International Journal of Geo-Information Article Heat Maps: Perfect Maps for Quick Reading? Comparing Usability of Heat Maps with Different Levels of Generalization Katarzyna Słomska-Przech * , Tomasz Panecki and Wojciech Pokojski Department of Geoinformatics, Cartography and Remote Sensing, Faculty of Geography and Regional Studies, University of Warsaw, Krakowskie Przedmiescie 30, 00-927 Warsaw, Poland; [email protected] (T.P.); [email protected] (W.P.) * Correspondence: [email protected] Abstract: Recently, due to Web 2.0 and neocartography, heat maps have become a popular map type for quick reading. Heat maps are graphical representations of geographic data density in the form of raster maps, elaborated by applying kernel density estimation with a given radius on point- or linear-input data. The aim of this study was to compare the usability of heat maps with different levels of generalization (defined by radii of 10, 20, 30, and 40 pixels) for basic map user tasks. A user study with 412 participants (16–20 years old, high school students) was carried out in order to compare heat maps that showed the same input data. The study was conducted in schools during geography or IT lessons. Objective (the correctness of the answer, response times) and subjective (response time self-assessment, task difficulty, preferences) metrics were measured. The results show that the smaller radius resulted in the higher correctness of the answers. A larger radius did not result in faster response times. The participants perceived the more generalized maps as easier to use, although this result did not match the performance metrics. Overall, we believe that heat maps, Citation: Słomska-Przech, K.; in given circumstances and appropriate design settings, can be considered an efficient method for Panecki, T.; Pokojski, W. Heat Maps: spatial data presentation. Perfect Maps for Quick Reading? Comparing Usability of Heat Maps Keywords: heat map; thematic map; user study; generalization with Different Levels of Generalization. ISPRS Int. J. Geo-Inf. 2021, 10, 562. https://doi.org/ 10.3390/ijgi10080562 1. Introduction Academic Editor: Wolfgang Kainz The world of small-scale mapping on the web is constantly evolving. The main aim of such cartographic representations is often to quickly and quite effectively present Received: 6 July 2021 geographical relations of both qualitative and quantitative character. For the latter, different Accepted: 15 August 2021 thematic map types are used, including those already well established, such as diagrams, Published: 18 August 2021 choropleth maps, dot maps, or isolines. In the age of neocartography and Web 2.0 [1,2], new map types have emerged, such as heat maps that allow point data density based on Publisher’s Note: MDPI stays neutral point-to-area estimation to be visualized [3–5]. From the point of view of the classification with regard to jurisdictional claims in of data presentation methods by MacEachren and DiBiase [6], heat maps can be classified published maps and institutional affil- as continuous and smooth maps. The growing popularity of heat maps comes from their iations. attractiveness and ease of creation, using various mapping libraries [5]. Despite being commonly applied, it has not been evaluated whether they are an effective solution as maps for quick reading in a web environment. Similarly, it has not been verified to what extent their level of detail (generalization) is a key issue. Copyright: © 2021 by the authors. Heat maps were imported into cartography from data visualization techniques, similar Licensee MDPI, Basel, Switzerland. to other map types already well established in cartography, such as diagrams, charts, dots or This article is an open access article choropleths [7]. Heat maps are visualizations for the graphical representation of the density distributed under the terms and of spatial phenomena, usually measured in points. De Boer [3] highlights the fact that conditions of the Creative Commons the term itself is not unambiguous; it can denote both the density map (regardless of Attribution (CC BY) license (https:// the method used) and the process of estimation of point-to-surface data (point density creativecommons.org/licenses/by/ estimation). Heat maps are not necessarily connected strictly to geography [8]. They are 4.0/). ISPRS Int. J. Geo-Inf. 2021, 10, 562. https://doi.org/10.3390/ijgi10080562 https://www.mdpi.com/journal/ijgi ISPRS Int. J. Geo-Inf. 2021, 10, 562 2 of 24 ISPRS Int. J. Geo-Inf. 2021, 10, x FOR PEER REVIEW 2 of 25 used in medicine [9], chemistry [10], biology and ecology [11], the social sciences for non- used in medicine [9], chemistry [10], biology and ecology [11], the social sciences for non- spatial data [12], and eye-tracker analysis [13]. In cartography, heat maps can be found in spatial data [12], and eye-tracker analysis [13]. In cartography, heat maps can be found in studies related to the spatial distribution of social issues [14–16], the visualization of routes studies related to the spatial distribution of social issues [14–16], the visualization of for runners and cyclists [17,18], and the analysis of road accidents [19,20]. The popularity routes for runners and cyclists [17,18], and the analysis of road accidents [19,20]. The pop- of heat maps is growing in the age of big data, as is the need for fast and attractive ularity of heat maps is growing in the age of big data, as is the need for fast and attractive visualizations [5,21,22] (Figure1). visualizations [5,21,22] (Figure 1). FigureFigure 1. 1.Selected Selected examples examples of of a heata heat maps maps with with Open Open Street Street Map Map base base maps maps ((A (()A heat) heat map map prepared prepared in ArcGISin ArcGIS Online Online [22], [22], (B) Location History Visualizer [21]). (B) Location History Visualizer [21]). TheThe designdesign of heat maps in cartography cartography ca cann be be considered considered from from various various perspectives: perspectives: mappedmapped data,data, estimationestimation methods, base map, color color scheme, scheme, legend, legend, and—last and—last but but not not least—generalization.least—generalization. InputInput data data are are usually usually referred referred to points,to points, and and less less frequently frequently to lines. to Methodslines. Methods of transition of transition fromsource from source data to data surfaces to surfaces is done isby done estimation, by estimation, usually usually Kernel DensityKernel Density Estimation Estimation or Point/Line or Point/Line Density Dens Estimationity Estimation [17,23 ,[17,23,24].24]. Most Most often, often, heat mapsheat comemaps with come spectral with spectral or hypsometric or hypsometric scales, butscales, single but colors single are colors used are as wellused [ 4as]. well As the [4]. maps As arethecreated maps are for created quick reading, for quick they reading, are not they always are supplemented not always supplemented by a legend, andby a the legend, colors areand self-evident the colors are (red self-evident = more, green (red == less,more, etc.). green Legends = less, canetc.). also Legends be ordinal/interval can also be ordi- and refernal/interval to “low-to-high” and refer values.to “low-to-high” The base maps values used. The for base heat maps maps varyused fromfor heat OpenStreetMap maps vary orfrom Google, OpenStreetMap via satellite or imagery Google, to via highly satellite generalized imagery topographicto highly generalized content—for topographic example, streets—especiallycontent—for example, in printed streets—especially maps [14]. in printed maps [14]. GeneralizationGeneralization plays an important role role in in every every map, map, including including thematic thematic maps maps [25]. [25 ]. TheThe detailednessdetailedness of of heat heat maps maps is isreflected reflected by by the the radius radius of the of thekernel kernel estimation estimation function: func- tion:the higher the higher the radius, the radius, the more the generalized more generalized the map, the and map, the “hot and thespots” “hot are spots” more areblurred. more blurred.Generalization Generalization is crucial, isespecially crucial, in especially non-interactive in non-interactive maps, which maps, cannot which be dynamically cannot be dynamicallyrescaled; this rescaled; factor influences this factor the influences effectiveness the effectivenessof web maps. of There web maps.can be Thereno effective can be nothematic effective map thematic without map simplifying without input simplifying data and input cartographically data and cartographically refining them. Raposo refining them.et al. [26] Raposo underline et al. [ 26the] underlinerole of generalization the role of generalization in thematic mapping in thematic by stating mapping that by “gener- stating thatalization “generalization is ubiquitous is ubiquitousand critical andin all critical cartography, in all cartography, and by corollary and bythat corollary it is an im- that itportant is an importantaspect of the aspect highly of thepopular highly themat popularic mapping thematic currently mapping capturing currently public capturing and publicotherwise and non-cartographer otherwise non-cartographer attention”. The attention”. authors The also authors applied also the typology applied the of general- typology ofization generalization operators operators(for content, (for geometry, content,geometry, symbol and symbol label) and proposed label) proposedby Roth, Brewer, by Roth, Brewer,and Stryker and