Feed on

The HeiGIT team had a great time at the State of the Map 2022 in Florence. Many thanks to the participants for the fruitful discussions at our workshop!

Despite the direct use of OSM as a map, more and more organizations use OSM as a source of data for activities, services, or applications.

“How good is OSM in Mozambique?”
“Can I use OSM for car routing in Mumbai?”
“Does OSM include the relevant healthcare facilities to plan a humanitarian action in Nepal?”

If you are a data user, these questions might sound familiar to you.
Data Quality and the concerns about “Fitness for purpose” are important issues.

The HeiGIT team creates tools to make it easier to answer these and other questions about OSM Data.

OSM data quality estimation is a complex task and the results are sometimes hard to interpret. The goal of one of our tools – the Ohsome Quality Analyst (OQT) – is to make this process as user-friendly as possible. Besides a Web Interface, OQT provides an even more powerful Web API to generate Data Quality Reports.

As a new feature, we started to integrate some of the functionality of OQT into our ohsome History Explorer (ohsomeHeX).

ohsomeHeX is a tool to visualize maps and timeseries of data development in OSM, such as health facilities, roads or buildings. Over time certain types of map objects are added to OSM and the data grows. When the growth rate of newly added features slows down, this so-called saturation can be an indicator of map completeness. The OQT API provides such a “Mapping Saturation” indicator, which you can now easily explore in ohsomeHeX to evaluate the Quality of a specific map topic in a certain region. A traffic light system will help the end-users to interpret the calculated indicator results.

With the new Mapping Saturation indicator, it’s quite easy to make an estimate of the quality by looking at the saturation of the data. For example, if a region has a  saturation of less than 30% in the last 3 years it’s denoted as Red, or in simpler words would mean this region is not saturated yet and a lot of features are not yet created in OSM.

[figure 1: Example of Low Quality (Red) OSM Data. Number of Hospitals in Mumbai]

[figure 2: Example of Medium Quality (Yellow) OSM Data. Number of Hospitals in Burundi]

[figure 3: Example of Good Quality (Green) OSM Data. Number of Hospitals in Heidelberg]

[figure 4: Comparing different Topics for their saturation. Comparing Number of Hospitals with Number of Clinics in Barcelona]

At the current stage ohsomeHeX started with only one quality indicator – Mapping Saturation – but there is more to come. OQT offers further intrinsic and extrinsic indicators on completeness and currentness and is still under heavy development. Already, additions such as the Building Completeness Indicator signal more to come!

So be ready for more useful integrations in ohsomeHeX for simple and convenient access to quality information and fitness for purpose of OSM data.

Related Links:

LiDAR data have become indispensable for research in archaeology and a variety of other topographic applications. To derive products (e.g. digital terrain or feature models, individual trees, buildings), the 3D LiDAR points representing the desired objects of interest within the acquired and georeferenced point cloud need to be identified. This process is known as classification, where each individual point is assigned to an object class. In archaeological prospection, classification focuses on identifying the object class ‘ground points’. These are used to interpolate digital terrain models exposing the microtopography of a terrain to be able to identify and map archaeological and palaeoenvironmental features. Setting up such classification workflows can be time-consuming and prone to information loss, especially in geographically heterogeneous landscapes. In such landscapes, one classification setting can lead to qualitatively very different results, depending on varying terrain parameters such as steepness or vegetation density. The paper focusses on a special workflow for optimal classification results in these heterogeneous environments, which integrates expert knowledge. It presents a novel Pythonbased open-source software solution, which helps to optimize this process and creates a single digital terrain model by an adaptive classification based on spatial segments. The advantage of this approach for archaeology is to produce coherent digital terrain models even in geomorphologically heterogenous areas or areas with patchy vegetation. The software is also useful to study the effects of different algorithm and parameter combinations on digital terrain modelling with a focus on a practical and time-saving implementation. As the developed pipelines and all meta-information are made available with the resulting data set, classification is white boxed and consequently scientifically comprehensible and repeatable. Together with the software’s ability to simplify classification workflows significantly, it will be of interest for many applications also beyond the examples shown from archaeology.

Ground point filtered DTM of a 1200 x 900m² large area at the foot of Mount Nakadake. The area is characterized by steep slopes and very dense vegetation. The created DTM performing well in the central area, where most of the kiln sites are located but was less successful in the border areas. 1) Sample area used in the paper containing kiln sites, 2) Paddy terraces, 3) Border areas.

