Changes in Road Centrality and Hospital Access Redundancy:
Impacts of the 2024 Flood in the Metropolitan Core of Porto Alegre, Brazil

Authors

Affiliations

Ricardo Ruiz Sánchez

Marcel Reinmuth

Heidelberg Institute for Geoinformation Technology gGmbH

Cristian Albornoz Espinoza

Sven Lautenbach

Heidelberg Institute for Geoinformation Technology gGmbH

Alexander Zipf

Heidelberg Institute for Geoinformation Technology gGmbH

Institute of Geography, Heidelberg University

Welcome

The Association of Geographic Information Laboratories in Europe (AGILE) has held annual conferences focused on Geographic Information Science since 1998. The next AGILE 2025 conference is held in Dresden, Germany, on 10-13 June 2025.

The conference theme “Geographic Information Science responding to Global Challenges” recognizes the pivotal role of geospatial data and analysis in informing decision-making for current global issues affecting humanity, the environment, and global resilience.

In this website you can find in detailed how the research design of the paper presented in AGILE 2025 was carried out in HeiGIT mostly in the Geoinformation for Humanitarian Aid department. The goal is to increase transparency, reproducibility, and openness of computational GIScience research as stated in the Reproducible Paper Guidelines from AGILE. For this purpose, the following files are available in the top right-hand corner of the web site:

• Paper official links to the pdf file of the review paper written in LaTeX using Overleaf, which is available on the github repository folder “paper_latex”.
• The paper rmarkdown contains the code to reproduce the results written in the paper, which is fully available in the github repository directory “paper_rmarkdown”. The Github repository also contains the rmarkdown file after it has been rendered as a PDF.
• The Github repository contains all the data for the creation of this website, the aim of which is to replicate the research methodology and the data produced in the process.
• Zenodo keeps a copy of the data needed to replicate the methodology and reproduce the results obtained, mainly the data folder in the github repository.

Body of Knowledge

The topic of this study includes Open Spatial Data, Open Science, Open-source Geospatial Tools and Applications , Disaster and Risk Management, Way Finding, Routing and Indoor Navigation or Spatial Databases and Data Management.

The submission type of this study is use case, while the Body of Knowledge (BoK) is the following:

• [GIST] Geographic Information Science and Technology
  • [AM] Analytical Methods. We derive analytical results from such as connectivity or redundancy from public Geospatial data.
    • [AM11] Network analysis. We used the road network of the Core Metropolitan area of Porto Alegre to derive results based on the graph theory. In the graph theory edges and nodes are represented as a system of intercon- nected networks (Freeman, 1977; Wasserman and Faust, 1994)
      • [AM11-2] Graph theoretic descriptive measures of networks. We used edge betweenness indicator to measure the level of connectivity on the road network of Porto Alegre
      • [AM11-3] Least-cost shortest path. We used pgrouting to compute the optimum path through a network with Dijkstras algorithm based on the weights given by OpenRouteService (ORS).
      • [AM11-7] Accessibility modeling. We defined a method for measuring redundancy on the access to hospitals.
  • [GC4] Open Science.We used open data such as OpenStreetMap (OSM), as well as services like OpenRouteService (ORS) and developed an open and reproducible methodology using software such as PostGIS, R or QGIS.

This study is located in the network analysis of the body of knowledge including centrality metrics of the road network and accessibility of the hospitals based on least-cost paths.

Data description

The table 1 summaries the initial data from which intermediate and final results are derived. The section link includes the origin of the data, so third parties to reuse. In addition, the public github repository with open data license also provide the data of the study.

The total built-up volume weighted the sampling of the origin and destination and is accessible via the following DOI:10.2905/AB2F107A-03CD-47A3-85E5-139D8EC63283. The degree of urbanization determined the urban dense settlement analysed contained the 9 municipalities that comprises the studied area named core metropolitan area and is accessible via the following DOI:10.2905/A0DF7A6F-49DE-46EA-9BDE-563437A6E2BA.Similarly, the subset of the flood is derived from the cheias_rhguaiba_2024_db_v11.gpkg accesible in the DOI:10.5281/zenodo.11164049. We also incorporated data from the Brazilian Institute of Geography and Statistics (IBGR) ¹ to add the GPD of the municipalities or from the Economy and Statistics Department of Rio Grande do Sul State(DEE-SPGG) ² about the population affected by the flood. For the hospitals, we used the dataset Hospitais_com_Leitos_de_UTIs_no_RS.geojson that included the bed capacity and is accessible via the DOI:10.7303/syn32211006.1. The location of the healthcare facility is obtained using the WFS provided by “Secretaria Estadual da Saúde/DGTI”.

