More

Update OSM data already imported in qGIS

Update OSM data already imported in qGIS


qGIS makes it really straightforward to import data from Open Street Map (there's a good tutorial here). However, despite this process is relatively straightforward, it takes some time to download the data, convert it, select the desired attributes to import, and then make queries to save a new layer.

Since I am working on a project that will take some time to develop, I am concerned about having to repeat steps or losing data if I have to make new queries from OSM to add new data. Do you know if there's any way to automate this process as much as possible.

Let's imagine that I have already downloaded OSM data, created several layers (with their corresponding attributes) and I spend some more time doing started doing some extra work (for example a heatmap). Now Imagine that I want to do the same with a more up-to-date version of OSM which will have more features. If I am not wrong I will have to repeat everything and it will be like starting from scratch. Do you know any way to just update all the layers without having to delete them and start over again? (I was wondering if there's any way to automate the process or create "smart" layers which query an specific layer or any way to update the original database)


Not that I'm aware of (although there may be some ways to automate the steps in that tutorial with a python script). If you wanted to go really high-tech (with a corresponding increase in the setup time) then you could setup a postgis database of the area you are interested in (with the tags you are interested in) using something like Osmosis or imposm and then apply daily, hourly or minutely diffs for it so that your database is kept up to date. This is quite a lot of work and a lot of data though, and it doesn't look like this would be worth it, unless you have to repeat the process above hundreds of times.


Spatially combining wood production and recreation with biodiversity conservation

Pine plantations established on former heathland are common throughout Western Europe and North America. Such areas can continue to support high biodiversity values of the former heathlands in the more open areas, while simultaneously delivering ecosystem services such as wood production and recreation in the forested areas. Spatially optimizing wood harvest and recreation without threatening the biodiversity values, however, is challenging. Demand for woody biomass is increasing but other pressures on biodiversity including climate change, habitat fragmentation and air pollution are intensifying too. Strategies to spatially optimize different ecosystem services with biodiversity conservation are still underexplored in the research literature. Here we explore optimization scenarios for advancing ecosystem stewardship in a pine plantation in Belgium. Point observations of seven key indicator species were used to estimate habitat suitability using generalized linear models. Based on the habitat suitability and species’ characteristics, the spatially-explicit conservation value of different forested and open patches was determined with the help of a spatially-explicit conservation planning tool. Recreational pressure was quantified by interviewing forest managers and with automated trail counters. The impact of wood production and recreation on the conservation of the indicator species was evaluated. We found trade-offs between biodiversity conservation and both wood production and recreation, but were able to present a final scenario that combines biodiversity conservation with a restricted impact on both services. This case study illustrates that innovative forest management planning can achieve better integration of the delivery of different forest ecosystem services such as wood production and recreation with biodiversity conservation.

This is a preview of subscription content, access via your institution.


Update OSM data already imported in qGIS - Geographic Information Systems

W e'd like to provide easy, simple, quick access to the "where" information about people, places, events and things. We're especially interested in connecting people in communities with information about the things happening around them. And we prefer to do it using and building on the ubiquitous, free and open source spatial tools that other communities have already built.

OpenEarth are a small consultancy formed in 1998 in Sydney, Australia and now back in New Zealand, based in and working from Christchurch.

We have been involved with spatial information from its very start and can draw on our experience in the US, UK and Europe as well as extensively throughout Australia.

We are particularly enthusiastic and knowledgeable about Open Source Spatial Systems and Data as we think they greatly extend the possibilities for making effective use of the world's ever-increasing spatial information and genuinely improving our societies.

  • How a spatial or mapping presence could be added to your existing information systems
  • An evaluation of your existing spatial systems to see if they could benefit from newer technologies and data sources
  • Determining how open source could be added to supplement or even replace existing proprietary solutions.
  • Analysis of your existing or planned spatial processes and data management
  • Development of specifications and architectures for new spatial systems
  • Spatial database design and optimisation
  • Spatial data sourcing, cleansing, structuring and presentation
  • Spatial Software and Services development and implementation
  • Hosting, Training, Documentation and on-going Support.

Much of our work can be and often is undertaken remotely but we can also travel in Australia / New Zealand to undertake agreed tasks and commissions.

PostGIS

P ostGIS is a spatial extension to the popular Open Source relational database, PostgreSQL. It implements a powerful set of spatial abilities including 2D and 3D Vector, Raster and Linear Routing.

OpenEarth has many years of experience in both PostgreSQL and PostGIS and can develop spatial storage and analysis functions that help you visualise and gain insight into your data in new ways.

We can also provide tutorials and training in PostGIS and PostgreSQL to help staff get up to speed on the spatial aspects of Relational Databases. Because most Spatial commands are based on the SQL/MM standards, spatial queries are relatively portable between Spatial-supporting SQL databases.

GeoServer

G eoServer is an Open Source Java-based server for sharing and presenting geospatial data. It can access a wide range of data sources, analyse and manipulate the spatial data and present it as interactive maps that can be used to query the underlying information.

