INOM
EXAMENSARBETE SAMHÄLLSBYGGNAD, AVANCERAD NIVÅ, 30 HP
STOCKHOLM SVERIGE 2019,
Custom Base Maps for Utility Network Applications
RASMUS EDUARDS
KTH
SKOLAN FÖR ARKITEKTUR OCH SAMHÄLLSBYGGNAD
Custom Base Maps for Utility Network Applications
Rasmus Eduards Stockholm 2019
Terms and Abbreviations
Base map The bottom-most layer in application or in layer scheme, also known as Background map.
Slippy map Internet map which allow zooming and panning.
GIS Geographical Information Systems.
OSM Open Street Map, an open source web map.
dpWebmap Web application developed by the company Digpro.
Mapbox GL JS Mapbox own graphical language for Javascript.
JavaScript Coding language for web client operations.
Utility Network A network that are used for a specific purpose, usually contains cables, pipe lines, roads etc.
Geo portal A internet accessed web platform which contains spatial data.
CSS Cascading Style Sheets.
HTML HyperText Markup Language
Abstract
Web maps are today used more than ever. These are often displayed in geographical information system solutions. Web maps are often constructed in the terms of different layers where the bottom-most layer is called base map or background map. A specific kind of base maps are those that are the canvas for utility networks. These base maps can look vastly different from each other.
Sometimes the base map can be overlooked. Currently there are different theories and opinions on how these base maps shall be designed and what information they shall contain.
This thesis investigates a suitable base map design for utility network management and also creates some prototypes as a proof of concept. The thesis focus mostly on three factors, colour, in- formation visualisation and symbology. This is investigated by using a user-centred design approach and comparing it to existing findings in literature and among map theories. The user-centred de- sign process involves a case study performed with participants that work with utility networks in web map solutions. The research method is an iterative process where the participants are answering three quantitative surveys. The surveys includes prototypes which is refined for each step after analysing the participants answers.
The thesis concludes that low saturated colours is good practise to enhance the network, this is both evident in literature as well as in the case study. Amount of lightness in the base map seem to be more of a opinionated matter and can highly differentiate depending on users device.
Using hue is recommended to make borders between map elements more distinct. The base map shall only obtain the most relevant information and more specific information is recommended to be added through additional layers. Further studies need to be done in order to investigate the networks design and there is also a need to investigate how a map solution with several layers shall be design and how the relationships between those layers shall be constructed for utility network management.
Acknowledgements
This M.Sc. thesis concludes my studies at the Degree Programme in Civil Engineering and Urban Management and the Master of Transport and Geoinformation Technology at KTH Royal Institue of Technology in Stockholm, Sweden.
The thesis has been conducted in cooperation with company Digpro AB and I would like to thank everyone there specially my supervisor Peter Markus whom offered feedback and support.
I would also like to thank all the customers of Digpro’s product that participated in the thesis surveys.
Addionally I would like to thank my academic supervisor at KTH Associate Professor Gyözö Gidofalvi for giving feedback and inspiration. Last but not least, I would like to thank examiner Professor Yifang Ban for giving feedback and examining this thesis.
Rasmus Eduards Stockholm, Sweden June 2019
Sammanfattning
Webbkartor används idag mer än någonsin. De är ofta visualiserade i geografiska information system. Webbkartor är ofta kontruereade i form av flera lager där det understa lagret benämns som baskarta eller bakgrundskarta. En specifik typ av baskarta är den som används som bakgrund för allmännyttiga nätverk. Dessa baskartor kan se väldigt olika ut och skilja sig från varandra. I vissa fall kan baskartor bli förbisedda. För närvarande finns det olika teorier and åsikter om hur dessa baskartor ska designas och vad för information det ska innehålla.
Denna uppsats undersöker en lämplig baskartedesignen för allmännyttiga nätverksapplikationer och tar fram några prototyper. Denna uppsats fokuserar framförallt på tre faktorer färger, infor- mationsåtergivning och symobologi. Denna utredning görs med en användarfokuserad approach och jämför dessa upptäckter med litteratur och andra kända teorier i området av kartografi och baskartor. Den användarfokuserade design processen involverar en fallstudie som utförs på delt- agare som arbetar med allmänytigga nätverksapplikationer till vardags. Metoden som används i denna fallstudie är iterativ och genomförs i tre delar. Fallstudien utvärdera också prototyper som förfinas i varje del efter feedback från deltagarna.
Uppsatsen sammanfattar att färger med låg färgmättnad är att rekommendera när det kommer till baskartor som ska användas till allmännyttiga nätverk, detta stärks både från fallstudien och litteraturen. Använding av ljusstyrka är med tvetydlig och verkar vara mer personlig och vari- era beroende på användarens arbetsplattform. Det rekommenderas att använda färgtoner för att förstärka gränser mellan olika marktyper. Baskartan ska bara visa den allra viktigaste informatio- nen och mer specifik information rekommenderas att läggas till i form av yttligare lager. Fortsatt arbete behövs att göras för att investigera nätverkets utformning och hur webbkarta med flera lager ska desginas samt hur dessa lager ska förhålla sig till varandra.
