How an algorithm from the 80s sets the new standard for modern spatial indices

Geospatial indices are all around us. They allow us to search through millions of points in an instant answering questions such as “find me the closest bike repair shop” efficiently.

And yet there are still forgotten gems in the archives of computational geometry: Space-filling curve based spatial indices. With an optimization going back to the year 1981 this spatial index delivers surprising efficiency; and yet it is rarely discussed in the geospatial community.

Let’s go back in time and rediscover how we can build a spatial index on top of a space-filling curve to accelerate geospatial queries and simplify geospatial indexing for location-based applications.

Small note based on reader feedback: If you know about Z-Order Curves or Hilbert Curves already I still recommend you reading this as the post below will show you how use Z-Order Curves efficiently as a spatial index and that is something you probably don’t know yet.

The Z-Order Curve

The Z-Order Curve is one of many space-filling curves transforming multi-dimensional data into a single dimension while preserving locality. In the geospatial domain it allows us for example to transform longitude and latitude into a single number while preserving geographic proximity.

For spatial queries such as “find me the closest bike repair shop” we want to find points within a rectangular area. To search a rectangular area on the Z-Order Curve (highlighted in gray below) we start at the top left Z value of the rectangular area and walk the curve until we’re at the bottom right.

Unfortunately the Z-Order Curve exhibits discontinuities between each Z block. If we walk the curve we might accidentally come across such a discontinuity and include irrelevant ranges:

In the example above when we walk the curve we exit the rectangular area (in blue) on its right and after scanning a vast amount of irrelevant data (in red) we enter it again on the left. In practice this means for some queries our spatial index is orders of magnitudes slower having to walk through vasts amounts of irrelevant ranges.

Fortunately there is an optimization from the ’80s pruning the search space from those irrelevant ranges.

BIGMIN, Tropf and Herzog, 1981

The year is 1981 and there’s still eight years until the release of Belgian techno anthem “Pump Up the Jam”.

In 1981 Tropf and Herzog show how to immediately jump back into a rectangular area on the Z-Order Curve:

Multidimensional Range Search in Dynamically Balanced Trees, H. Tropf, H. Herzog

I won’t go into the implementation details here (see references at the end) but the next Z value that jums back into the rectangular area can be computed efficiently based on the Z-Order Curve and the query rectangular area’s top left and bottom right Z values: It’s a function of those three numbers.

This optimization, called “BIGMIN”, or in later years by others also called GetNextZ-Address, or jumpNextIn, allows us to prune the search space and turn any spatial query into a few very efficient linear memory scans.

Spatial Index: Z-Order Curve + BIGMIN

Equipped with a Z-Order Curve and the BIGMIN optimization, a spatial point index can now be implemented as follows

1. Sort points by their value on the Z-Order Curve
2. For a nearest neighbor query within a rectangular area
   1. Find the top left Z value (zmin) and bottom right Z value (zmax) e.g. with two binary searches on the sorted points (highlighted in gray below)
   2. Start the linear scan through the range [zmin, zmax] as all nearest neighbors are within this range by design (highlighted in blue below)
   3. Whenever we leave the rectangular area compute BIGMIN and immediately jump to it, skipping over irrelevant ranges (highlighted in red below)

In the example the query turns into scanning eight points in memory, then computing BIGMIN to skip over an irrelevant range, and then scanning another eight points in memory.

Building the spatial index is a simple sort. Querying the spatial index is two binary searches followed by linear memory scans and potentially binary searches when having to jump back in.

Incredibly efficient in 1981, incredibly efficient in 2025.

Summary

We’ve gone back in time and rediscovered how we can build a spatial index on top of a space-filling curve using an optimization technique from the ’80s.

The geospatial community has mostly overlooked this historical algorithm. Revisiting the Z-Order Curve based spatial index reveals its potential; and we’re only touching the surface with point queries.