It can also supply raw image and vector based spatial information to a wide range of existing map clients (both Open Source and Proprietary) using the widely-adopted standards of the Open Geospatial Consortium (OGC)

Open Street Map

O pen Street Map (OSM) is a community or crowd-sourced, Open Source map set of the entire world created by thousands of contributors throughout the world.

Spatial Odd Jobs

O penEarth is a small, responsive consultancy that can quickly and easily provide solutions to problems which would be difficult or expensive to justify requesting from a larger organisation.

We have many years of experience in being a flexible, adaptable resource "on tap" for other organisations to quickly solve spatial and web problems of various sizes but especially those that are urgent or time-sensitive, and might require the application of a wide range of skills. We can also take on small jobs which might consist of as little as a few hours work. With our experience and skill set, we can accomplish, in hours, that which might take days internally.

OpenEarth runs its own servers and so can make results, as they are developed, easily available for viewing and discussion.

Typical requests might be to:

  • Take the data from this spreadsheet and make a map with icons and colours reflecting particular attributes from the spreadsheet.
  • Make the map interactive and access data from this other resource when a user clicks on an icon
  • Tell me the number of 30-55 year old males living within 15km of all these addresses
  • Create an API so my software can send an address and it responds with which of these sales areas that address is within
  • Load all this data in Excel, CSV and PDF formats into a database, investigate potential ways to link the data, geocode the addresses and provide a web interface for creating customised PDF reports
  • Create Microsoft Word docs that can automatically be populated with text and map images from a spatial database to create customised reports

Opening Closed Spatial Systems

I t's noticable that even the predominant proprietary spatial systems are now taking Open Source Spatial tools and solutions seriously. Their newest offerings are often similar to, have learnt from or are even built on existing Open Source alternatives - javascript libraries such as JQuery, C3, D3, OpenLayers and Leaflet, extensive use of popular open browser technologies such as HTML5, CSS3, SVG and Web/GL, Open Source middleware such as GeoServer and Nodejs and open source spatial data stores such as PostGIS/PostgreSQL.

OpenEarth has quietly been ahead of the curve when it comes to Open Source Spatial Innovation. From a Google Maps-like web mapping solution, SPLASH, (Spatial Planning Linkup Around Sydney Harbour) implemented in NSW's Planning Department years before Google Maps itself, to early adoption and adaption of many Open Source tools such as PostGIS/PostgreSQL, MapServer, SVG and WebGL.

OpenEarth can "add space" to your existing alphanumeric solutions or "add openness" to your existing spatial solutions. We can provide Open Source functions to access your existing local or remote data, improve its usefullness by adding location information and spatial context such as addressing and administrative area (suburbs, postcodes) and present the enhanced information in many ways including the results of spatial analysis (How close. how many. where. ) interactive 2D and 3D maps, business graphics or high resolution spatial images suitable for printing. We can also take data from your existing GIS and present it cheaply and efficiently to a wide audience of users without the need for user and licensing management.

Evaluating Existing Spatial Systems

O penEarth can review your existing spatial systems without a vendor's bias. While we obviously charge for our time, the products we propose are free and ubiquitous with excellent, world-wide support through an active user and development base.

Everything we develop for you is yours so you can continue to modify and improve it yourself or hand it on to others with the same skills. Open Source means you're not tied to specific, vendor-approved help.

  • Inventoring existing systems - products, versions, compatibilities, licensing, training and support costs etc.
  • Understanding how the various components are being used. Are they being utilised to their maximum capabilities?
  • Identifying gaps and duplications in systems
  • Proposing, if appropriate, alternative or supplementary tools and approaches
  • Designing and documenting the elements necessary to make the improvements
  • Writing and implementing the proposed changes
  • Hosting development and test sites and, if required, hosting the final solutions or add-ons

While Open Source spatial tools are some of the most powerful and flexible available and are constantly improving, there are cases where an existing proprietary system "just works" and is the best solution for a specific task. OpenEarth can determine how well your current solution meets the requirements of staff, customers and other users. We can suggest specific areas where an alternative technology could improve things in terms of performance, flexibility, scaleability, usability etc.

Design of New Systems

O penEarth have many years experience in designing and implementing spatial solutions in a wide range of markets, especially for Telecoms.

We can consult, design, build and host a solution to meet your exact requirements using an agreed set of components.

We can also integrate new functionality into or beside existing systems and replace old functions with faster, more efficient and open alternatives.

OpenEarth have produced complete start-to-finish spatial systems for telecoms, software companies and transport fleet management, amongst other commissions.

Spatial Sourcing

T he volume, quality and availability of Spatial Data is growing by the day. Every satellite, every geographically-distributed network, creates large volumes of valuable and interesting data. OpenEarth can help search, discover, evaluate and utilise the right datasets to suit your specific needs.

We can determine the best options for data access based on issues such as currency, accuracy and precision, volatility and likely network impact and then implement the best local or remote, static or dynamic, file, database or API-based methods of data storage and access.

Dynamic Design

D ynamic Design is an international company, with headquarters in Switzerland, providing a range of solutions for managing networks, especially in the telecoms market.