Contents
1 Introduction 7
1.1 Background and Rationale . . . . 7
1.2 Research Objectives . . . . 7
1.3 Scope and Limitations . . . . 7
1.4 Disposition . . . . 8
2 Defiintions and Related Works 9 2.1 Definitions and Concepts of Map Design . . . . 9
2.1.1 Map Colour . . . . 9
2.1.2 Map Symbology and Information Displaying . . . . 10
2.1.3 General Practises and Recommendation of Map Design . . . . 11
2.2 Related Studies . . . . 12
3 Methodology 14 3.1 Research Strategy . . . . 14
3.2 User-Centred Design . . . . 14
3.3 Identification of Influential Factors . . . . 15
3.4 Choice and Validation of Methodology for Data Collection . . . . 16
3.5 Risks and Undertakings with Research Strategy . . . . 16
4 Data collection and Implementation of Methodology 17 4.1 Case Study . . . . 17
4.1.1 Survey 1 . . . . 17
4.1.2 Survey 2 . . . . 18
4.1.3 Survey 3 . . . . 19
4.2 Data Collection Integration . . . . 19
4.3 Data Processing . . . . 19
4.4 Prototyping . . . . 20
5 Results 21 5.1 Survey 1 . . . . 21
5.1.1 Personal Information . . . . 21
5.1.2 Colour . . . . 21
5.1.3 Information and Symbology . . . . 23
5.1.4 Qualitative Data . . . . 24
5.2 Survey 2 . . . . 24
5.2.1 Information Quantity . . . . 24
5.2.2 Degree of Map Components in Colour . . . . 25
5.2.3 Amount of Chroma and Lightness . . . . 26
5.2.4 Qualitive Data . . . . 27
5.3 Survey 3 . . . . 28
5.3.1 Final Prototype . . . . 28
6 Discussion 30 6.1 Implementation of User-Centred Design . . . . 30
6.2 Optimal Base Map Design . . . . 30
6.3 Prototype Evaluation . . . . 31
7 Conclusion and Future Work 32 7.1 Conclusion . . . . 32
7.2 Future Work . . . . 32
1 Introduction
1.1 Background and Rationale
Web maps are now widely used in general public thanks to geoportals and online mapping plat- forms. These web maps usually consist of several layers and can have many utility areas. The base map is the bottom-most layer and consist of static information, that helps the user to navigate in the environment. Base maps can consist of different symbology and different generalisation for a specific design. As well, the specific colours of the entities have to be considered. Today, in different GIT systems the base maps are sometimes taken for granted and have not been given enough thought or rationale in its design.
When creating and designing a base map generic cartography design cannot be directly applied, since the base map will not be the main focus of interest. Instead, its function is to enhance the layers of interest that is stacked on top of the base map. A certain kind of base maps are those that apply to the field of utility networks. Utility Networks are distribution lines, often displayed in a vector layer on top of the base map. In the working field of utility networks the user have to be able to navigate quickly and get a clear overview of their systems with ease. One example of a system that utility network companies work with today is dpSpatial developed by Digpro AB.
Digpro AB deliver their web platform with the open source developed OSM (OpenStreetMap) as their base map, which is not specialized as a base map to manage utility networks systems with.
However, it is not certain how a base map shall be constructed in these cases and how one shall arrive at a design, deciding colour schemes and what information shall be displayed. The focus of this thesis is to investigate if there is an optimal way of constructing base maps and how one shall arrive at that design.
Below are some examples showing the standard layer of OSM at certain zoom level. However, the information shown differentiate depending on the zoom level. This is called generalization and is another important factor to consider when developing so called web maps (web maps where pan and zoom is a possibility). Important to notice is the amount of information that is shown.
Figure 1: Example of a zoomed in OSM in Digpro’s product dpWebmap.
1.2 Research Objectives
The objectives of this thesis research are three folds:
• Investigate a suitable base map design for digital utility network management
• Investigate an user involved design process and compare findings to existing literature and theories
• Develop a few base map prototypes as proof of concept
1.3 Scope and Limitations
This study is delimited to base maps used for utility network management only. The study will not delimit itself to a certain field of utility networks since utility networks do not differ significantly.
This study will focus solely on the base map and not the additional layers and functions that the geoportal consists of. The vector layer displaying the utility network is also of interest. However, general recommendations regarding colour schemes will be considered.
Further, there will be three specific focus areas regarding the topic of base maps: map colours, map information and map symbology. Other map elements such as generalization, text placement and style of text and symbols will be considered and will be based on existing recommendation and common use. However, it will not be the area of main focus.
1.4 Disposition
This thesis will firstly cover the existing literature as well as definitions of map concepts in Section 2. Related studies will also be covered in Section 2. Further, Section 3 will present the research strategy for this study and reasoning for why it was chosen. Followed by Section 4 explaining how the case study was conducted and implemented in to the research strategy. The result is presented in Section 5. Lastly the discussion in Section 6 and conclusion and future work in Section 7 is presented.
2 Defiintions and Related Works
This section covers the basic concepts and definitions of cartography in Section 2.1. Related Work is further discussed in Section 2.2, where progression of similar studies will be presented.
2.1 Definitions and Concepts of Map Design
2.1.1 Map Colour
According to Robinson et al. [3] colour have three dimensions. However, these can be measured in different colour modelling systems. There is mainly three common digital systems: RGB Color Model, HLS Colour Model and HVC colour model.
RGB Colour Model The RGB Colour Model consists of the three colour Red, Green and Blue.
The system may be visualized similar to a cube with the coordinate X, Y, Z and each coordinate controls the intensity of Red, Green and Blue. The value of these coordinates can be between 0 and 255. Where 0, 0, 0 is black with no light and where 255, 255, 255 corresponds to the other corner of the cube which displays white with maximum light.
HLS Colour Model The HLS Colour Model consists also of three parameters Hue, Lightness and Saturation. It was developed in 1970s by Tektronix Inc. It often gets visualized with double sided cone where white is at the top and black in the bottom. Hue is represented as the angle of the circle within the cone and the saturation as the radius which is the biggest between the two cones when the there is as much white as black.
HVC Colour Model The HVC Colour Model consist of the three parameters Hue, Value, Chroma developed by Tektronix Laboratories. Which is a electronic system developed from the classic Munsell System. Each hue is measured by its angular distance from red. Value is on scale from 0(black) to 100(white) and Chroma varies from 0 at the center of the axis to 100 at outer circle. This can be displayed in Figure 2.