Find all details in the full paper:

Doneus, M., Höfle, B., Kempf, D., Daskalakis, G., & Shinoto, M. (2022). Human-in-the-loop development of spatially adaptive ground point filtering pipelines — An archaeological case study. Archaeological Prospection, 1–22.


Code and data availability:

The software project of AFwizard is open-source and publicly available on GitHub. The data used in the case study are available online under a creative common licence here. We are convinced that it will help archaeological prospection to become significantly streamlined and dramatically increase in quality. Not at least, the software will be of great use to anyone who has to deal with classification and ground point filtering.



The project is a joint collaboration between the Department of Prehistoric and Historical Archaeology, University of Vienna, the 3D Geospatial Data Processing Research Group at Heidelberg University, the Scientific Software Center of the Interdisciplinary Center for Scientific Computing (IWR) at Heidelberg University and the Institute for Prehistory, Protohistory and Near Eastern Archaeology, Heidelberg University.

The software development work described in this manuscript was carried out by the Scientific Software Center (SSC) of Heidelberg University in the framework of the project ‘Human-in-the-Loop Adaptive Terrain Filtering of 3D Point Clouds for Archaeological Prospection’ led by Maria Shinoto. The Scientific Software is funded as part of the Excellence Strategy of the German Federal and State Governments.

(Text: Sylvia Pscheidl)

Open healthcare Access Map

Last year we started the Open Healthcare Access Map.  Initially, only a few countries and later on continents were featured. Today, we are pleased to announce that we are releasing a number of previously missing countries, achieving global coverage.

The Open Healthcare Access Map uses healthcare facilities extracted from OpenStreetMap, the isochrone method from openrouteservice and population distribution from WorldPop. Isochrones represent an area that can reached from a location within a specified time and means of transport. The isochrones from each facility are intersected with population data. This way, statements can be made about how many people potentially live within how much time distance from a health facility.

The web application is available here: https://apps.heigit.org/healthcare_access/.

Please note that this is still a prototype and feedback on improvements and desired functionalities is very welcome.

A key objective of the project is to make the data more accessible to users. Therefore, we will make the data available on HDX. Please refer to our organizational page on HDX here: https://data.humdata.org/organization/heidelberg-institute-for-geoinformation-technology. HDX, is an open platform for sharing and disseminating data across crises and organizations. It is managed by the United nations Offfice for the Coordination of Humanitarian Affairs (OCHA) and features more than 19,000 datasets from more than 300 organizations.

The first country we are publishing data for is Pakistan. Pakistan has been hit by severe flooding over the past few weeks.The published datasets provide aggregated information at different scales(administrative and regular hexagons) on the coverage of the population with health services. The data can be used to compare the coverage of different regions and identify potential gaps in coverage.

Here is the directlink to the Pakistan dataset: https://data.humdata.org/dataset/pakistan-healthcare-facilities-accessibility

Open healthcare Access Map

Related works:

Environmental developments create an ever increasing need for monitoring and protection measures. These efforts are often based on digital or technical solutions like data analyses or modelling. Yet, in order to enable effective, reliable and large scale environmental monitoring and protection techniques, large information volumes are required.

In a recently published paper, we analyse different integration workflows for bird observation images. Multiple data sources are combined to create a larger and temporally, geographically and semantically more extensive and complete dataset. The workflows leverage the power of deep learning image analyses models in combination with user generated information and citizen science projects.

A subsequent quality analyses revealed that the integration of social media images not only made a large impact in terms of data volume but also had a positive effect on data quality.

Fig. 6

Hartmann, M. C., Schott, M. , Dsouza, A., Metz, Y., Volpi, M. and Purves, R. S. (2022): A text and image analysis workflow using citizen science data to extract relevant social media records: Combining red kite observations from Flickr, eBird and iNaturalist, Ecological Informatics vol 71, https://doi.org/10.1016/j.ecoinf.2022.101782 (open access)

Related work

Lee, H., Seo, B., Cord, A.F., Volk, M., and Lautenbach, S. 2022. Using crowdsourced images to study selected cultural ecosystem services and their relationships with species richness and carbon sequestration. Ecosystem Services 54