Their products are based totally on proprietary tools database, middleware, spatial tools (GIS) and user interface. While these tools were appropriate in 1993 when the company was founded, and have supported development through the years, the ever-growing issues of licensing costs and management requirements version managment of multiple products a lack of control over changes or improvements to these essential components and a growing international customer base meant that easier, simpler alternatives were being actively canvassed.

OpenEarth provided consultancy into Dynamic Design in its Melbourne offices to establish just what could be accomplished with Open Source alternatives to their existing proprietary tools. After some initial familiarisation, OpenEarth was able to show how the Spatial Database, Spatial Server, User Interface and Data Import and Export tools could all be replaced or supplemented with Open Sources equivalents. OpenEarth was also able to provide browser-based proof-of-concept vector graphics using Open Source graphics libraries essentially equivalent to, and as functional as, those of the existing Dynamic Design desktop solution.

The consultancy proved conclusively, that Open Source Spatial solutions could replace most if not all their existing proprietary tools and remove entirely, issues of licensing on customer systems. Customer Support would also benefit as full control was available over all components of their solution, along with the active, vibrant world-wide support community that characterises the best Open Source solutions.

It showed that by adopting Open Source alternatives for some or all of its external tools, Dynamic Design would be able to provide its solutions without external and third-party licensing requirements, making the setting up of new clients considerably easier and cheaper. In addition, by changing to a browser-based User Interface using Open Source graphics libraries, it became realistic to consider providing it products as Services delivered over the web. Because Open Source Spatial solutions provided Dynamic Design developers full access to source, they could also optimise the tools themselves, enabling better control and so a better experience for their users.

Virgin Mobile Location Services

O penEarth was approached by an Australian Mobile Telecoms company, Virgin Mobile, with a spatial problem. Virgin wished to sell a home broadband solution Virgin Broadband At Home which utilised their existing 4G network and included a home phone with an allocated land-line number.

Australia's Emergency Services conventions required that every Land-Line number (+61 x xxx xxxx) be spatially located and provided to a centralised database, the Integrated Public Number Database (IPND) from which emergency operators throughout the country could automatically identify a location simply from the telephone number from which a call was being made. This meant that the At Home devices needed to remain at the location provided when the devices were purchased and that an subsequent changes of address were noted and the new address provided to the IPND. The problem was that the At Home devices were small and portable and would work anywhere. The challenge was to identify when devices moved outside their approved location and communicate, as automatically as possible, with the device owners.

  • Accessed Virgin Customer Management systems to extract a list of new, changed and dropped At Home devices.
  • Use the provided address to geo-locate the device's authorised location for subsequent comparisons
  • Updated its own Open Source Spatial database, PostGIS/PostgreSQL so that customer details properly reflected Virgin's Customer Database.
  • Nightly use Virgin's GSM Location Services API to interrogate the location of each At Home device.
  • Determine, using a flexible set of algorithms, whether the device location was within a designated area. This was based on distance of the device from cell towers, location accuracy, the device's history of previous movements, whether the device location crossed various administration boundaries and a number of other contributing factors.
  • If the device was out-of-place, to issue SMS and/or Email messages requesting the device to be returned to its authorised location - a sequence of different messages were sent depending on whether the device was still out of place.
  • Finally escalate issues, if necessary, to Virgin Customer Service from where staff then contacted the device owners
  • Provide a Map-based tool for Virgin staff to monitor all devices
  • Provide various reports in Excel, CSV and PDF formats for input into other Virgin Services

The system was successfully implemented and hosted by OpenEarth. It ran for approximately five years until the At Home product was replaced with new technologies that allowed handsets to become WiFi Points of Presence and use of landline numbers was no longer required.

Optus Location Services

B ased on previous, successful services provided to Virgin Mobile, OpenEarth developed a similar device location and reporting service for Optus Networks. The system provided comprehensive monitoring and reporting facilities customised to the Optus Customer model.

The Optus system performed a function similar to that developed for Virgin which was monitor, control and report on devices that were contractually required to remain in designated "Home Areas".

Optus Reseller Location Services

B ased on previous, successful services provided to Virgin Mobile, and Optus, OpenEarth was further asked to develop and host a customisable device location and reporting service for Optus Networks Resellers. This allowed the service to be customised to meet the various needs of each of the Optus Resellers.

The system provided comprehensive monitoring and reporting facilities customised to each Optus Reseller's model. The system performed a function similar to that developed for other products which was to monitor and control devices that were contractually required to remain in designated "Home Areas".

Nangi Health Database

N angi ("to see" or "to look" in the language of the Ngunnawal people of Canberrra) was developed for the peak Aboriginal Health body, NACCHO to bring together a wide range of information on the Aboriginal peoples of Australia to understand where they were physically placed and to identify the Health resources available to them.

OpenEarth consulted, designed, developed and hosted the Nangi Site allowing NACCHO member organisations an integrated, map-centric view of the state of the State of Australian Aboriginal Health.

Quality Use Of Medicines

Q MAX represents a key factor in ensuring good health outcomes - ensuring a regime of proper and timely medication. QUMAX helps patients with complex medications using a number of simple measures along with careful monitoring and measurement.