Hue According to Robinson et al. [3] when people in common tongue talk about a thing having different colour they actually refer to a thing having different hues, i.e yellow, red and blue.
Lightness According to Robinson et al. [3] lightness is the amount of lightness or darkness reflected from a certain hue. Usually when the lightness is high it refers to the colour being light and when the lightness is low it refers to the colour being dark. This can also be referred to as brightness, value or luminosity.
Chroma Robinson et al. [3] describes chroma as measurement for how colourful a specific hue is. Chroma ranges from low chroma with grey hue with no apparent colour to high chroma with clear hue with no grey tones in the colour. Intensity, saturation and richness is other terms to refer to this phenomena. Dent [7] describes it as measurement of the purity of the colour.
Visual Efficiency According to Robinson et al. [3] the visual efficiency is the amount of re- flectance or luminosity between colour of the background and the featured symbol. Robinson et al. continues to say that colour contrast is the basis of a good figure-ground relationship and that lightness is the primary colour dimension cartographers use to promote figure fore-ground.
Robinson et al. mentions that there are different type of colour hierarchies where the eye picks up certain colours as closer than other. Examples of colour which is observed as being close is yellow, orange and red. While green, blue, brown and grey is observed as being far away. However, according to Bertin [4] hue is not a necessary tool to use. Instead Bertin sees hue as a compliment to rank different colours. Colour contrasting is another viable option to promote figure fore-ground according to Robinson et al. An important feature is to have low chroma colours as foreground and high chroma colours as promoting figure in the map. Another way to increase visual efficiency is to have areas with patterns. Patterns is according to Robinson et al. something the eye picks up as figure, being of high focus. The amount of focus of the pattern can be manipulated by choosing a colour with high or low chroma and choosing a hue with high or low lightness. Robinson et al.
Figure 2: Figure displaying The Munsell System: Hue, Value(Lightness) and Saturation(Chroma).
further mention it can be challenging to have pattern-background. Because it is hard to make both pattern and background seen as one entity.
Colour Preferences According to Robinson et al. humans in general prefer certain colours.
However, this can vary slightly from culture to culture. It also varies between children and adults, e.g. children prefer colours with high chroma i.e. pure hues(high chroma) with the rainbow spec- trum. Adults however generally prefer lower chroma colours with less intensity. Blue is specifically preferred while pure greenish-yellow is disliked the most.
Robinson et al. mentions that psychologists have studied preferred colour combinations for adults. The studies showed that pleasant colour combinations is the once which stand out from the fore-ground. This is obtained most commonly by shifting the lightness of the fore-ground and the colour. One being much lighter and one much darker. It is with other words highly connected with visual efficiency.
Special colour design problems Robinson et al. also mention that one has to bare in mind that colour-deficient vision exist, Robinson et al. mentions that about 8 percent of men and 0.4 percent of women. One fifth of those people are dichromats, i.e. suffer severe limitations to distinguish certain colour combinations.
2.1.2 Map Symbology and Information Displaying
Classes of Symbols Robinson et al. [3] refer to classes of symbols as a combination of primary and secondary visual variables. The primary variables according to Robinson et al. contains the three essence of colour (hue, lightness, chroma), shape, size and orientation. While the secondary visual variables contains arrangement (pattern), texture (pattern) and orientation (pattern). These components can then create what Robinson et al. calls the classes of symbols.
• Point-emphasizing symbols, usually refers conceptually to a location and can consist of signs, for example triangles and dots. It could represent intensity at a place, location of a feature or simply the positioning of map element.
• Line-emphasizing symbols map elements that are depicted as lines. Represents geographical phenomena like rivers, roads and political boundaries. Do not have to be linear, for example could represent contour lines.
• Area-emphasizing symbols, marks that an area have a common attribute. Could be anything from water to a measurable statistic. The area should have homogeneous character like a colour or a pattern.
• Volume-emphasizing symbols, represent intensity dimension through space of a spatial phe- nomenon. Examples could be height data or shifts in the shades of grey to display the intensity differences.
In Figure 3 shows example of qualitative and quantitative distinctions of the four classes of symbols.
Figure 3: Robinson et al.[3] figure showing the four different classes of symbols.
2.1.3 General Practises and Recommendation of Map Design
There are not many rules of cartography however some researchers have presented some guidelines.
According to Robinson et al. [3] the main objective of map design is to evoke the viewers environ- mental image appropriate to the maps purpose. One recommendation is explained by Shneiderman [17] and is called The Mantra of Visual Seeking. The mantra says: "Overview first, zoom and filter, then details-on-demand". Kenneth Field and Damien Demaj [10] investigate design relevance in cartography. The study mentioned several important factors:
• A map should be adapted to the user
• A map should be simple to use
• A map should be accurate, with no errors, distortions or misrepresentation
• The language shall relate to the entities of the map
• A map should be clear and pleasant to look at
• A map ideally should enable interaction with user for user customization.
2.2 Related Studies
Nadia Panchaud and Lorenz Hurni [14] from the institute of Cartography and Geo Information pre- sented an article where they presented a smart cartographic background symbolization algorithm.
The algorithm improves visual hierarchy by desaturating the background layer. They further tested the algorithm through a survey with 21 participants, which showed the algorithm to be successful.
The study emphasized the importance of desaturating the background layer while keeping the hue as colourful since having colourful hues enhances the our vision to distinguish between different map elements. Hence some amount of chroma is preferable.
The phenomena of desaturating the background layer was also prominent from a study by Toomanian et al [2]. Toomanian et al. emphasised that the use of highly saturated colours in base maps is a common problem. In the study a model was developed to desaturate the base map elements.