Lee, H., Seo, B., Koellner, T., and Lautenbach, S. 2019. Mapping cultural ecosystem services 2.0 – Potential and shortcomings from unlabeled crowd sourced images. Ecological Indicators 96: 505–515

In den vergangenen Wochen hat die Monsunflut in Pakistan über 1.000 Tote gefordert und Millionen Menschen ohne Dach über dem Kopf hinterlassen. Da Berichte und Fotos von rapide steigenden Lebensmittelpreisen, Bewohnern, die durch schultertiefes Wasser waten und Satellitenbilder von überfluteten Flüssen das Ausmaß dieser Katastrophe unterstreichen, hat das HeiGIT den Disaster openrouteservice für diese Region aktiviert. Derzeit werden die Daten alle 10 Minuten aktualisiert, um die laufenden Bemühungen zur Bearbeitung von OpenStreetMap, die sich stetig ändernde Situation und die betroffene Infrastruktur widerzuspiegeln.

Darüber hinaus werden die vom Copernicus Emergency Management Service, Activation EMS629, identifizierten Überschwemmungsgebiete als Overlay auf der Basiskarte angezeigt. Mit diesen Ressourcen können Nutzer, wie erste-Hilfe-Organisationen, diese Gebiete vermeiden, indem Routen berechnet werden, die überschwemmte Gebiete umgehen. Dazu können sie ein Polygon um die Überschwemmungsgebiete ziehen und die Route wird automatisch neu berechnet.

Zuvor wurde dieser Dienst für Ereignisse wie Zyklon Idai aktiviert, ein tropischer Wirbelsturm der Kategorie 4, der über 1.300 Todesopfer forderte und als der tödlichste tropische Wirbelsturm gilt, der jemals im Südwesten des Indischen Ozeans aufgezeichnet wurde. Idai löste eine verheerende humanitäre Krise in Mosambik und anderen Teilen Südostafrikas aus. Die Kartierungsbemühungen für dieses Projekt werden von Tausenden von Mappern unterstützt, die Geodaten zur globalen OpenStreetMap (OSM)-Community beitragen.

HeiGIT entwickelt weiterhin Tools, um Geodatenlösungen für humanitäre Katastrophen in Echtzeit bereitzustellen. Weitere Projekte im Zusammenhang mit Hilfe und Katastrophenschutz finden Sie hier.

For the English version see below.

Over the past weeks, monsoon flooding in Pakistan has left over 1,000 dead and millions without homes. As reports and photos of skyrocketing food prices, residents wading through shoulder-deep water, and satellite images of flooded rivers highlight the magnitude of this catastrophe, HeiGIT has created an activation of the Disaster openrouteservice for that region. Currently, data is being updated every ten minutes to reflect the continued efforts of editing OpenStreetMap to reflect the dynamic situation and affected infrastructure.

In addition, the flood extents identified from the Copernicus Emergency Management Service, Activation EMS629 are shown as an overlay on the base map. With these resources, users can avoid those areas by calculating routes that avoid flooded areas. To do so, draw a polygon around flood zones and the route will be automatically recalculated.

Previously, this service has been activated for events like Cyclone Idai, a category 4 tropical cyclone that caused over 1,300 fatalities and ranks as the deadliest tropical cyclone recorded in the South-West Indian Ocean. Idai began a humanitarian crisis in Mozambique and other parts of Southeast Africa. Mapping efforts for this project are supported by thousands of mappers contributing geodata to the global OpenStreetMap (OSM) community.

HeiGIT continues to develop tools to provide geodata solutions to real-time humanitarian disasters. Other projects related to aid and disaster response can be found here.


Since the release of the ohsome quality analyst (short OQT) in February of last year, we have been working on developing an accessible yet powerful toolkit which provides end users insights on the quality of OpenStreetMap (OSM) data. An overview of the functionality additions made to OQT over this period of time can be found in the changelog. However, for this blog post, we wish to introduce one of the new major additions to OQT from release 0.10.0, the new Building Completeness Indicator. Before diving in on how to use the new indicator in practice, the underlying principles on how the indicator has been constructed will be discussed, as to facilitate an understanding behind the working of the Building Completeness Indicator.