There is more work to do: The Z-Order Curve and the BIGMIN optimization generalize to arbitrary dimensions. Can we make use of this e.g. to index line segments? If we index a line segment’s mid point, angle, and distance on a four dimensional Z-Order Curve we could use the very same technique to accelerate geospatial queries such as “find nearby roads pointing south-west”.

An an example while working through this I have implemented the Z-Order Curve based spatial point index in tinygraph to power a graph routing engine’s location lookups.

I encourage the geospatial community to revive this forgotton gem together.

References

On spatial search algorithms

• https://blog.mapbox.com/a-dive-into-spatial-search-algorithms-ebd0c5e39d2a

On Z-Order Curve and its relationship to space-partitioning trees

• https://en.wikipedia.org/wiki/Z-order_curve
• https://en.wikipedia.org/wiki/Quadtree
• https://en.wikipedia.org/wiki/Quadtree#Compressed_quadtrees

Implementation in the tinygraph project

• https://github.com/tinygraph/tinygraph/pull/68
• https://github.com/tinygraph/tinygraph/issues/22
• https://github.com/tinygraph/tinygraph/issues/70

Z-Order Curves for Amazon DynamoDB

• Z-Order Indexing for Multifaceted Queries in Amazon DynamoDB 1 2

Going deeper on space-filling curve based indices

• Using a Space Filling Curve for the Management of Dynamic Point Cloud Data in a Relational DBMS 1 2 3
• Z-ordered Range Refinement for Multi-dimensional Range Queries
• Towards a general-purpose, multidimensional index: Integration, Optimization, and Enhancement of UB-Trees

Interactive Z-Order Curve visualization, the screen grabs above are from here

• https://blocks.roadtolarissa.com/daniarleagk/278709dedf09451b794ff72c4c05cda1

Implementation: BIGMIN, GetNextZ-Address, jumpNextIn in the wild

• geomesa
• PGM-Index
• zCurve

Nice service

The State of the Map Organizing Committee provides a Travel Grant Programme (TGP) to facilitate accessibility and diversity at the global SotM 2025 Conference, which will happen in Manila, Philippines, October 3-5.
Taking into consideration the success of last year, the programme will also dedicate a portion of the budget available to support remote attendance (e.g. by providing access to a reliable internet connection) for those who cannot travel to Manila because of imposed restrictions, travel costs, or other reasons.

The call for applications to the Travel Grant Programme will be open until May 16,12:00 UTC.

It is fundamental to read all the information on this page before submitting an application.

For applicants

This section sets out the benefits of a Travel Grant to attend SotM 2025 and the process and criteria by which applicants will be selected. The TGP is administered by the State of the Map Organizing Committee.

Travel Grants

The Travel Grant will include a full conference ticket and a lump-sum contribution towards travel, accommodation and other expenses.

There will be five (5) different sizes of travel grants available:

The size and number of grants may be adjusted at the discretion of the Organizing Committee, depending upon the level of funding raised and the number of applications received. Applicants will only be able to apply for one size of travel grant, and they should apply for the grant depending on their country of residence. The World Bank rating is available online

Anyone applying for more than one grant will be excluded from the application process.

For categories C, D and E, a maximum of 2 winners will be chosen for each country of origin.

Online Attendance Support Grant

For those interested in attending the conference remotely but with limited internet connectivity, smaller grants will be made available to subsidise the cost of purchasing mobile data passes. The size of each grant will vary depending on the country and mobile data packages available, but it’s expected not to exceed £ 30.

With an estimated 1GB per hour of streaming videos in HD quality (720p), it’s expected that a maximum of 40GB will be sufficient to cover the entire conference. Arrangements will be made with grant recipients to purchase the data packages directly or to transfer the amount needed via mobile money.

Criteria

In order to guarantee a transparent and fair allocation of funds, the travel grant programme uses a points system for ranking applications. Points are awarded if applicants match specific criteria. After the application phase is over, all applicants are ranked by points, and the applicants with the most points will receive funding.

