Conclusion

In conclusion, the analysis traffic changes in the canton of Geneva until 2050 reveals significant trends and patterns as forecasted by the National Traffic Model of Switzerland. The following sections will delve into the detailed findings related to Research Question 1 and 2. The first section will be focusing on the specific changes in rail and road traffic across different areas of the canton (research Question 1). The second section will focus on the change related to each of the municipalities in Geneva (research question 2).

The rail network in the canton of Geneva is affected by many changes. The frequency of use is increasing, especially in the northern part of the canton up to the city center. The rail network along Lac Leman, across the French border, is particularly affected. Usage of the rail network in the city center is also expected to grow. A single decrease is to be expected in a small section of the network in Sécheron-Prieuré. The growth is greatest in the city of Geneva, particularly on the rail lines running south through the city, with a forecast growth of over 1000%. However, no clear change in traffic behavior in terms of usage can be seen across all rail sections. The values of the type of use mostly increase proportionally.

The road network in the canton of Geneva is undergoing major changes. For the most part, greater road use is to be expected. The range of percentage changes is very wide. In the city center, the more heavily trafficked roads have mainly seen growth of between 0-64%. Neighborhood streets (many of which currently have little traffic) can expect a decrease. This means that in the city, primary growth can be expected on the already heavily frequented roads. In rural areas and on the outskirts of the city, very high increases are to be expected in some cases. However, these increases are mainly due to the originally low level of traffic. In the rural peripheral areas of the canton, declines are forecast in some cases. However, no clear change in traffic behavior in terms of use can be seen across all road sections. The values mostly increase or decrease proportionally.

To answer research question 1, one can clearly see patterns of change in the canton of Geneva, with a clear increase in traffic for rail and roads.

For the road traffic in the municipalities, it can generally be said that in the more densely populated urban areas of the canton, especially in the city of Geneva itself the traffic change is smaller than in the more rural areas. As it can be seen in the Atlas statistique du canton de Genève [2]: Four of the five municipalities with the lowest population density have a change in traffic volume of +20% or less (the exception being Vernier with +26%), while all the municipalities with a change in traffic volume of 38% or more have a low population density.

For the Rail traffic it’s more the urban areas that have a bigger positive change in traffic volume. But for the railway it can be said that most municipalities have an increase, with two exceptions: On one hand the municipalities on the south-eastern line from Geneva to Annemasse (France) have a reduction in traffic volume and on the other hand the municipalities on the line to Bernex have also a negative change. According to the SBB online timetable the route from Bernex to Geneva is only served by a single tramline [3]. So, it can be assumed by looking at the maps that the municipalities along the lines from Geneva in direction to the rest of Switzerland and from Geneva in direction to Lyon and Paris experience a increase in traffic volume, while the municipalities along the other lines experience a decrease. But the strongest increase in volume is in the densely populated urban municipalities of Lancy and Carouge, which are not located on one of the two lines. From our map it is not possible to tell what the reason for the high numbers might be there.

Growth of the railroad

In the canton of Geneva there is a pronounced growth of the railroads. Growth is particularly strong in the city of Geneva, with the rail lines running south through the city being particularly affected.

Shift in road use

The city center has mainly experienced growth of between 0-64% in the more heavily trafficked streets. Neighborhood streets in the city will see a decrease. Less frequented roads in the countryside will also become busier in the future.

Less growth in urban areas

Municipalities with high population density experience a weaker increase in road traffic volume than municipalities with a low population density.

Increase along major rail lines

Municipalities on the lines connecting Geneva to the other Swiss cities and to big French cities like Lyon and Paris generally have an increase in traffic volume.

Methods

In the following section, we can see the methods used by our group to produce our final project.