A significant upgrade to the QUMAX system was designed and implemented by OpenEarth for NACCHO, the National body responsible for QUMAX Administration. It supports the development of Workplans covering the seven key QUMAX areas along with two Progress Reports each year which ensure the workpans are being effected.

Cadastral Management

O penEarth consulted with the New South Wales Land Information Centre (LIC) to design a service that allowed surveyers and developers to submit plans and surveys online with automatic detection of CAD features and reading of the drawings into a spatial database.

Part of the work was an evaluation of two Cadastral (Property Title) databases one at LIS and the other at Sydney Water. The analysis highlighted differences and errors between the two cadastres such that a successful merge of the two cadastres could be implemented.

The Panorama system demonstrated a full workflow of new property development from acceptance of CAD drawings through incorporation into provisional spatial database layers and publishing of the new data versions in formas that could be used to update assets tied to the cadastre (cadastral shift).

National Address Management Framework

O penEarth provided consultancy to the Australia New Zealand Land Information Council (ANZLIC) to review all aspects of addressing in Australia, interview key participants and provide three deliverables:

A Standard Address Data Model.
A Standard Address Transfer Format.
A Standard Address Web Services Interface.

OpenEarth researched and conducted interviews for all three deliverables but was most involved in designing, implementing and documenting the Web Services Interface which provided API access to NAMF-compliant Address Services.

Aboriginal Community Controlled Health Services

T here are approximately 150 ACCHS across Australia providing Primary Healthcare to Indigenous Australians. OpenEarth developed a custom tool to allow the gathering, analysis and reporting of information on these organisations and on the communities within which their services were offered. The ACCHS site locations themselves as well as the locations of outstations and clinics were mapped so that their potential population catchments could be estimated and reported on as an aid in planning future requirements.

The ACCHS Tool collated a large amount of previously-scattered data and so helped the organisations better understand their areas of responsibility (local, regional, state and national) and so imporove the coordination of their efforts, resulting in significant savings in both time and resources.

MapChat

M apChat is a simple service that allows multiple map viewers to be coordinated across the web so that a map-enhanced and informed conversation can be held about particular locations. As MapChat is still in development, the best way to show its early capabilities is with the short (6min) video below. The communication mechanism between Map Clients is very quick with minimal overhead. In the video, the messages between the various browser pages (in Christchurch NZ) are being transmitted through a server in Melbourne, Australia.

There are many possible use cases for a map collaboration facility such as MapChat. OpenEarth are always keen to hear how the underlying technologies can be used to solve other user issues.

If you'd like to try the actual MapChat tool, assuming the Server is running, you can access the client here. You are best off using the Google Maps client at the moment as the other clients have varying amounts of functionality. The Cesium 3D client allows three dimensional navigation but does not yet have the drawing functions enabled.

Spatial Publishing

W hile the creation and analysis of spatial data is usually a specialist task, the results of those activities need to be easily available to a wide range of users with the ability for them to further use the results in their own ways. Typically a small number of analytical processes generate data that many more users want access to.

OpenEarth can provide customised services that "publish" the results using Open Source tools that require no licensing and can be freely scaled to meet user demands.

Smart Images

I mages are a great way to convey information and are in use throughout the web. But they are just pictures - arrays of coloured dots. So providing further information about what the image represents requires considerable extra work.

But what if the image itself could supply that other information? This is the problem Smart Images solves. Smart Images are ordinary images in every way but have extra content that can be used to make the image interactive and Smart

To better understand how Smart Images can be used, please select from these short video presentations that illustrate some uses of SmartImages.

Spatial New Zealand

I magine a simple tool that lets people throughout New Zealand cooperate in space. By that, we mean a set of spatial functions that can be connected together in various ways so that each participant can see and interact with what others are doing. Spatial systems that cooperate with each other and with multiple data sources to form a single, virtual system that everyone can use. The idea is similar to MMORPG - an acronym from the Video Games world meaning "Massively Multiplayer Online Role-Playing Game"

Spatial New Zealand is the idea of building an integrated, customised suite of spatial tools using the very best open source products and then integrating them with a layer of connectivity so that they form a single, cohesive model of the entire country, a model which can be viewed initially as the familiar interactive map that everyone knows, but eventually can become an continuous immersive Virtual Reality and/or Augmented Reality experience.


Agent-based models for infectious disease epidemiology

Agent-based models (ABMs) are a type of computer simulation composed of agents that can interact with each other and with an environment. An agent can be anything from an individual to an organization, or body, such as a nation state. The actions of agents are governed by a set of coded rules. At each time step an agent decides what it will do: the actions can be as simple as defining which direction an agent will move in based on some simulated perception or, the actions can be more complicated such as searching for agents with certain characteristics within a given radius and socially interacting with them [2]. ABMs can capture unexpected aggregate phenomena that result from combined individual behaviours in a model [3]. Although agent-based models have been around for some time, with one of the earliest published models appearing in 1971, it was not until the late 1990s that they began to gain popularity in the social sciences. This was mainly driven by the introduction of platforms such as Netlogo, Swarm and Repast that were designed to enable non-computer programmers to create and understand ABMs. As the platforms improve and computing power expands ABMs are being applied ever more broadly [4].