1. Convert the colour format from RGB (Red, Green, Blue) to HSV (Hue, Saturation, Value) 2. Decrease the saturation according to the predefined ranges (e.g. 50 procent)
3. Reconvert the new colour to the RGB format
Another study by Lafay et al. [19] introduced a way of automatic mapping and innovative on-demand mapping services at IGN France. This study presented a method for the user to design the colour theme of the map by manipulating the WMS. This made it possible for the user to have more influence of the design of the map. The result showed that the user went for more intense colour themes then the standard more desaturated maps. Hence, this study contradicts previous recommendations to desaturate the base map to some extent.
A study from Buard, E. and Ruas, A. [5] tested colour perception on personnel with cartography experience. The study showed that colours with the same chroma and different hue was observed as being different pure or strong. Hence the study recommended to not saturate the hues in the map equally without evaluating it.
Weihua Dong et al. [20] presented a study where they had been using eye tracking to investigate enhanced imagery in web mapping. In this study they focused on enhancing satellite imagery.
They found the tool of having eye tracking as something beneficial since it limits the amount of interpretation that is otherwise needed in a questionarie.
Hoarau et al. [6] published a study investigating colour use of hue, lightness and chroma. The study focused on how to limit the battery used for displaying colours on a mobile device while still fulfilling cartographic semantic relationships. The study concludes that the background layer does not have to be light and can be represented by dark colour as long as it easily differentiated to the symbology. Hence this would result in a reduction of energy consumption. Further Hoarau et al. claims that conventional colour use is not needed as long colours are selected that enhance comprehension of the map.
In Figure 4 Kenneth Field and Damien Demaj [10] visualize the complexity of map design.
According to Kenneth Field and Damien Demaj map design is a matter of three important com- ponents art, science and technology. Increasing one of these component may effect another.
Figure 4: Kenneth Field and Damien Demaj [10] view of map design in cartography.
3 Methodology
This section will cover the methodology and research strategy used for this study.
3.1 Research Strategy
There are plenty of user involved design approaches. Lucas, HC, jr. [16] recommended directly receiving data on attributes and information from the user. Pinandito, A et al. [15] argues that user involved design improve the efficiency of content delivery for mobile devices in their study.
Another well documented user involved design is that of Contextual design. Originally devel- oped in 1988 by Karen Holtzblatt and Hugh Beyer. Used in many industries and reconstructed at many occasions [9]. Contextual design is a step-by-step process carried out in the field to collect data in order to design a technical product. However, being in the field collecting data for a study is time consuming and the amount of participants becomes limited. The in-depth knowledge of being with the participants is great however, it is easy that the data is not specific enough and there is potential risk that the data becomes distorted in the interpretation process. This method also is highly qualitative and qualitative data is harder to process and collect on a larger scale.
Robinson C. et al. [1] developed a user centred design with six steps and user participation at all time. It involves Work Domain analysis, Conceptual Development, Prototyping, Interaction Usability Studies, Implementation and Debugging. The process is suitable for the objective since it will involve prototyping of base maps. Roth E. et al. [18] presented a reconstructed version of the methodology where the prototyping would be the first step followed by Interaction Usability studies. However this approach would not be beneficial in this case since user data is needed before the process of prototyping.
3.2 User-Centred Design
Robinson C. et al. user-centred design is a six-stage design process.
1. Firstly a Work Domain Analysis is carried out, conversation with the user regarding how the work element shall be constructed.
2. Further the Conceptual Development is performed, identifying important features and design aspects.
3. Prototypes are then made based on previous steps.
4. Interaction and Usability Studies are then performed to test the prototypes.
5. Then implementation of the final product is made.
6. Lastly there is debugging of the final product. This is done while having user participation at each stage.
This study will not include stage 4 and 5 due to time restrictions. Implementation and debug- ging of the prototypes takes a couple of months.
Figure 5: Robinson et al. six-stage user-centered design process.
’
Following text will explain each step. However, how the case study will be implemented into this research strategy is explained in Section 4.1.
Work Domain Analysis According to Robinson. C et al. [1] this phase represent the general ideas of the participants and the developer in conduct of the thesis. This can be done through interviews, survey or other forms of conduct.
Conceptual Development According to Robinson. C et al. this step in the process sums up and outlines what have been studied in th Work Domain Analysis. The idea for the prototypes are constructed as well as layout and architecture design is drawn. This part can be iterative after getting feedback from the participants as opposed to the Work Domain Analysis.
Prototypes At this stage fully working prototypes are created.
Interaction and Usability Studies This stage is crucial since this is the part where the par- ticipants interact with the prototype and the designer retrieve feedback. This is done with the presenting of the prototypes. Often done through interviews or focus groups. Robinson. C et al.
recommend that there could be informal assessment since the end-users are asked to "try out" the prototypes.
3.3 Identification of Influential Factors
There are some important influential factors that have to be acknowledged when collecting data with this method. Since this method is focused to investigate the realm of web maps it will be seen from different devices. This could be computers, tablets and phones. This will affect view scopes and responsiveness and also size of device. It is of high importance to identify these factors. Another important factor is the environment the platforms is used in. Some phones will be operated outside and which will change light perception of the device and hence the map. As well, it is of high importance to identify that the participants are using different settings on their screens. This will greatly effect what hue, lightness and chroma the participants will prefer. A
certain light setting on their screen can be totally different on another screen. This is difficult to predict and impossible to change, as well if the research was carried out on the same device the end-user would not work in the same conditions and on that device. Hence the research will be carried out in their own environment and on their own devices, which will be in different sizes.
3.4 Choice and Validation of Methodology for Data Collection
The data collection will be carried out in the procedure of three surveys through the service Google Forms[8]. This is beneficial since it is easy to carry out and construct. As well, it is easy for the participants to undertake since it can be done anywhere with an electronic device with connection to the internet. Lastly it enables easy and fast access to the results.
The surveys have the user to choose between alternatives to make the data collection quanti- tative. Since quantitative data is more easy and decisive to interpret and also more objective then qualitative data collection. This is of high importance since map design can be a highly personal matter. Hence quantitative research is beneficial to identify common factors. As well, quantitative data allows data processing and more tools to process the data.