Please note that the travel grant programme aims to enhance diversity at the conference. We might reserve a few programme spots for people with unique and outstanding applications. However, most programme spots will solely be determined by the point ranking.

To ensure the participation of OSM contributors who have not attended any in-person SOTM conference, the in-person Travel Grant winners of the previous two SOTM (SOTM 2024 Nairobi and SOTM 2022 Firenze) will not be eligible for this Travel Grant.

Furthermore, once the ranking has been drawn up and the scholarships assigned based on it. The next two people for each grant size will be informed of their position. If one of the first winners communicates by the 22nd of August that he will not be able to participate, the scholarship will be provided to the people who are in this second ranking. However, those people will receive a free online ticket to attend the conference virtually. Moreover, if they will be able to attend the conference in person, they could receive a free ticket if they provide support volunteering.

Applicants receive points for the following criteria:

Nationality

The allocation of points is based on the World Bank country rating by income.

• Applicants from Low-income economies receive 1.5 points.
• Applicants from Lower-middle-income economies receive 1 point.
• Applicants from Upper-middle-income economies receive 0.5 points.
• Applicants from High-income economies receive 0 points.

Residence

The allocation of points is based on the World Bank country rating by income.

• Applicants who are currently living in Low-income economies receive 1.5 points.
• Applicants who are currently living in Lower-middle-income economies receive 1 point.
• Applicants who are currently living in Upper-middle-income economies receive 0.5 points.
• Applicants who are currently living in High-income economies receive 0 points.

Gender and Minorities

In order to create a more diverse community, women and applicants from other underrepresented groups will receive 2 points. The minority criteria will be considered with respect to the other candidates of the same category.

Students

Students, including PhD students, receive 1 point. You may be asked to provide evidence of enrollment.

First State of the Map

Applicants who attend SotM for the first time receive 1 point.

Amount required
The expected cost amount requested is to indicate in pounds sterling (£). Applications without an explicit request of the amount required will not be considered. The minimum between the expected cost and the grant size based on the residence country will be assigned.

OpenStreetMap and Community
Details of their past contributions to OpenStreetMap projects or community activities: up to a total of 4 points. Up to 2 points for mapping activities based on your OpenStreetMap username and up to 2 points for other contributions to the project as you will describe.

You & OpenStreetMap

Present the most valuable experience, mapping activities, or community moment that describes the travel grant experience in the OpenStreetMap world. This is not the call for participation; the call will be open, and there will be the possibility to submit talks and workshops. So, this is not an instrument to present talks; you are invited to participate in the call for talks and workshops if you would like to present your activities.
Please provide a short abstract about what you would like to present, with a maximum length of 100 words. No points; the abstract will be used as a discriminant for the final decision.

Volunteering

The travel grant winners will be asked to volunteer for a time proportional to 1/3 or less of the conference length, depending on the number of other volunteers. Indicate the roles that you are confident to do: Live Video Cutting, Remote/Pre-Recorded Talk Assistant, Session Assistant and Venueless Assistant (only for remote grants). For more detailed info on the roles, see the wiki pages of volunteers for SotM 2024. No points.

Payment

You must be able to fund the direct costs of your travel to SotM and be in possession of a valid passport and entry visa and other documents (for example, vaccinations). Successful applicants will receive a conference pass and become eligible for reimbursement of expenses up to the limit of the travel grant when they check in at the conference registration desk. Reimbursement of expenses will be by electronic means (details to be advised). Reimbursement may take up to two months after the conference date, provided that the travel grantee has submitted all the necessary information before the conference commences.

Only in the case of demonstrated needs and exceptional cases will the travel grant amount be provided in advance or used to cover specific travel costs directly.

How to Apply

If you wish to be considered for a SotM travel grant, please fill out the application form.