Building Completeness Indicator – The Concept

The core functionality of the Building Completeness Indicator is to use proxy variables to predict an expected building area in a given area of interest (AOI) and to compare this prediction to the current building area mapped in OSM. The prediction is based upon a number of covariates which include; population and population density, settlement typologies, (both based on the Global Human Settlement LayerGHSL GHS-POP R2019AGHSL SMOD R2019A), the subnational Human Development Index and nighttime lighting as a degree of urbanity (EGO – Nighttime Lights VNL V2).

The indicator relies on a Random Forest Regressor and the prediction is performed on a hexagon grid basis. As such, your AOI is split into several smaller grid cells. The overall completeness is derived as the weighted average of all the hex-cells, where the weights are defined by the predicted building area. For release 0.10.0, the new indicator only provides reliable results within Africa, as the building data set (Open Buildings by Google) used to train the Random Forest regressor is limited to Africa. However, in future releases we plan to extend our coverage by using multiple open data sets such as the ones provided by Microsoft.

Building Completeness Indicator – In Use

By using a small case example, it is possible to observe how this weighted average operates in determining “building completeness“. The aim of this small case example is to determine the Building completeness for Lesotho. Lesotho is prone to droughts, our hypothetical research analysis is thus aiming to determine which population centres have access to plumbing and which areas do not. For doing so, we would like to establish what the quality of OSM building data is in Lesotho, in which building completeness acts as a relevant factor.

Fig. 1: Calling the Building Completeness Indicator in an API POST request (/indicator) in the web Application

It is possible to access the Building Completeness Indicator in either the web application or Command Line Interface. Using the web application, it is possible to call the indicator by defining building_area as layer.name in an API POST request (/indicator) with a custom AOI within Africa (in this case Lesotho). When wanting to obtain a visual report it is possible to request an SVG (“includeSvg”: true) and, since release 0.9.0 it is also possible to request a HTLM snippet (“includeHtml”: true). When having executed the API request, the response should look similar to the figure below.

Fig. 2: Building Completeness Graph

For the bounding box of Lesotho, we observe a variability in building densities which is relevant for the weighted average given due to how the prediction is executed, namely per hex-cell. If hex-cells which show low building count with a low total building area mapped in OSM compared to the predicted building area, then they should weigh in less heavily as hex-cells with a low completeness ratio and a larger number of buildings.

Fig. 3: A visual such as presented in the figure above is not produced in standard API requests.

In the figure above it is possible to observe that although there are a number of hex-cells which show a low completeness ratio, they only represent a small area of the total building area and thus play less role in the weighted average than the hex cells which present a completeness ration between 80 – 90 %, as they amount to a larger percentage of the total building area.

For our hypothetical case example, the results imply that overall a high degree of building completeness can be determined for Lesotho, as the report states, a weighted average of 88%. It seems that hex-cells that have a higher degree of urbanity and thus a higher share of the total building area are relatively well mapped in OSM compared to a number of hex-cells which occupy a lower share of building area. This could imply, for the research team, that the quality of OSM data in some rural areas is lacking. When combined with other OQT tools and other services based on the ohsome framework such as ohsomeHeX, the research team could get further insights on the quality of OSM data in the region, which could provide useful insights into how to develop a methodology or interpret their conclusions.

Final Notes

In addition to those new features to OQT various smaller changes have been made, all to be found in the changelog. A recent addition, from release 0.10.1, allows the user to request a “Multilevel Mapping Saturation” report which indicates the mapping saturation of four Map Featuressimultaneously. If you have any ideas, feedback or would like to contribute to OQT feel free to contact our team via ohsome@heigit.org. OQT is Open Source and the development is done on GitHub, where contributions are welcome.

The further development of OQT of course does not end here. We have a couple of interesting new additions in the pipeline, including, as mentioned earlier, an expansion in the use cases for the Building Completeness Indicator, where multiple data sets will be utilised to allow worldwide access.

Related Work

OQT relies on OSM data processed by the ohsome framework developed at HeiGIT. The aim of the ohsome framework is to make OSM’s full-history data more easily accessible for various kinds of data analytics tasks, such as data quality analysis, on a regional, country-wide, or global scale. Here you find a list of related blog posts and publications:

Further Reading on the Building Completeness Indicator



Creating tools to map and navigate the urban (and not-so-urban) jungle has been a specialty of the HeiGIT and GIScience teams since their founding. We’ve plotted our way to the closest park through meinGrün, fostered awareness of barriers for limited-mobility travelers with CAP4Access, and helped find shortcuts through the unprecedented heat with HEAL. The latest addition to that list is Gefahrenstellen.de, which has been generating buzz and headlines in its native Bonn and beyond. The product of three enterprising brothers, the project seeks to help parents and schoolchildren select the path to school with the least traffic and risk, a task that currently weighs on parents’ minds as the new school year begins. For their routing service, the project employed ORS, the openrouteservice client developed by HeiGIT.