Qualitative data is also collected as compliment. Since it is a user-centred design process with surveys it is of high importance to acknowledge that the constructed forms can direct the answers and select which domain shall be discussed with which alternatives. Qualitative sections will aid that problem and let the participants direct the research to issues which is not in the realm of the researchers grasp. As well, it functions similarly to a outlet for the participants where they can express there opinions and give feedback. Additionally, as previously stated Robinson. C et al.[1]
recommended to have an informal part in the Interaction and Usability Assessment part, which one can argue the qualitative part will represent in this study.
3.5 Risks and Undertakings with Research Strategy
There is certainly risk with chosen approach. One of the biggest risks is the risk of not being able to collect any data due to users not responding. This is a risk with any user-centred design process which relies on individuals to answer in some way. In this case there is also a risk of participants dropping off to a big extent between the surveys. To address this issue a few procedures have been adapted. Firstly the first conduction is personal. There is a big risk of the participants not feeling selected and that this is massive dispatch, hence resulting in participants not answering.
As well, it is important for the information in conducting phase to be short and concise. Surveys shall as well be rather short and easy to understand and complete. If the surveys are too long the participant can become impatient. In order to make as many participants to prosecute with the case study only 3 surveys are conducted. It will limit the number of iterations of prototypes however it will increase the chances of participants not dropping off. Questions will be kept simple and basic. There is no need to ask about very specific information and details since this tend to complicate both the structure of the questionnaire and also the interpretation of the data. Lastly technical problems can occur since it involves visualization of map prototypes and the user being able to operate in the environment.
4 Data collection and Implementation of Methodology
This section explains how the data has been collected and how the data collection was implemented in the research strategy.
In this thesis the components of the colour will be mentioned as hue, chroma and lightness since it is applicable to the main working program of Mapbox and is comprehensive. Working within the colour system of RGB is not preferable since the value of red, green and blue does not give the reader a comprehensive understanding of how the elements looks.
Hue for different map land cover types was set similarly to the ink colour and screen tint standards for features on U.S Government maps and other hue conventions described by Robinson et al. [3] mentioned in 2.1 Definitions and Concepts of Map Design.
4.1 Case Study
The target group for the case study consisted of people already working with GIS platform (product from Digpro) to manage a utility network system. This means that all of them work with some sort of base map and hence it is likely they are familiar to the concept of base maps. However, they work within different fields of utility networks. The main domain of work fields that is targeted is heating-, water-, power-, fiber- and gas management. All which include managing an utility network.
During these surveys the participants operated within prototypes displayed on a web page.
Hence they were free to operate at any area and any scale. However, the starting areas was set to Stora Essingen, Stockholm. Since Digpro had test data of that area which helps the users to evaluate the base map in a realistic area.
4.1.1 Survey 1
There was 126 participants in the first survey. The first survey consists of three parts.
1. Information About the User.
2. Colours.
3. Information and Symbols Presentation.
Information about the user is necessary in order to find correlation in preferences further dis- played in two next sections. Colours, information and symbol presentation is a part of the three important components of map design stressed by Robinson et al. [3] colour, pattern and typogra- phy.
Information About the User The information about the user is kept short and basic with only four questions. The participant is asked within what field of utility networks they work with in order to identify difference between the different fields of utility networks. Secondly the participant are asked where they work with the GIS system (out in the field or at the office). Third what platform they work on mostly and fourth if they are colour blind.
Colours A website was developed and published as part of the colour assessment in the form.
This was done solely with Javascript, HTML, CSS and did not include any database management.
The website included a interacting test where the user was presented a basic OSM map with overlaying electric utility network. The data of the network was test data from Digpro used as real demo data for clients and very similar to the utility network that the participants are use to work with. The demo data that was chosen contained a network displayed in red. This was chosen due to Robinson et al. [3] stating that red, orange and yellow appear as being closer then other colour hence the network will be seen as overlaying feature. The user could then change the outcome of the colour representation of the base map through changing two parameters, lightness and chroma.
Where both parameters set to 100 percent represented a standard colour representation of a OSM.
The user was not permitted to change hue. The reasons for this are there is little reason to let the user change the standard hue conventions with small exceptions. This is argued by Robinson [3]. Robinson claims that blue is the one universal colour for water bodies and rivers. Robinson argues that other colour conventions are less strict and can vary from culture to culture. However mentions that in the western culture following hue conventions are used:
• Green for vegetated areas
• Brown for land surface symbols such as contours
• Yellow or tan for dry, sparsely vegetated areas
Hence letting the participants change the hue parameter would result in all land covers to change hue in regard to each other. With other words the participants could violate these very basic principles. Hence the parameter was not included to create less distractions and focus on the lightness and chroma of the map.
The reasoning behind collecting this very specific data is that as mentioned earlier in related work by Nadia H. Panchaud and Lorenz Hurni [14] which developed a formula reducing saturation and changing the lightness of the base map to enhance the featured layer on the map. In the paper it is mentioned that the goal of the calibration is to avoid grey colours, keep enough chroma to be able to distinguish between different hues and avoid greyish colour schemes. This test address this very issue and is aimed to determine that range of a values for chroma and lightness.
Below in Figure 6 a screenshot from this test i shown. As seen the lightness parameter varied from 0-150 and the chroma from 0-200. There was no point of having an option of choosing lightness above 150 since the map went completely white at this kind of values.
Figure 6: The specific website where the participants could select a certain chroma and lightness
4.1.2 Survey 2
There was 55 participants in the second survey.