All data will be held confidential and only used for assessing TGP applications. Please ensure that the email address that you provide is regularly monitored, as the selection committee may wish to seek clarification of responses. If no replies have been provided to requests via mail for the Travel Grant Committee within 5 days, the Travel Grant will be removed from you and assigned to the next applicants in the ranking.

Applications must be received by the May 16 at 12:00 UTC.

The selection committee will aim to notify recipients of grants by the May 31.

Supporting the Travel Grant Programme

The travel grant programme is funded through the State of the Map Sponsors.

The State of the Map Working Group

The State of the Map conference is the annual, international conference of OpenStreetMap, organised by the OpenStreetMap Foundation. The OpenStreetMap Foundation is a not-for-profit organisation, formed to support the OpenStreetMap Project. It is dedicated to encouraging the growth, development and distribution of free geospatial data for anyone to use and share. The OpenStreetMap Foundation owns and maintains the infrastructure of the OpenStreetMap project, is financially supported by membership fees and donations, and organises the annual, international State of the Map conference. Our volunteer Working Groups and small core staff work to support the OpenStreetMap project. Join the OpenStreetMap Foundation for just £15 a year or for free if you are an active OpenStreetMap contributor.

OpenStreetMap was founded in 2004 and is an international project to create a free map of the world. To do so, we, thousands of volunteers, collect data about roads, railways, rivers, forests, buildings and a lot more worldwide. Our map data can be downloaded for free by everyone and used for any purpose – including commercial usage. It is possible to produce your own maps which highlight certain features, to calculate routes etc. OpenStreetMap is increasingly used when one needs maps which can be very quickly, or easily, updated.

Test

Overpass turbo https://overpass-turbo.eu/?Q=%5Bout%3Ajson%5D%5Btimeout%3A25%5D%3B%0A%2F%2F%20Gather%20results%20within%20the%20current%20map%20view%0A%28%0A%20%20node%5B%7E%22%5Eaddr%3A.*%22%7E%22.%22%5D%28%7B%7Bbbox%7D%7D%29%3B%0A%29%3B%0A%2F%2F%20Print%20results%0Aout%20body%3B%0A%3E%3B%0Aout%20skel%20qt%3B&C=37.317027%3B-122.072031%3B16

At this point I’ve officially mapped out half of Bell Island, Newfoundland. Although I’m geographically halfway, the southern portion which I’ve mapped out is much less geospatially dense than the northern portion, so I’d estimate I’m about a third of the way through this project.

The ID Editor has been very intuitive so far in this process, and I have also learned a bit on how to use the JOSM editor for more specific geospatial processes. One thing I have yet to figure out is the display order of overlapping features, however I believe this is tweaked through the relations mechanic in the ID editor.

So far the minimal existing geospatial data that is present in this area has been accurate for the most part, with only slight alterations required to match the current satellite data that I’m basing my new features off of. That being said, quality varies, and it’s clear that some of these features were one-off additions by likely inexperienced users.

I think I’ll be able to keep up this mapping pace for the foreseeable future, so this project will hopefully get within the mopping up phase within a couple of months.

Belfast Maine - Little City by the Water

Hello! I have just started getting into OpenStreetMap in the last few days, I absolutely love the way it works, and love to map out different towns. I lived in the town of Belfast, Maine for many years and visit this town so it holds a special part to me.

I’ve been adding all the “Downtown” houses that aren’t mapped, they are just Address points, and different new Houses, roads ETC. I just found out last night that I can actually MERGE address points with areas for houses so it’ll speed the process up a lot for me.

Whoots is a simple tile server proxy for WMS servers. WMS > TMS. So if you have an application that only works with ZYX Google-style tiles and all you have is a WMS server, you can use it to re-route the request.

It was created way back in 2010! Here’s the post announcing it: WhooTS a small wms to tile proxy – WMS in Potlatch

There’s been few recent changes.