Fig. 1: Report Gefahrenstelle Interface

To learn more about the innovative and timely project, we spoke to Arno Wolter, one of the three founders and CEO of the Initiative for Safer Roads which developed Gefahrenstellen.de.

Fig. 2: Screenshot of gefahrenstellen.de

Can you describe your background?

We (I and my two brothers) have been working on developing our own web- and app-based projects for more than 20 years now. In that role, we have been working on user-friendly and needs-based solutions and have been crunching lots of data, but not so much in the traffic/mobility sector before, but in other areas such as travel, mobile phones, insurances.

What inspired you to develop Gefahrenstellen.de?

Due to painful experiences with road accidents in our personal environment a couple of years ago, we thought nowadays with the amount of data available and the existing technologies, it must be possible to invent solutions which tackle road safety more proactively and identify dangerous spots before accidents occur. That is exactly what we did with the research project FeGiS+ (or EDDA+ which stands for “Early Detection of Dangerous Areas in Road Traffic using Smart Data”). Gefahrenstellen.de is the result and platform of that research project.

What was it like working with your family on this project?

We have a well-structured split of responsibilities between the 3 brothers in the company: Henrik runs all the IT, Jörn is in charge of content and public relations, and I (Arno) look after planning, partnerships and collaborations, i.e., our activities and skills are complementary. Sometimes we have heated debates on strategies and directions, but that makes the result even better and more thought-through. We always end up with good solutions, as Gefahrenstellen.de shows.

What are your goals for the project?

Our goal is two-fold: on the one hand, we would like to raise awareness for dangerous areas in the road networks so that road users can avoid dangerous road sections or adapt their behaviour accordingly (and the safer routing option for school children is a good example for this). On the other hand, we would like to provide detailed insights on critical road sections to city authorities and to the police as additional input to prioritize their road safety work and adapt the road infrastructure before serious accidents happen.

What has been the response from users?

The user response has been extremely positive. The user reports on dangerous spots which we receive on Gefahrenstellen.de are of very high quality. A detailed study from RWTH Aachen University and the German Police University have confirmed the validity and relevance of the road user reports. Also, the new service to search the safest route to school has been already used more than 10,000 times in only six weeks after the introduction of the new service.

Fig. 3: Screenshot of routing option on gefahrenstellen.de

How did you incorporate ORS into the project?

Right now, we use ORS for the “safer school routing” on our platform. School children and their parents can search for the safest route on the platform, which avoids hazardous black spots and proposes alternative routes. Of course, such routing can only serve as a suggestion and first indication, and it is up to the parents to discuss the final school routing with their kids. ORS is the perfect fit for us as it also provides dedicated routes, e.g., cyclists also showing smaller streets or pathways.

What do you envision for the future of the project?

We will further develop the features of the platform. For example, we could also imagine a safer route navigation system for cars, as an additional option next to the already-existing fastest and fuel-optimized routes. Another priority will be the implementation of the Gefahrenstellen.de approach in other European countries. We are currently in discussion with two to three other countries which plan to introduce the approach for their road networks in 2023.

Is there anything else you would like to mention?

Every lost life in road traffic is a tragedy and, in general, accidents leave a long-lasting impact on all those involved. It remains our long-term ambition to make a contribution with this project to the VISION ZERO concept. We believe that in today’s advanced world it is not acceptable to have so many road fatalities and it is important to make best use of existing innovations and technologies to avoid accidents.

We look forward to seeing the project in use this coming school year, plugging in our own routes, and keeping up with the future innovations sure to come from Arno Wolter and his enterprising brothers.

We are happy to announce that, this year, GIScience and HeiGIT has had four contributions to the FOSS-Community “conferencathon” in Florence. As QGIS-Meeting, State of the Map, HOT unSummit and FOSS4G all took place consecutively, the schedule was shaping up to be extremely packed. We’d like to highlight some specific events and topics that you should definitely revisit. For any further questions you are invited to reach out to us at any time and via any channel.