Since the first survey was lacking any type of prototypes the participants was limited to visu- alisations of maps in their questionnaire. Hence, the second survey consisted of several prototype pairs which was identical in all regards except one category. This was done simply in-order for the user to be able to choose between one of the prototypes and only being influenced by the variable of interest. Hence, when looking at the result preferences could be limited to one factor. The three categories are:
• Information Quantity
• Degree of Map Components in Colour
• Amount of Chroma and Lightness
The first two pairs which is different in the information quantity differ by one having road number for bigger roads, displaying more symbols and names of parks and institutes. In this section it there will also be question regarding having contour lines since this was requested from the qualitative data.
The second pair which differ in degree of map components in colour differ by one having coloured areas for hospitals, schools, bigger roads, industrial areas. Both having colour for water, green areas and buildings.
The third pair which differ in the amount of chroma and lightness. Where the one prototype will have a higher chroma setting of about 80 percent chroma where 100 percent is maximum. And the other prototype having a chroma between 33 and 50 depending on land cover since different colours appear different at the same chroma level as stated by Lafay et al. [19]. The lightness was set between 80 and 90 percent for both prototypes which was about the estimated lightness collected from the first survey.
In survey 2 the user was able to navigate in these prototypes in order to experience generaliza- tion and contrast on different zoom levels. The idea was that the visualization would give them a more in-depth view and increase the accuracy of there preferences since it can be hard to visualize information without any visual representation.
4.1.3 Survey 3
There was 48 participants in the second survey.
Survey 3 was ultimately a review of the iterated prototypes in-order to measure the satisfaction of the participants. The survey was kept short and basic. The participants were asked to rate and review the prototype. A basic OpenStreetMap was used as a comparison.
A final prototype was constructed. The final prototype was based on Survey 1, Survey 2 and the literature. The information that was displayed on this final prototype was: road names, address number, hospitals, place of worship, viewpoints, town halls, national parks and nature reserve. Following areas was highlighted in colour: water, green areas, hospitals, roads, schools and buildings.
The different areas was divided into primary features and secondary features. The primary features are water areas, parks and roads that was identified in survey 1 as more interesting features for the participants. Had the setting of chroma at 32(except roads) and lightness at 75.
However, the road which is in yellow and orange had a chroma setting at 70 since these colours appear grey at the chroma value of 32. The secondary features are buildings, hospitals, schools and road segments under ground level. These were set to chroma at 32 and lightness at 85. Land was assigned its own category since it only serves as canvas to display the other colours on and shall take as little attention as possible. It had chroma at 12 and lightness at 92 which it appear as a grey hue.
4.2 Data Collection Integration
As seen in Figure 7 the study started with the Work Domain Analysis. The Work Domain Analysis was based heavily on Survey 1. When the data was collected Conceptual Development started to form the prototypes. The prototypes was then tested with Interaction Usability Studies which was firstly done with Survey 2. This was followed by one iteration where Conceptual Development was formed again from the result of Survey 2. Prototyping was conducted again and the final Interaction Usability Studies was performed with Survey 3. Further, the prototypes was finalized.
4.3 Data Processing
The data from the surveys is processed with known measurements such as:
Standard Deviation =
qP(X−X)2
n
Median = (n+1n )thterm
Average =
P10 i=1xi
n
Figure 7: Robinson C. et al. research methodology combined with the three implemented surveys.
[1]
As well data was processed through divided subset in regard to the users preferences and users profiling, line of work, if they are coloured impaired etc. Which allows for a more in-depth analysis.
This was done by importing google form answer to Excel.
4.4 Prototyping
The prototypes are constructed through OSM data. The OSM data is free to use and crowd source founded. There is a few programs which simplify the process of manipulating the design of the OSM tiles. Two of them are Mapbox Studios [11] and Maputnik [13]. Both free to use, however Mapbox Studio have a limitation on amount of tile loading. In this case Mapbox Studio was chosen since it has its own graphical language the Mapbox GL JS [12]. This is important since the prototypes will be displayed and operated in web solutions including HTMl, JavaScript and CSS.
The prototypes are made with regard from the surveys and the literature. Mapbox and Maputnik allow the users to design the base map in specific details such as deciding on what zoom level information shall be shown and how it shall be displayed. As well, colour can be exactly specified in the HSL (Hue, Saturation, Lightness) system. Each set of information layer can be designed specifically and with the use of classes and types one can restrict or allow information in each layer.
Both programs create a JSON styling sheet which can further be implemented to style a tile server with OSM data. The page showed the base map prototype as well as a vector layer displaying a test utility network. To be able to evaluate the base map in a close to a realistic working scenario.
5 Results
Following section will present the results and findings of this study.
5.1 Survey 1
5.1.1 Personal Information
Majority of the participants worked with electrical networks at 63,3 percent, followed by fiber at 25 percent. Majority of the participants worked mostly at office at 78,1 percent, as well a big majority mostly used computer as their main device when working at 89,1 percent. 3,1 (four participants) percent of the participants claimed to be colour blind. These groups was categorised, however no apparent pattern was evident. Hence there will be no data displayed.
5.1.2 Colour
91,4 percent of the participants preferred having a bright background colour over having dark base colour. Of those who chose a bright background colour 90,9 percent preferred having map type elements in colour (not in grey scale) and 88 percent of the participants choosing a dark background. 13,2 percent of the participants preferred having grey scale instead of having some map type elements in color.
When asked what land covers the participants would prefer in colour, the two highest selected alternatives was water 61,9 percent and green space 47.8 percent. 27,4 percent selected no land cover shall have colour or some range of chroma, hence black/white map. This question was only asked to the people who preferred having some land covers in colour/some value of chroma.
Following test data is in relation to standard OSM where its standard colours are displayed at value of 100 chroma and 100 lightness.
As seen in Figure 9 the majority of the participants preferred a lightness between 90 and 110.