Topic Search

Internet, Books, Articles, Journals

In the first step, suitable topics were searched for. Various sources were consulted, and analyses were conducted to identify relevant areas for the project in regards to the UN Development goals.
Data Retrieval

Online (see Sources)

The data was gathered from various online sources (see here). It was imperative to ensure comprehensive data retrieval to support the project's objectives.
Preprocessing

QGIS

The data was preprocessed in QGIS. All the necessary information and graphs were precomputed to ensure smooth operation of the webmap with such a large amount of street data. The image above provides insight into the preprocessing step using the integrated graphical modeler in QGIS. Data was initially gathered from various sources, then joined and clipped to create one comprehensive dataset. Once completed, all statistics were precomputed and exported as a CSV file.
Preprocessing, Analyze

R

After preprocessing in QGIS, the data was summarized into 8 separate CSV files, each for communal and cantonal levels. These CSV files were then utilized in the creation of the webmap.
Website and Map Sketch

Paper and Pen

We chose the R Shiny framework to create our interactive web map. We tried to replicate the initial sketch of the web map and adapted the map accordingly. Potential problems were carefully discussed within the group.

Want to see the sketches? Click here.
Map Creation

R Shiny / Leaflet

We opted for the R Shiny framework to produce our interactive webmap. We endeavored to replicate the initial sketch of the webmap and adjusted the map accordingly. Potential issues were carefully discussed within the group.

Want to know more how we designed our map? Click here.
General Information

Our map is based on the transport model of Switzerland 2050, with the focus of the map on the canton of Geneva. An important reason for this decision is the location itself. With its small size but high international appeal, Geneva is constantly subject to change. Geneva is well suited for a spatial analysis of traffic development due to its clear size. Our map has a very specific target group, i.e:
- Spatial planning offices
- Office for Road and Shipping Traffic (Canton of Geneva)
- Federal Department of the Environment, Transport, Energy and Communications
With the map it should be possible to successfully answer our research questions, which are as follows:
1. How and where does Rail and Road traffic in the canton of Geneva change until 2050 according to the National Traffic Model of Switzerland?
2. What are the key differences between various municipalities in terms of traffic development?
The two layers "Rail" & "Road" are used to answer RQ1. They visualize the change in traffic on the respective road section. A good visualization should make it possible to identify significant changes and, if present, to recognize spatial patterns.
Map Layout

The aim of the map is to do justice to our target group and research questions. Spatial patterns play a decisive role for us, so the map takes up a lot of space, it is the heart of the design. On the left-hand side there is a column with additional (more precise) information. One can select the desired layer to be displayed on the map. In addition, the two plots (Graph 1 and Graph 2) provide further information on the selected sections. The plots should react dynamically and display the information on selected sections.

Design Choices

The main strengths of our design are clarity and the reduction to important and essential information. The map card reader should immediately understand the intuitive operation and receive the essential information. The focus is on the relevant data and not on detailed data analysis.

Design of the "Municipality" Layer

The Municipality layer is intended to answer research question 2 and therefore the data is aggregated into the 45 municipalities of the canton of Geneva. In a first step the user can choose with radio buttons whether the municipal statistics for roads or railways should be displayed. If not mentioned otherwise everything in about the municipality layer is applicable for both the road and railway statistics.

The starting view

The starting view of the municipality layer should provide an overview over the municipalities of Geneva. On the map all the municipalities are shown and it displays how much change happened in each municipality. By hovering over the respective municipalities a pop-up displays the name of the municipality and the exact and the exact rate of change. In the left panel a bar plot shows the traffic volume in 2017 and 2050 for each municipality.

Municipality View

By clicking on a municipality, it gets highlighted and the map zooms to the selected municipality. In the left panel the bar plot now shows the traffic volume in 2017 and 2050 by usage zone in the respective municipality. We decided to abbreviate the usage zones in the plot, because otherwise the long name compressed the plot and made it hard to read. When a municipality is selected we therefore added a text field below the plot to explain the abbreviations. For the names of the usage zones we are aware that a mixture of German and English is not ideal, but while trying to translate the zones we came to the conclusion that an attempt to translate them, could make it harder to understand what is meant by them. Because we assume that the majority of the people who will look at our map will know some German, we therefore decided to stick with the German names.