Our team thoroughly enjoyed the newly in-person State of the Map 2022 in Florence. We attended workshops, talks, and had the opportunity to present our own poster about the accessibility of abortion clinics in Germany, which will be discussed in this post and can be found here.

In the following blog, we’ll explore more information and maps on the topic beyond the poster’s content.


On June 24, 2022, Roe v. Wade was overturned by the U.S. Supreme Court. With this decision, US states can decide independently whether to ban abortions completely. Sooner or later, this will result in a large number of women in the U.S. facing severe barriers to accessing safe abortions.

Coincidentally, on the same day, a law on abortion was repealed in Germany, which is widely perceived as a liberalizing move¹³. Paragraph §219a prohibits physicians from informing and “advertising” about abortion. Doctors are now allowed to provide information on their websites about whether they perform abortions and what methods they use. Previously, doctors themselves had to ensure that they were on the German Medical Association’s list of the doctors willing to perform abortions. Easy and quick access to this information is of enormous importance to the women concerned because, the earlier the abortion is performed, the less invasive it must be¹². Increased public information can also contribute to ending the stigmatization of doctors and the feelings of guilt felt by the women concerned.

Inspired by maps of the New York Times¹ and the Katapult Magazine², we performed an analysis of the accessibility of abortion services in Germany.

Background on abortion in Germany

Abortions are not a rare medical intervention in Germany. As reported by the Federal Statistical Office, a total of 94,596 abortions were registered in 2021. In the first quarter of 2022, 4.8% more abortions were performed than in the first quarter of the previous year. In 2021, the total number of abortions was even 5.4% lower than in 2020. However, abortion is still actually illegal⁴, with the criminal law on abortion codified in Section 218 has existed since 1871. Since 1972 (DDR) and 1976 (BRD), abortion is allowed until the twelfth week of pregnancy⁷. The paragraph further imposes mandatory consultation with a recognized counseling center and a three-day reflection period. After the 12-week period, abortions are allowed only in exceptional cases such as “[..] to avert a danger to life or the danger of a serious impairment of the physical or mental state of health of the pregnant woman [..]”⁹

Abortion laws in the EU are not uniform. Almost all countries allow abortion, but have different regulations regarding the period limit. In the Netherlands, abortion is legally possible up to the 24th week. Every 3rd to 4th woman living in Germany who has an abortion after the first 12 weeks decides to have an abortion in the Netherlands. However, this figure is collected from the reports of doctors rather than an empirical study. The research on so-called “abortion journeys” within the EU is still in its infancy and is funded by the European Research Council¹⁰.

Usually abortion can be performed either by medical or surgical intervention. Surgical abortion (aspiration method) is the most common and mildest variant, which is performed under local or general anesthesia. It should be noted that the patient is not allowed to drive a car afterwards and is therefore dependent on other people or public transport. Since not all people have a supportive environment or do not want others to know about the procedure, public transportation plays a particularly important role in this context. In the case of medical intervention, the pregnancy is terminated by taking the hormone mifepristone. For this method, three visits to the doctor or clinic are necessary¹¹. Due to possible serious side effects, longer journeys can be particularly stressful for the patient so extended travels represent an additional burden. After the third month of pregnancy, the aspiration method is performed in Germany only in isolated cases. Rather, a birth is induced by means of labor pills¹⁰.

Another aspect of abortion is the financial expense of the procedure. Without medical or criminological indication, all costs must be covered by the patient. There is the possibility of financial support from health insurances, but in any case the patient must pay a minimum amount of €200-600 themselves¹¹.

In addition, the number of doctors is declining. For example, in 2020, 39 physicians in Munich had permission to perform abortions. Half of them were over 60 and a few even over 70. In the last 10 years (as of 2020), 20 physicians are no longer available to perform abortions, whereas only six new applications for a permit have been submitted⁶. One gynecologist attributes the lack of successors to prices that are too low for too high a risk. He can charge his patients between €225 and €470 for an abortion, depending on the method. However, the doctor must provide a room, an anesthesia team, materials, anesthetics and a recovery room for the procedure. In addition, he must be available 48 hours after the procedure in case of complications. In the case of a medicated procedure, patients must be seen individually for four hours at all three sessions6.

Figure 1.: Bivariate map on the distribution of women and the number of clinics that perform abortions within a 1-hour driving range.