As seen in Figure 10 the data for selected chroma values is almost evenly distributed across the range. There is a clear peak at the interval of 100 to 125. However, it is worth noticing that 24 out of 112 that answered this question (other 14 participants not being able to see the content due to technical issues or prohibited by their company) answered 100 which is 21,6 percent of the answers. It is not normal distributed since it increases on both sides after a decrease. Table 1 shows that there is clear difference in selected values for chroma and lightness between the grey scale preferring group and the colour preferring group. Aswell, there is higher correlation between lightness and chroma for the preferring grey scale group compared to the colour preferring group.
This is visualized in Figure 11. 17,9 percent of the participants chose a chroma setting of 150 or higher, which is a high chroma setting. Below in Figure 8 one can see the result from what landcovers the participants liked having in colour, where water and green areas was the most popular.
Figure 8: Result from the question "What land covers would you like to have in non-grey scale (with some range of chroma). From top to bottom: Water, Green areas, Fields, Schools/Hospitals/Institutions, Property, Big roads, Non.
Table 1: Table showing average, median and standard deviation from the web map test. Where it is divided into three groups. Participants preferring having map type elements in colour, participants that prefer grey scale and all participants combined.
Basic Statistics
Colour Category Average Median Std Dev
Chroma(Colour Only) 105,2887 100 36,3719
Chroma(Grey Scale Only) 67 50 46,8746
Chroma(Colour and Grey Scale) 100,4505 100 39,7045
Lightness(Colour Only) 90,5567 95 14,8527
Lightness(Grey Scale Only) 83,1429 90 24,0763
Lightness(Colour and Grey Scale) 89,6216 93 16,3446
Table 2: Table displaying correlation between chroma and lightness divided into the different groups.
Correlation
Correlation Variables Correlation
Chroma and Lightness (Colour only) 0,0136 Chroma and Lightness (Grey Scale) 0,3077 Chroma and Lightness (Colour and Grey Scale) 0,0793
As seen in Table 2 the correlation is low and no correlation near 1 or -1 which is perfect linear correlation.
Figure 9: Histogram displaying preferences for Chroma, both colour and grey scale preferences.
The histogram is set to a bucket-size of 10 and divergent percentile to 0 percent. 2 outliers were identified and removed at values of 34 and 120
Figure 10: Histogram displaying preferences for lightness, both colour and grey scale preferences.
The histogram is set to a bucket-size of 25 and divergent percentile to 0 percent.
Figure 11: A. Average Chroma and lightness participants preferring colour, B. Median Chroma and lightness participants preferring colour, C. Average Chroma and lightness participants preferring black/white. D. Median Chroma and lightness participants preferring black/white.
5.1.3 Information and Symbology
In the question asked “What information do you find necessary in a base map?” where the partic- ipants could select multiple alternatives the most selected alternatives was bigger roads at 85,9 per-
cent, address numbers at 82,8 percent, smaller roads at 76,6 percent, Big parks/hospitals/institutions at 57 percent. As seen in figure 12. When asked about information representation of parks, insti- tutes etc. 40,6 percent preferred symbol and text, 55,5 percent preferred only text and 3,9 percent only text.
Figure 12: Number of votes when asked: What information do you find useful to show in a base map? From top to bottom: Big roads names, Small roads names, Commuting information, Road direction, Shopping Entertainment, Bigger parks and institutions, Smaller parks and institutions, Address number, parking area, gas station and lastly none of the alternatives.
5.1.4 Qualitative Data
Last question of the survey presented qualitative data. There was some mixed opinions expressed.
However, one general concensus appear to be coincidental with multiple participants. Many par- ticipants mixed the concept of the base map and a map with multiple layers. They expressed the need of being able to filter out or add information of the base map on the fly. With other words to work with the base map in multiple layers which enable the user to select what information shall be shown. One user says “The most important is that the base map is based on layers. That every information have a correspondent layer”. Another person says “One idea could be to have some sort of quick option where I could minimize information in the map to make panning faster and later being able to select more information when I want to look on the map in smaller scale”.
Another user stressed that the matter of map information and colour is highly personal and also stressed the option of being able to choose while working. As well property number was heavily requested by many participants to be displayed as information in the base maps.
5.2 Survey 2
Bare in mind that the following prototypes where operate by the participants at any scale and area of there liking. Following figures are screenshots from these prototypes.
5.2.1 Information Quantity
In Figure 13 the two prototypes are depicted which shows the one with more information on the left and the one with less information displayed on the right. 67,4 percent of the participants preferred the prototype with more information displayed, while 32,6 percent preferred the one with less information.
Following data was collected from the question regarding having contour lines in the base map.
11,1 percent of the participants would like to have it as a part of the base map, 7,4 percent would not like to have it and 81,5 percent would not like to have it as part of the base map however consider it as layer that would be able to turn on and off.
Figure 13: Depiction of two screenshots from the prototypes with different amount of information displayed and a overlaying network in red. The one on the left with more information displayed such as road signs, hospitals, parks, landmarks etc.
5.2.2 Degree of Map Components in Colour
In Figure 14 the two prototypes are depicted where the one on the right have more map components in color and the one to left have less map components in colour. 69,2 percent of the participants preferred the prototype with more map components in colour while 31,8 percent preferred the one with less map components in colour.
Of the 32,6 percent that preferred the prototype with less components in colour. 68,8 percent preferred having the same amount of components in colour, 18,8 percent less components in colour and 12,5 percent preferred more components in colour.
Of the 67,4 percent that preferred the prototype with more components in colour. 69,4 percent preferred having the same amount of components in colour, 30,6 percent would have preferred more components in colour and 0 percent would preferred less components in colour.
Figure 14: Depiction of two screenshots from the prototypes with different amount of map type elements in colour and a overlaying network in red. The one right with roads, school areas, airports and hospital areas in colour
5.2.3 Amount of Chroma and Lightness
In Figure 15 and Figure 19 two prototypes are depicted where the right prototype have higher value of chroma and the left lower value of chroma. 20,8 percent preferred the prototype with higher chroma and 79,2 percent preferred the prototype with lower chroma.