Originally it was possible to change between the railway and road statistics in the municipality view without resetting the selection, which would have allowed an easier comparison of the two plots in one municipality. But this led to problems when switching from the municipality to the Rail or Roads layer, therefore we decided to remove this.

Classification

As classification for the positive values on the map we chose national breaks (Jenks) which tries to group the most similar values together, while maximizing the differences between classes [1]. The calculations of the classes were originally mad in R, but to simplify our code we chose to take the calculated breaks and just use them instead of dynamically setting them each time the input is changed. For the road statistics we decided to start the legend from the value 12, because there are no values lower than that. This way we stuck with the exact class ranges calculated before. For the rail statistics the positive values would have started at 15, but because there are negative values in the rail statistics, we decided to extend the lowest positive class down to zero.

For the negative values, which are only present in the rail-statistics, we chose to just group them into a single class. This decision was made, because there are not many negative values in our dataset.

Color scheme

As a color scheme on the map a sequential single hue color scheme with different values of blue was chosen. This color scheme was chosen from the Colorbrewer website, where it was categorized as colorblind safe. For the negative values, which are only present in the rail-statistics, a red color was picked. We also made sure with the Color Oracle that all the colors chosen in our map are colorblind safe.
Design of the "Rail" and "Roads"

The design of the two layers "Rail" and "Road" is almost identical due to the same focus on (Research Question 1). They should give the card map reader an immediate feeling of belonging together.

Data Structure

The two layers are line data.

Dynamic Zoom

Fig: Desired size of lines at a resolution of 1:2’500
We were unfortunately not able to adjust the size of the lines dynamically based on the zoom level. The figure above shows the optimal size of the lines when zooming closer to a map part. In our case, when the zooming in, the lines are appearing a bit too small.
Classification

The percentage change in traffic is displayed on the map. Negative and positive values occur accordingly. There are 4 different classification options to choose from:

Natural breaks

Equal interval

Quantiles

Mean standard deviation

Before one can make an exact classification, you have to look at the distribution of the data. The histogram of the "Rail" data has few negative values and shows many positive values in the range from 0% to approx. 200%. Followed a few values that are very high and go up to almost 2000%. Based on this data distribution, we decided to use natural breaks (Jenks) for all the positive values. It allows the most "natural" grouping of the data. Due to the few negative values and the problem that natural breaks can create a class with positive and negative values, we decided to manually class all the negative values into one class. This allows the clusters for the low values to be better emphasized and larger outliers to be better summarized. It will also be important to visualize the transition between negative and positive values with the right color (more under Visual Variables -> Color).

The data for the roads is distributed similarly to the rail data. However, it has many more data points and a more consistent distribution in the lower part. It has few negative values and many values that are slightly positive. The tail on the right indicates a few very high values. Due to the similar distribution as with the rail data, natural breaks (Jenks) are used again for the positive values. It is easy to distinguish between positive and negative values, peaks are highlighted and the few very high values are grouped together in a class.
Number of Classes

There are several possibilities for the number of classes, we also want a differentiation in the near range (distribution slightly negative to slightly positive, with peak). We have therefore opted for 5 classes. As a rule, 7 (+/- 2) is the best number of classes [1]. For our classification, 5 is best as it highlights the clear differences in the data without appearing confusing.

Visual Variables

The size of the variable would be one way of displaying the change (higher values = larger, lower values = smaller). The main problem with this method is the lack of clarity. With this method, spatial patterns can no longer be easily recognized, negative values are almost no longer visible and higher positive values are displayed too dominantly. Due to these problems, we have decided against this method.