Input data

library(DT)
library(tidyverse)
input_data_csv <- read.csv("~/agile-gscience-2024-rs-flood/data/source_data/input_data_csv.csv")
input_data_csv$Link <- paste0('<a href="', input_data_csv$Link, '">', 'Link to input: ',"<br />",input_data_csv$Data, '</a>')
input_data_csv |> 
DT::datatable(extensions = 'Buttons',
              filter = "top",
              options=list(
                          dom = 'Bfrtip',
                          columnDefs = list(list(className = 'dt-center', targets = c(1:3))),
                          pageLength = 7,
                           buttons = c('copy', 'csv')),
              escape = FALSE) 

Computational Workflow

Computation steps

This document is written following the Reproducibiltiy paper guidelines for the AGILE conference 2025 (Belliard et al. 2019).The following flowchart explains the methodology to analyse the centrality and resilience of the road network and healthcare facilities in the core metropolitan area of Rio Grande do Sul.

1. Select Area of Interest (AOI) includes importing the flood extent, municipalities and the road network affecting the core metropolitan area of Porto Alegre. We name core metropolitan area to the 9 municipalities in the urban dense human settlement that intersects with the municipality of Porto Alegre.

2. Obtain routable network created the two output post-disaster and pre-disaster network. Firstly, OpenRouteService is used to transform the OpenStreetMap data into a routable graph named pre-disaster network. Extracting the difference between the flood extent obtained in the previous lane with the pre-disaster network generated the post-disaster network. Both road networks are graphs, which allow to use the extension pgrotuing.

3. Analysis centrality and resilience is carried out calculating shortest paths with the inputs weighted and regular origin-destination. While calculating connectivity using the edge betweenness indicator provides results to answer the 1st and 2nd research question, calculating the lack of redundancy with alternative paths provides results to answer the third research question.

Fig.1 The research design illustrates how three activities are required to calculate centrality and resilience providing the information to answer the question of how the flooding in Rio Grande do Sul affected the road network and hospital in the core-metropolitan area of Porto Alegre. We used open geospatial tools and data such as PostGIS, pgrouting, ORS ,R and OSM

Computational environment

• PostgreSQL 15.3 is used from the docker file container that includes the extension PostGIS 16-3.4-v2024.03.17 and pgrouting 3.5.
• The libraries that includes the study analysis, visualization and creation of this documentation is available with the renv file.
• The study required a CPU Intel(R) Core(TM) i5-4300U CPU @ 1.90 GHz with 15 Gi model HP EliteBook 820 G1.
• QGIS Version 3.38 ‘Grenoble’
• RStudio 2023.06.1+524 “Mountain Hydrangea”
• openrouteservice:v8.0.0 used following the described docker file container instructions. Specifically, we used the docker settings found in the directory “~/agile-gscience-2024-rs-flood/data/docker_settings/docker-compose.yml” on the project’s github.

Expected execution times

While calculating the connectivity took 81 and 54 seconds in total, the computation of the lack of redundancy was the most expensive process requiring 52 minutes (3097 seconds). Another demanding process was exporting the ORS file from the docker to the R environment using the get_graph.R script. We included the computation times of the main processes to obtain the connectivity and lack of redundancy results. Providing data to support the research question 3 related to the lack of redundancy took 76% of the time required, while calculating the connectivity in a intracity (IC) and core metropolitan (CM) scale only took 1.32% and 2% respectively.

library(DT)
library(tidyverse)
computation_agile <- read.csv("~/agile-gscience-2024-rs-flood/data/computation_times_agile_2025 _fresults.csv")
computation_agile_dt <- computation_agile |>
                    mutate( 
                            time_sec = round(time/1000),
                            total_perc=round(time/sum(time)*100,2)) 
DT::datatable(data=computation_agile_dt,
              colnames= c("Lane","Activity","Description","time (s)", "time (%)"),
              extensions=c("Buttons",'RowGroup'),
                                 options=list(rowGroup=list(dataSrc= 1),
                                              dom="Bfrtip",
                                              pageLength=11,
                                              buttons=c("copy","csv","pdf","colvis"),
                                              columnDefs =list(list(visible=FALSE, targets=c(1,3)))),
                                  selection="none") |>
  DT::formatStyle("total_perc",
              background=DT::styleColorBar(c(range(computation_agile_dt$total_perc)[1],100),'lightblue'),
                backgroundSize = '98% 88%',
  backgroundRepeat = 'no-repeat',
  backgroundPosition = 'center') 

Sponsors

Belliard, Frederique, Rusne Sileryte, Anita Graser, Karl Broman, Marta Teperek, Carlos Granell, Barbara Hofer, et al. 2019. “AGILE Reproducible Paper Guidelines.” https://doi.org/10.17605/OSF.IO/CB7Z8.

---

1. https://www.ibge.gov.br/cidades-e-estados/rs/

2. https://mup.rs.gov.br/