• Some validations to the code was added to make it a bit more secure.
• image/png and image/jpeg will now work. Defaults to png. Optionally pass in ?format=image/jpeg for jpeg
• You don’t need to have a map= param in the URL for it to work now.
• Puma server configs added
• new php port of the code
• The server at whoots.mapwarper.net was moved to a shared host and is now running the php version

The code is at https://github.com/timwaters/whoots

I have mapped out multiple different areas in Kilmeaden, Portlaw and Kilmacthomas. Quite a bit of these areas were already mapped out so I modified and updated many roads and buildings.

I have mapped out different areas in Kilmeaden and Portlaw. Quite a bit of these areas were already mapped out so I modified and updated many roads and buildings.

Khumalo 01

Overview

This is inspired by laambda19’s diary post on their mapping project, I want to document my work the same way with clear examples of before and after, adding more detail to the Bulawayo map. Aerial photography and local knowledge are the main driving force in mapping. There is scarce information available online, what little there is often requires a Facebook account.

Changes made

*I added colour to buildings and roofs

before & after

Nothing to see on Carto renderers but in simple3D renderers

This is a bug fix release which contains the following:

• refuse to update database when the flatnode is missing
• fix regression where writing to tables without a managed id column was not possible

Templo parroquial

Overpass turbo code https://overpass-turbo.eu/?Q=%5Bout%3Ajson%5D%5Btimeout%3A25%5D%3B%0A%28%0A%20%20way%0A%20%20%20%20%5B%22tunnel%22%5D%0A%20%20%20%20%5B%22tunnel%22%7E%22%5Eyes%24%7C%5Eculvert%24%22%5D%0A%20%20%20%20%5B%21%22culvert%22%5D%0A%20%20%20%20%28%7B%7Bbbox%7D%7D%29%3B%0A%20%20relation%0A%20%20%20%20%5B%22tunnel%22%5D%0A%20%20%20%20%5B%22tunnel%22%7E%22%5Eyes%24%7C%5Eculvert%24%22%5D%0A%20%20%20%20%5B%21%22culvert%22%5D%0A%20%20%20%20%28%7B%7Bbbox%7D%7D%29%3B%0A%29%3B%0Aout%20body%3B%0A%3E%3B%0Aout%20skel%20qt%3B&C=37.34314%3B-121.941376%3B11

This is a overpass turbo script https://overpass-turbo.eu/index.html?Q=%5Bout%3Ajson%5D%5Btimeout%3A25%5D%3B%0A%2F%2F%20fetch%20waterways%20excluding%20those%20tagged%20intermittent%3Dyes%0A%28%0A%20%20way%5B%22waterway%22%7E%22%5E%28stream%7Cditch%7Criver%7Cdrain%7Ccanal%7Cyes%29%24%22%5D%5B%21%22intermittent%22%5D%28%7B%7Bbbox%7D%7D%29%3B%0A%20%20relation%5B%22waterway%22%7E%22%5E%28stream%7Cditch%7Criver%7Cdrain%7Ccanal%7Cyes%29%24%22%5D%5B%21%22intermittent%22%5D%28%7B%7Bbbox%7D%7D%29%3B%0A%29%3B%0Aout%20body%3B%0A%3E%3B%0Aout%20skel%20qt%3B

Zamboanga city

03/04/2025-09/04/2025

Mapping

• Please cast your vote in the proposal to map grab rails in (wheelchair accessible) toilets.
• Comments are being requested for these proposals:
  • A tagging scheme proposal for windmills and watermills that aims to clarify the distinction between building=windmill and man_made=windmill. It also proposes the introduction of a new key, windmill=*, to further refine the classification of these structures.
  • armrest:*=* for tagging whether of not a toilet is equipped with arm rests.