ABMs are becoming popular in infectious disease epidemiology as the models can capture the dynamics of disease spread combined with the heterogeneous mixing and social networks of agents [5]. To realistically model an outbreak, and to be useful in real world scenario, an ABM needs to model characteristics of a disease (such as infection rates), as well as characteristics of the agents and their environment, all at an appropriate level of detail [6].

One way to categorise ABMs used for infectious disease modelling is into those that use data and those that do not. It is possible to capture the dynamics of a system, such as the spread of an infectious disease, without the use of data. For example, the Dunham [7] model does not use any data to set up their population or run the model. However, for infectious disease modelling an ABM that does not use data has a disadvantage concerning applicability. While the model may be used to better understand the general dynamics of a disease, many things affect an outbreak including the characteristics of the population and the environment. It would be impossible to capture the effects these would have on the outbreak without the data to create them. Many models, such as Rakowski et al. [8] and Crooks and Hailegiorgis [9], use data sources to set up their model. Rakowski et al. [8] use both Polish census data and landscan data (https://web.ornl.gov/sci/landscan/), which is a global population distribution dataset, to create their influenza simulation. Crooks and Hailegiorgis [9] use data from a refugee camp and GIS elevation data for a model on the spread of cholera. Because both models use data the results can be directly applied to a real population and can help to influence future policy.

Infectious disease ABMs can be categorised into models that are created to simulate a specific disease or specific outbreak and those that are created to simulate general disease dynamics [6]. There are many ABMs that focus on specific strains of influenza such as H1N1 [10] or H5N1 [11], or treat influenza generally [8]. Other agent-based models have been based on specific outbreaks, for example the model by Merler et al. [12] that simulates the Ebola outbreak in Liberia. Not only does this model include specifics to how Ebola spreads, such as contact at funerals, but the model is specific to Liberia including the number of hospital beds that were used for Ebola patients over the course of the epidemic [12]. Specific ABMs have also been created to determine the effects that the government mandates had on the spread of the H1N1 virus in Mexico [10] and how vaccination programs affect the incidence rate of Human papillomavirus (HPV) in Denmark [13].

Modelling frameworks can be used to study disease dynamics, for example the models by Duan et al. [14] and Dunham [7]. These more frameworks are commonly used to influence public policy. For example, FRED (A Framework for Reconstructing Epidemiological Dynamics) is an agent-based modelling system that is used to support research on the dynamics of infectious diseases particularly for state and county public health officials to evaluate the effects of interventions [15].

In order to create the models described in this paper different types of data are needed including population statistics, GIS data, school and workplace locations and vaccination data. The majority of data used comes from Ireland’s Central Statistics Office (CSO) [16], but other sources are also used. The following sections outline the sources of the data used in the model.

Population statistics

Population statistics are used within the model to create a realistic population of agents. Real data is used to determine the age and gender breakdowns of our populations along with correct distribution of household size and other household characteristics such as child age. The CSO provides a wealth of open access data. The data is taken from the results of the Irish census which occurs every five years. The data used for our model is from the 2011 Irish census, data from the 2016 census has recently been made available, however the 2011 data is more suitable for the outbreak we attempt to simulate as it occurred in 2012. The census data is organized into fifteen different themes each with a set of tables containing information on the population of Ireland under that theme. The themes are described in Table 1. The themes used to create the model are theme 1, theme 4, theme 5 and theme 8.

Data can be downloaded at multiple geographic levels, the lowest being the small area [16]. Small areas are areas of population that contain between 50 and 200 dwellings. We base our simulations on data at the small area level. The CSO make available a data set (delivered in csv format) for all small areas in Ireland containing data for each table within each theme. When simulating a specific town the small areas related to that town and the necessary tables can be selected from the data set. The small area boundary file discussed in the next section provides a mapping between small areas and towns. Table 2 contains the information on the different CSO tables that were used to create the simulation.

GIS data

Various sources of GIS data are used in our models. GIS data not only gives us the town boundaries but also residential, commercial and recreational areas within the town that help to define where the agents live, work, and travel.

The CSO provides access to boundary files from the 2011 census. The files contain the boundaries at different levels including provinces, counties, electoral divisions, towns and small areas [16]. The data set downloaded from the CSO website contained small area information for all of Ireland: the QGIS [18] software was used to select only the small areas that overlapped with the town being simulated so the data could be loaded into Netlogo. The small area boundaries do not always match town boundaries, thus the small area data set could potentially cover more area than the town being simulated.

Zoning data is taken from two sources: Open Street Maps [19] and Myplan.ie [20]. Myplan.ie gives the shape files that include local area development plans. While Open Street Maps provides land use data. The land use data is a shape file that provides information on if the land is used for residential, commercial, retail or industrial purposes. The data set can also provide more detailed information such as if the land is used for religious purposes, sports pitches, cemeteries or reservoirs. Neither source is comprehensive and there are some areas in the towns for which zoning data is not available. The different zoning and land use types are sorted into six categories: open, town center, community, residential, commercial and mixed.