Method and results

To create our map, we downloaded a list of all clinics and physicians that perform abortions in Germany and are registered with the German Medical Association³. Before §219a was repealed, a central list by the association was created in 2019 to improve the access to information about abortion, which is updated on a monthly basis. Registration for the list is voluntary. As of June 2022, a total of 369 locations are available. For each location, we calculated isochrones with varying travel time intervals using the openrouteservice. A single isochrone depicts the area accessible from a location within a certain time threshold. We assume a best-case scenario with a car driving routing profile. The isochrones were combined with a 1km grid from the 2011 census bearing population and demographic information for each cell. This allows us to estimate the ratio of women and doctors or clinics that can potentially be reached in a time interval for every square kilometer in Germany. Based on additional information for each clinic or physicist, we have created two more such maps: one that considers only those locations that indicates the ability to speak a foreign language and another one that includes only locations that perform surgical abortions. Surgical abortions can be the only option for procedures conducted later in the pregnancy, often past the 9-week mark.

Figure 2.: Accessibility of clinics that indicate the ability to speak at least one foreign language.

Figure 3.: Accessibility of clinics that indicate the availability of the surgical abortion method.

Most metropolitan areas have a high density of access to abortion services. Figure 1-3 all use the same symbology to visualize the combination of both the distribution of women and the distribution of doctors within a 1-hour car driving distance. Figure 2 and 3, however, use filtered clinic locations. The brighter the map unit, the fewer doctors available. The darker the red, the more women inhabit the area. The bivariate map shows poor coverage in large areas in the north and east of Bavaria. In general, the south stands out with the largest gaps in coverage. Densely-populated areas such as Nuremberg, Freiburg and the Lake Constance region are each covered by a single clinic only. Still, the coverage looks worse when we consider Figures 2 and 3 with the filtered clinics. For access to clinics that indicate the ability to speak at least one foreign language, coverage in rural areas is very thin to non-existent. Areas between Berlin and Hamburg and all of Bavaria except for Munich are not covered. Only access coverage for clinics that perform the surgical method is poorer. Filtered for this property, only four clinics remain in the entire South.


It is important to note that in this analysis, we refer only to those 368 clinics and physicians registered with the federal medical association. It is possible that other physicians perform abortions without advertising them before §219a was abolished. The completeness of the clinics’ characteristics regarding foreign language skills and abortion methods are subject to uncertainty. In our analysis, we only looked at those clinics that clearly provided information on one of these characteristics, while many clinics did not provide such information and were omitted from the analysis. Additionally, our accessibility model assumes a car routing profile, which certainly does not reflect the average patient of an abortion clinic. Therefore, the analysis at hand investigates a best-case scenario for reaching a location for safe abortions. Other important dimensions of access, such as social or financial, were not considered.

Related works:


[1] https://www.nytimes.com/interactive/2021/05/18/upshot/abortion-laws-roe-wade-states.html

[2] https://katapult-magazin.de/en/article/wer-abtreiben-will-macht-es-auch-illegal

[3] https://www.bundesaerztekammer.de/themen/aerzte/schwangerschaftsabbruch

[4] https://www.destatis.de/DE/Themen/Gesellschaft-Umwelt/Gesundheit/Schwangerschaftsabbrueche/_inhalt.html

[5] https://www.deutschlandfunk.de/schwangerschaftsabbrueche-in-deutschland-warum-immer-100.html

[6] https://www.sueddeutsche.de/muenchen/muenchen-abtreibung-aerzte-mangel-1.5121832

[7] https://www.bpb.de/shop/zeitschriften/apuz/290795/kurze-geschichte-des-paragrafen-218-strafgesetzbuch/

[8] https://www.gesetze-im-internet.de/stgb/__218.html

[9] https://www.gesetze-im-internet.de/stgb/__218a.html

[10] https://taz.de/Spaetabtreibungen-in-Deutschland/!5681768/

[11] https://www.profamilia.de/themen/schwangerschaftsabbruch/

[12] https://www.spiegel.de/politik/deutschland/abtreibung-abschaffung-von-paragraf-219a-fuer-die-muendige-frau-kommentar-a-784cd403-f279-4124-a216-e320042d1719

[13] https://www.tagesschau.de/inland/219a-gestrichen-101.html

« Newer Posts - Older Posts »