Of the 20,8 percent that preferred the prototype with higher chroma 45,5 percent would keep the same intensity, 36,4 percent would decrease the intensity and 18,1 percent would increase the intensity in the colours.
Of the 79,2 percent that preferred the prototype with lower chroma 69 percent would keep the same intensity, 26,2 percent would like to have decreased intensity of the colours and 4,7 percent would increase the intensity of the colours.
Figure 15: Depiction of two screenshots from the prototypes displaying two different sets of chroma values and a overlaying network in red. The prototype on right with higher chroma.
Figure 16: Depiction of two screenshots from the prototypes displaying two different sets of chroma values and a overlaying network in red. The prototype on right with higher chroma.
5.2.4 Qualitive Data
Some participants claimed they could not open the web site which provided the prototypes. The participants mentioned that their companies did not allow them. Some participants expressed that it can depend a lot in what branch your working and hence information shall be provided through a filter option which allows to turn on and off the information. As well in the section of degree of map components in colour participants expressed that it could be very situational how many colours shall be on display.
5.3 Survey 3
In survey 3, 80,4 percent of the participants answered that they preferred the final prototype over the OpenStreetMap as a base map for a utility network and 20,6 percent preferred the Open- StreetMap. The participants were asked how well the prototype fulfilled there requirements for a base map on scale of one to five where one was not at all and five very well. The map got rated 3,98. 93,6 percent would consider having the prototype as base map in there application today and 6,4 percent would not.
When it came to the qualitative question, there was mixed answers. Several answers however, mentioned issues regarding the roads. Expressing the limitation of roads in rural areas. Where some where not visualized at all or that the road section could be more clearly visualized. Another participants expressed satisfaction regarding the prototype however expressed that it needs to be several base maps for different purposes.
5.3.1 Final Prototype
Following figures display some screenshots taken of the final prototype. Which can be compared with the standard Open Street Map in Figure 6.
Figure 17: Final prototype at zoom level 12.57, depicted area is Stockholm, Sweden
Figure 18: Final prototype at zoom level 13.62, depicted area is West Stockholm, Sweden
Figure 19: Final prototype at zoom level 15.66, depicted area is Stora Essingen, Sweden
6 Discussion
This section will analyse and discuss the result of the study. Three main points will be discussed;
the implementation of user-centre design, the comparison of the case study and the literature and the prototypes.
6.1 Implementation of User-Centred Design
The implementation of the user-centred design was implemented succesfully, even though in this study it was adapted to more quantitative approach. However, more iterations would be desirable to obtain a more validated prototype. Since more data could be collected and further analysis of the final prototype could be done. As discussed in Section 3.5 it would have been hard to obtain the focus of the participants in such long process. In addition it would put immense pressure on the time schedule with a time constraint of five months. It was challenging to apply the research technique of user-centred design on the matter of cartography since it is such a highly personal matter. It was a challenge to stay objective and not direct the research too much. Of course it is influenced by the researcher. However, in this thesis it was attempted to be as low as possible. Another challenge was to not over weight the qualitative data. The qualitative data suggest improvements and changes. However, it is important to bear in-mind that these suggestions are from specific individuals which the majority of people may or may not agree with. Another challenge was that the participants seemed to be more interested to discuss the whole GIS platform rather then only the base map. This is of course natural since the base map will be heavily complemented by other layers and feature. However, it can delude the result and creates distractions and mistakes in the study when the participants keep confusing the concepts. This continued to happen through out the surveys even though it was addressed by informing the participants of this fact in survey 2 and survey 3. As identified in Section 3.5 participants did drop off during the case study. In the first survey 126 participants answered and in survey 2 55 participants answered. This could be due to many reasons. However, one of them was clearly that some participants were not allowed to open the internet links which displayed the prototypes. Survey 2 was solely built on the participants being able to open these links, hence participants with this dilemma could not answer. Other reasons could be lack of interest, survey 2 and 3 not having personal conducting methods (since it was sent out to all people who answered survey 1) and more.
6.2 Optimal Base Map Design
The matter of what is an optimal base map design can still be disputed. However this thesis can recommend some good practises. Previous literature recommend the base map to be in lower chroma, however not to the extent where the colours are not distinct-able. From the case study this seems to be something that can be agreed upon. A general good practise seem to be in the range of having the chroma setting between 30-50 percent of the maximum intensity and a lightness setting between 75-90 percent of the maximum intensity. However, there is certain percentage that prefer stronger colours and other that like grey scale. As stated in the result about 10 percent of the participants prefer black or dark base colour. About the same ratio prefer having a base map in complete grey scale. On the other side of the spectrum 17,9 percent preferred having a value of 150 or more chroma in the colour test where Standard OSM is at value 100. This amount of chroma present very vibrant colours and 17,9 percent is presentable group of people. This is somewhat surprising since it the colours will have the same amount of intensity as the network which speaks against the literature. However this seem to be highly effected on what screen an individual are using. Since they will differentiate in both chroma and lightness. This was seen from standard deviation which showed a wide spread of the answers. However as showed in the litterateur a lower lightness will save battery. It shall be acknowledge that from the case study performed there is a certain group that prefer colour settings outside the norm. The colour test did not show any significant correlation between the participants preference of chroma and lightness with Pearson correlation coefficient of 0,0136 for all participants. However, when divided in to groups of colour preference and grey scale preference. It is apparent that there was some correlation in the group that preferred grey scale the correlation coefficient was 0,3077 which is distinctively higher.
However, it shall be noted that this group was represented by 13,2 percent which corresponds to rather small group of participants and lower the legitimacy of that answer. There seem to be no apparent correlation in the survey for the working field, working environment or preferable device.