School locations

In order to determine both the number of schools in a town and their locations we use data from the Department of Education and Skills in Ireland. They provide data on individual schools, including enrollment and type of school (primary or secondary) [21]. The data set includes the longitude and latitude of the schools. These are then geocoded in QGIS [18] in order to create a GIS shape file that can be combined with the town boundary and land use shape files and loaded into Netlogo.

Vaccination data

Vaccination statistics are used to determine the number of agents in our model who have been vaccinated and thus are immune to the disease. Vaccination statistics for Ireland can be found on the childhood vaccination schedule. Statistics are presented for Ireland as a whole and broken up into Health Service Executive (HSE) regions. Vaccination uptake statistics for the whole of Ireland and by HSE region are available on Ireland’s Health Protection Surveillance centre website going back to 1999 [22]. The Organisation for Economic Co-operation and Development (OECD) reports vaccination rates for Ireland back to 1983 [23]. When initiating the model the choice of all Ireland vaccination rates or vaccination rates for a specific HSE region based on the town being modelled must be made. Further discussion of how vaccination rates are used to in the model can be found in the “Society” section.


On the nexus between landslide susceptibility and transport infrastructure – an agent-based approach

Road networks are complex interconnected systems. Any sudden disruption can result in debilitating impacts on human life or the economy. In particular, road systems in mountain areas are highly vulnerable, because they often do not feature redundant elements at comparable efficiencies.

This paper addresses the impacts of network interruptions caused by landslide events on the (rural) road network system in Vorarlberg, Austria.

Based on a landslide susceptibility map we demonstrate the performance of agent-based traffic modelling using disaggregated agent data. This allows us to gain comprehensive insights into the impacts of road network interruptions on the mobility behaviour of affected people. Choosing an agent-based activity-chain model enables us to integrate the individual behavioural decision-making processes into the traffic flow model. The detailed representation of individual agents in the transport model allows optimisation of certain characteristics of agents and including their social learning effects into the system.

Depending on the location of the interruption, our findings reveal median deviation times ranging between several minutes and more than half an hour, with effects being more severe for employed people than for unemployed individuals.

Moreover, results show the benefits of using agent-based traffic modelling for assessing the impacts of road network interruptions on rural communities by providing insights into the characteristics of the population affected, as well as the effects on daily routines in terms of detour costs. This allows hazard managers and policymakers to increase the resilience of rural road network systems in remote areas.

Infrastructure networks and related assets support the delivery of essential goods and services to society (European Commission, 2017 Mejuto, 2017 Gutiérrez and Urbano, 1996) . In particular, the functionality of socio-economic systems in modern communities heavily depends on extensive, interconnected transport networks because any disruption may cause rippling effects, eventually entailing instability of other critical infrastructure – both domestically and beyond (Bíl et al., 2015 Jaiswal et al., 2010) . The main challenges are negative socio-economic consequences (high direct and indirect losses) to societies as a result of hazard events (Bordoni et al., 2018 Rheinberger et al., 2017 Pfurtscheller and Vetter, 2015 Kellermann et al., 2015 Pachauri and Meyer, 2014 Schweikert et al., 2014 Pfurtscheller, 2014 Meyer et al., 2013 Pfurtscheller and Thieken, 2013 Nemry and Demirel, 2012 Taylor and Susilawati, 2012 Rheinberger, 2011 Jenelius, 2009 Koetse and Rietveld, 2009) .

The impacts caused by severe weather events and associated hazards underline the importance of resilient and reliable transportation infrastructure (Eidsvig et al., 2017) , especially in complex landscapes such as the European Alps where the topography impedes redundancies and alternative routing. Failure and disruption of transport infrastructure can therefore affect a broader environment due to cascading effects which result from the dependence of economies, institutions and societies on such networks (Kellermann et al., 2015 Doll et al., 2014 Keller and Atzl, 2014 Pfurtscheller, 2014 Meyer et al., 2013 Kappes et al., 2012) . This is especially true under severe weather conditions triggering disasters, because reliable networks are crucial for emergency response to avert further damage, save lives and mitigate economic losses. Network reliability in this context is defined to comprise network availability and network safety. Non-reliable transportation networks and the associated overall societal loss introduced by destructive incidents considerably exceeds the mere physical damage to such infrastructure. Apart from an impairment of roads – which results in maintenance and reconstruction efforts to be carried out by road operators (cf. Donnini et al., 2017) – secondary effects such as intangible and indirect costs of damage to infrastructure networks have to be considered in a broader economic context and lead to considerable vulnerability of societies affected (Klose et al., 2015 Pfurtscheller and Thieken, 2013 Meyer et al., 2013 Fuchs et al., 2011 Fuchs, 2009) . Consequently the assessment of transport network systems has gained relevance in academia as well as the policy agenda of authorities across all scales (Pant et al., 2018 Unterrader et al., 2018 Bíl et al., 2017 Pregnolato et al., 2017 Winter et al., 2016 Rupi et al., 2015 Jenelius, 2009 Taylor et al., 2006 Zischg et al., 2005a, b D'Este and Taylor, 2003 Berdica, 2002) .

