equal intervals - III World population in 2005 - equal intervals Paolo Zatelli | Università di Trento | 44 / 58 Thematic maps | Attribute classification | Natural breaks - Equal intervals - Standard deviation Standard deviation - I for normal distributions It is used when attribute’s distribution is nearly Gaussian. Intervals are defined around the mean value, with amplitudes that are multiples of the standard deviation (e.g. 1, 2, 3 times the std. dev.). It is useful to highlight the distribution of the values around the mean value. Paolo Zatelli | Università di Trento | 45 / 58 Thematic maps | Attribute classification | Natural breaks - Equal intervals - Standard deviation Standard deviation - II for normal distributions m + 1 std. dev. Standard deviation - 5 classes Values 1-30 m - 2 std. dev. frequency Std. dev. = 6.65 mean m - 1 std. dev. Mean = 18.30 class 1 class 2 class 3 class 4 class 5 data Paolo Zatelli | Università di Trento | 46 / 58 Thematic maps | Attribute classification | Natural breaks - Equal intervals - Standard deviation Standard deviation - III World population in 2005 - standard deviation Paolo Zatelli | Università di Trento | 47 / 58 Thematic maps | Attribute classification | Quantiles - Equal areas - Progressions Outline 1 Thematic cartography 2 Data, features and representation Data Representation Points, lines and areas 3 Attribute classification Natural breaks - Equal intervals - Standard deviation Quantiles - Equal areas - Progressions User defined classes Paolo Zatelli | Università di Trento | 48 / 58 Thematic maps | Attribute classification | Quantiles - Equal areas - Progressions Quantiles - I Classes with equal numerosity Classes are chosen so that they have the same number of elements. It is useful for distributions not having similar values (in this case the amplitudes of the classes are very different). In some cases, artificial discontinuities result, because similar values can be placed in different classes (near the boundary between two classes). Paolo Zatelli | Università di Trento | 49 / 58 Thematic maps | Attribute classification | Quantiles - Equal areas - Progressions Quantiles - II Classes with equal numerosity Quantiles - 5 classes Data 240 Classes 5 frequenza Elements per class 48 class 1 class 2 class 3 class 4 class 5 dati Paolo Zatelli | Università di Trento | 50 / 58 Thematic maps | Attribute classification | Quantiles - Equal areas - Progressions Quantiles - III World population in 2005 - quantiles Paolo Zatelli | Università di Trento | 51 / 58 Thematic maps | Attribute classification | Quantiles - Equal areas - Progressions Equal areas Classes with equal surface A similar approach is to use equal areas: classes are chosen so that for each class the surface is as much as possible the same. When areas have similar surfaces, the result is the same of using quantiles. When areas have different surfaces, the differences between smaller areas tend to vanish, because these areas are grouped together. Paolo Zatelli | Università di Trento | 52 / 58 Thematic maps | Attribute classification | Quantiles - Equal areas - Progressions Progressions Classes with variable width Attributes with arithmetic or geometric progressions can be represented with classes with this type of distributions. The width is different for each class, but this is systematic. This approach is often used for distributions with many values on the tails. Paolo Zatelli | Università di Trento | 53 / 58 Thematic maps | Attribute classification | User defined classes Outline 1 Thematic cartography 2 Data, features and representation Data Representation Points, lines and areas 3 Attribute classification Natural breaks - Equal intervals - Standard deviation Quantiles - Equal areas - Progressions User defined classes Paolo Zatelli | Università di Trento | 54 / 58 Thematic maps | Attribute classification | User defined classes User defined classes User defined classes: it is used when the other approaches do not lead to satisfactory results (e.g. “difficult” distributions) it is used after the application of the other methods to round up classes’ boundary values, to make the more readable it can (but should not) be used to highlight or hide a certain feature of an attribute (e.g. to “hide” areas with higher values) the result depend on the user experience Paolo Zatelli | Università di Trento | 55 / 58 Thematic maps | Attribute classification | User defined classes Comparison of thematic maps Different classification methods Natural breaks Equal intervals Standard deviation Quantiles Paolo Zatelli | Università di Trento | 56 / 58 Thematic maps | Attribute classification | User defined classes Surfaces proportional to the attribute normalized Some representation techniques distort areas so that their surface represent the attribute’s value with respect to the total value. World population in 2000 - surface of each area is proportional to its population c 2006 SASI Group and Mark Newman Paolo Zatelli | Università di Trento | 57 / 58 Thematic maps | Appendice | License c This presentation is 2009 Paolo Zatelli, available as Data for thematic maps are available at “the Mapping Hacks website”: http://www.mappinghacks.com/data/ with Creative Commons Attribution-Share Alike license. World population in 2000 map is available with Attribution-Noncommercial-No Derivative Works 3.0 Unported license, c Copyright 2006 SASI Group (University of Sheffield) and Mark Newman (University of Michigan). All maps and charts have been created using only Free Software under GNU license. Paolo Zatelli | Università di Trento | 58 / 58