The value of a variable/class can be displayed with the appropriate color. Since we have a classification and no continuous dating, we can use a two-level color gradient for the classes. The two colors are used to differentiate between positive and negative values. Once colors are involved, it is also important to find a suitable color pattern. Colors can lead to false associations, which must be avoided. Color blindness also occurs again and again, so it is important to refer to it in order to avoid complications. A good solution for this is a website called "ColorBrewer", which helps with the right color selection and takes into account the number of classes, color blindness, etc. Thanks to the "Colorbrewer", we decided on the following color gradient:

© Cynthia Brewer, Mark Harrower and The Pennsylvania State University

The value of a variable/class can be displayed with the appropriate color. Since we have a classification and no continuous dating, we can use a two-level color gradient for the classes. The two colors are used to differentiate between positive and negative values. Once colors are involved, it is also important to find a suitable color pattern. Colors can lead to false associations, which must be avoided. Color blindness also occurs frequently, so it is important to refer to it in order to avoid complications. A good solution for this is a website called "ColorBrewer", which helps with the right color selection and takes into account the number of classes, color blindness, etc. Thanks to the "Colorbrewer", we choose the color gradient visualized below. That color scheme has also been checked with “Color Oracle”.
Basemap (Background)

For the background map, we have opted for the official swisstopo map in black and white. The advantage of this map is the information content and presentation. The surveys are accurate and up-to-date, and our target group is particularly familiar with the official surveys. In addition, the background map has a good synthesis with our layers and shows the important things for geolocation without being overloaded and distracting from the main content. The black and white color selection should contribute significantly to this, it allows for a greater color selection of the layers and ensures fewer potential problems due to the color.

Additional Features

The maps have many additional features that are worth mentioning but do not fit into an additional section due to their size. These are the following features:

Box for operating assistance

Each layer has an information box. This provides information on what the map user needs to do. This increases user-friendliness and helps inexperienced/new users to find their way around more quickly.

Limitation of map displacement

The area of the survey is only in the canton of Geneva, but the whole of Switzerland is shown on our map for orientation purposes. However, the other countries are not visualized. As a result, the map user can quickly get lost if they click incorrectly and can no longer find their way back to the desired map section. To avoid this, we have decided to limit the map displacement. This makes it impossible for the user to get lost in the map.

Zoom limitation

We have decided to limit the minimum zoom. This allows us to limit the zoom options to a reasonable scale. Otherwise, ill-considered zoom changes can quickly lead to disorientation (map no longer visible because zoomed out too far). We can prevent this problem with this option.

Notification about longer waiting times

Longer waiting times may occur with the Road layer due to the large amount of data. This can quickly lead to uncertainty or annoyance on the part of the card reader. To avoid such misunderstandings, a notification message appears stating that the layer will take longer to load.
Website Creation

HTML/CSS/JS (Bootstrap 5 Framework)

Initially, we attempted to utilize existing website templates, but soon realized they were not optimized for mobile use. After conducting internet research, we decided on the Bootstrap 5 framework, renowned as the world's most popular framework for building responsive, mobile-first sites. Fortunately, there are many valuable resources available for this framework.Its main advantage is its pre-configured CSS and JavaScript files. We programmed this website from scratch.

Want to know more? Click here.

Bootstrap 5

Bootstrap 5 is a free and open-source front-end framework developed by Twitter. It provides HTML, CSS, and JavaScript-based design templates for creating responsive and mobile-first web applications. Bootstrap includes a 12-column grid system that adapts to different screen sizes, ensuring responsive web design. It offers various classes to create flexible and adaptive layouts. The framework provides a comprehensive set of CSS styles for typography, forms, buttons, tables, navigation, and other interface components. These styles ensure a consistent and modern look across different browsers and devices. Bootstrap also includes numerous reusable components such as alerts, modals, tabs, tooltips, and carousels. These components are designed to be easily customizable and enhance the user experience. Bootstrap 5 offers JavaScript plugins to add interactivity to web elements.

Website https://getbootstrap.com/

© Bootstrap grid example
Improvement

Feedback, Literature, Lecture Slides, Presentations from our peers

After the initial stages of development, we gathered feedback from our peers and studied the literature, lecture slides and presentations of our peers in order to optimise both the website and the webmap across a range of topics.
Final Product

Go Back to the Map