Since no context-free definition of road network vulnerability exists, the respective methodological approaches (even if highly sophisticated) remain fragmentary and repeatedly tailored to individual settings (Bagloee et al., 2017 Eidsvig et al., 2017 Mattsson and Jenelius, 2015 Rupi et al., 2015 Fuchs et al., 2013) .

Berdica (2002, p. 119), for example, suggested that network vulnerability should be understood as “susceptibility to incidents that can result in considerable reductions in road network serviceability”. This includes a focus of assessment on the most critical hotspots (links or nodes) within a current network system, where the highest socio-economic impact can be observed, which – according to other scholars – equals exposure (Unterrader et al., 2018 Khademi et al., 2015 Jenelius et al., 2006) . On the other hand, Taylor et al. (2006) understood network vulnerability as a concept close to network weakness and thus as the consequence of failure to provide sufficient capacity for the original purpose of the system, that being to transfer people and goods from point A to point B. This already shows the close connection of network vulnerability to other terms, such as accessibility, remoteness or robustness, which is linked to the idea of network performance (Yin et al., 2016 Taylor et al., 2006 D'Este and Taylor, 2003) . In sum, the idea behind vulnerability is a decline in the original capacity to handle the network flow based on disruption (Yin and Xu, 2010) . Nevertheless, in the literature two main directions within network vulnerability assessment can be distinguished: (1) topological vulnerability analysis, which includes the assessment of real transport network systems (represented in an abstract network) and (2) system-based vulnerability analysis, which focuses on the structure of the network within supply and demand models (Mattsson and Jenelius, 2015) . In the context of the present paper, we understand vulnerability as the assessment of the disruptive impact based on a certain event (incident) which causes a malfunction or breakdown in the current road network system (Postance et al., 2017 Pregnolato et al., 2017 Klose et al., 2015 Mattsson and Jenelius, 2015) . The potential disruption may span from natural hazard events to terrorist attacks, infrastructure collapses or ordinary traffic accidents (Bagloee et al., 2017 Unterrader et al., 2018 Vera Valero et al., 2016 Mattsson and Jenelius, 2015 Koetse and Rietveld, 2009 Zischg et al., 2005b Margreth et al., 2003) . Depending on the threat, the potential consequence can be additional travel time from some minutes to total cut-offs of several days of a community (Rupi et al., 2015 Taylor and Susilawati, 2012 Jenelius, 2009 Zischg et al., 2005b) . Therefore, a central goal of vulnerability assessment is the identification of the critical links within the current network system that are highly susceptible to such disruptions (Gauthier et al., 2018 Jenelius et al., 2006 Berdica, 2002) . In contrast to the ongoing vulnerability debates in natural hazard and risk management of buildings (see for example Papathoma-Köhle et al., 2017 Fuchs et al., 2011 or Fuchs, 2009 ), however, network vulnerability usually does not account for any probability of disruption within the assessment (Rupi et al., 2015) .

Two main methodological approaches on how to assess road vulnerability exist (Mattsson and Jenelius, 2015 Hackl et al., 2018) . The first is a topological one which focuses on characteristics of the road network's links. It is based on graph theory, which is widely used in various disciplines, such as computer science, physics, sociology and transportation (Heckmann et al., 2015 Phillips et al., 2015) , with the aim to assess and understand networks and their individual properties (Slingerland, 1981) . Using graph theory in vulnerability assessments of road networks generally means focussing on specific graph edges (links) and nodes, their criticality or redundancy to reflect resilience and interdependencies between parts of the network, as well as potential cascading effects (Pant et al., 2016 Rupi et al., 2015 Tacnet et al., 2013 Jenelius et al., 2006 Meyer et al., 2013) . This approach, however, is limited by the reduction in connectivity within a network, therefore not including the behavioural aspects of transport network users.

A second group of models bridge this gap by considering link properties and traffic demands on the links of traffic networks. The network loads, together with appropriate traffic dynamics that result in alterations of network properties, which gives rise to various stress response effects that can also be observed in real-world traffic. These models differ with respect to the chosen granularity and can be divided into macro-, meso- and microscopic models (Treiber and Kesting, 2013 Hoogendoorn and Bovy, 2001) .

Macroscopic traffic models stem from the concept of flow theory and consider aggregate continuous flow densities of anonymous users on the network. They can be applied to find equilibrium loading states within these networks, as well as to describe dynamic effects within the flow continuum. Their application usually requires solving systems of coupled equations. In contrast, the fine-grained microscopic traffic models consider each transportation network user an individual entity (an agent, i.e. vehicle or pedestrian) with separate interaction details and decisions. These models are implemented as simulation frameworks, iterating the entire network evolution over time steps. Thus, individual entities (agents) retain their specific characteristics throughout the traversal of the network and therefore can react to different circumstances based on these characteristics. Mesoscopic traffic models are hybrids between macro- and microscopic models. They are less fine grained and borrow some characteristics from both approaches, offering a description that is less detailed in time or space, but also less demanding regarding the computational requirements. Depending on the implementation, micro- and mesoscopic models can be “agent-based”, thus retaining the individuality of their agents throughout the model evolution. A more detailed conceptual distinction of agent-based models regards the scheduling of mobility demands. Simpler approaches define individual (or multiple unrelated) trips between origin and destination pairs (“trip-based”), whereas more recent frameworks allow the expression of agent activity plans or chains (“activity-based”) to be fulfilled by adaptively traversing the transportation networks of the simulation.

The change between levels of granularity in the description of model entities is referred to as (dis-)aggregation for (increasing) decreasing detail.

With the modelling discrimination provided above, the approaches of the second group of road vulnerability assessment methods allow the effects of landslide events to be explored on a given population and its subgroups with respect to their mobility requirements. The main drawbacks of aggregated (macroscopic) traffic models in that context include (1) loss of population individuality therefore (2) a lack of behavioural alterations and co-dependent learning effects of individuals (3) more time-averaging aspects, prohibiting re-decisions based on incidents (4) more space-averaging aspects, prohibiting investigation of localised events without rebuilding the overall model (e.g. new zoning structure) (5) connection to unavailable consequences of precise socio-demographic measures (6) trip-based macroscopic models, considering individual journeys instead of whole day-plans and (7) macroscopic models that are adaptable to the increasing level of detail available through continuously improving data by layering of multiple models. Choosing an agent-based activity chain model, which integrates the dynamic aspects of each agent, can overcome these limitations. The vulnerability assessment utilising activity chain traffic modelling allows simulations that integrate multiple phenomena to understand the dynamic interactions of human behaviour and the environment in the sense of consequences for households or wider socio-economic systems.

The focus of this paper is on landslide hazards, which repeatedly jeopardise the integrity of road infrastructure by causing structural damage and interruptions (Postance et al., 2017 Klose et al., 2015 Bíl et al., 2014) . In the Austrian Alps, 1444 damaging events to rural roads were recorded in the provinces of Salzburg (2007–2010) and Styria (2008–2011), and debris flows and landslides caused nearly 50 % of the recorded damage costs (König et al., 2014b) . The prevailing hazard potential caused by landslides is aggravated by findings of several other recent studies which have shown that landslide activity and thus related damage will most probably increase with progressing climate change (Schlögl and Matulla, 2018 Gariano and Guzzetti, 2016 Bíl et al., 2015 Klose et al., 2015 Strauch et al., 2015 König et al., 2014a Keiler et al., 2010) . Similar results are available from other mountain regions (e.g. Postance et al., 2017 Unterrader et al., 2018 Meyer et al., 2015 Fuchs et al., 2013).

So far, most studies have mainly focused on primary road networks (Postance et al., 2017 Taylor et al., 2006) and urban areas (Gauthier et al., 2018) , while federal and local road networks have been largely neglected. Mountain roads, in contrast to lowland roads, are highly vulnerable due a higher probability of climate-driven hazard events and the inherent obstacles of implementing redundant systems (Schlögl and Matulla, 2018 Matulla et al., 2017 Schlögl and Laaha, 2017 Doll et al., 2014 Eisenack et al., 2011) . Consequently, misleadingly termed “forgotten road systems”, local road networks in fact connect rural communities in various ways – from supply reliability over public health and tourism to all sorts of economy. Furthermore, issues mostly on technical realisation of mitigation and road maintenance have been addressed, rather than socio-demographic impacts on communities or exposed societies (Mattsson and Jenelius, 2015) . This paper partly contributes to closing the gap by including the full road network system. In particular, the relation between infrastructure and communal development in mountain areas is not one-directional, meaning that it is only the former that can impact the latter instead, the influence is two-way (Jaafari et al., 2015) .

The presented approach is complementary to previous studies because of the consideration of whole-day travel plans (as opposed to a focus on peak traffic flow periods on the investigated network), with these plans stemming from the underlying agents' activity chain model. This schedule of activities, which is far less dependent on fixed locations, allows for a more inclusive and flexible reassignment of mobility needs and resulting traffic demands. Therefore, integrating transport route finding and satisfaction of individual activity needs in one single simulation framework facilitates a more detailed and realistic representation of traffic loads on the network. We demonstrate an appropriate methodological response to foreseeable demands imposed by the increasing detail of available mobility data, which brings about particular relevance of this approach for future applications.

The applicability of the approach is demonstrated by the example of Vorarlberg, the westernmost province of Austria (Fig. 1). While being the second-smallest federal state, the population density of Vorarlberg is only surpassed by Austria's capital, Vienna, which indicates the need for a resilient transport network. The main traffic artery in this almost completely mountainous area is the connection from Germany to western Austria, via the Rhine Valley, Walgau, Klostertal and the Arlberg massif. Apart from this link, which is realised as a motorway (A14 and S16), rural roads prevail in the complexly structured topography of Vorarlberg. Because Vorarlberg is almost entirely surrounded by mountain areas and considerable exposure to extremely high rainfall (with average annual precipitation totals exceeding 2000 mm), the transport system of Vorarlberg is highly exposed to landslides. The combination of (i) being characterised by high landslide susceptibility, (ii) exhibiting a high population density and (iii) lacking alternative routes on the rural network due to the mountain orography makes Vorarlberg a perfect case study.

Methodologically the approach presented in this paper is divided into two modelling sections:


Watch the video: Download OSM data in Qgis