Community

• Pedro Tharg wrote ►, in his diary, about using MapBiomas data in OpenStreetMap and suggested tagging for different land cover and land use classes. MapBiomas is a collective project that officially began in Brazil in July 2015 and is organised collaboratively by NGOs, universities, laboratories, and technology start-ups.
• Looking for a holiday destination for hiking or cycling? How about a region where many paths in OpenStreetMap don’t yet have any surface information? Nakaner’s presentation at FOSSGIS 2025 and the maps he produced will show you where your help is needed. The slide deck from the presentation and the map are available, and you can watch his video.

OpenStreetMap Foundation

• Michael Spreng, from the OpenStreetMap Foundation’s Membership Working Group, announced that OSMF members will now need to renew their memberships manually, as the foundation has temporarily halted automatic renewals via PayPal. The decision follows ongoing issues with the automated system, where payments often fail to register correctly.

Events

• Dani Waltersdorfer, a board member of the OpenStreetMap Foundation, has shared her insights following her participation in the State of the Map Latin America 2024 conference, held in Belém, Brazil.
• FOSSGIS announced ► the successful completion of the 2025 FOSSGIS Conference, held in Münster from 26 to 29 March. Co-organised with the OpenStreetMap community and the University of Münster’s Institute for Geoinformatics, the event took place at Münster Castle, with a large tent on the castle square accommodating attendees and hosting exhibits and posters. Over 1000 participants joined, including 750 on-site and more than 350 online.
• The recordings of the FOSSGIS 2025 conference are available . These include vehicle localisation in DB Regio [1] , coordinate reference systems for D-A-CH [2] and 2.5D indoor maps based on OpenStreetMap data [3] .
• Séverin Ménard will conduct a virtual workshop, in English, on mapping building damage using post-disaster imagery in OpenStreetMap on Thursday 17 April at 7:00 PM UTC via BigBlueButton. The session will cover mapping techniques using JOSM and mark the kick-off of a mapathon focused on the French overseas department of Mayotte, which was damaged by Cyclone Chido at the end of 2024.
• Volker Krause shared his insights and personal notes on OSM indoor navigation, the Transitous public transport routing service, and OpenTransport, following his participation in the FOSSGIS Conference 2025.
• The State of the Map Organising Committee has launched a travel grant programme, aimed at enhancing accessibility for participants attending the State of the Map 2025 Conference, set to take place in Manila from 3 to 5 October. Applications for the grant are now open and will be accepted until Friday 16 May at 12:00 UTC.

Education

• Mapeducação, a programme that aims to map public and private schools and educational institutions has started ► in the Municipality of Jacobina, Bahia, Brazil. The programme aims to facilitate access to the schools’ data, helping to facilitate the enrolment process, for example.

Maps

• OpenLocalMap is a map of Catalonia originally created [ca] from a project run [ca]► by the non-profit association la Konfraria de la Vila del Pingüí, from la Palma de Cervelló. Now maintained by OSM-Catalan, the map incorporates data from la Palma de Cervelló and Vilanova i la Geltrú from OpenStreetMap. The code is available on GitHub.

OSM in action

• ^[1] Herr Voeglein tooted that the latest issue of Fahrrad Zukunft featured ► an in-depth article on BRouter, a customisable, elevation-aware bike routing app that uses OSM data. The article also outlined how to connect the BRouter Android app to Osmand, OruxMaps, or LocusMap.

Software

• Jake Low’s pull request has resulted in OSMCha now getting changes from an OSM adiff service, rather than using the Overpass API. This code change should result in OSMCha being able to handle displaying larger changesets more quickly.
• Ralph Straumann reported that Fabian Rechsteiner has developed a performance benchmark comparing six open-source vector tile servers that are used to deliver tiles from a PostgreSQL/PostGIS database.

Programming

• Nick Walker has added Overpass QL syntax highlighting support for .overpassql files on GitHub. It is based on the Overpass QL syntax VS Code extension and was made possible by standardising the file extension (we reported earlier).
• Kumakyoo continued his series on his new OMA file format with a description of how relations are handled.

Releases

• MapLibre has published its March Newsletter with notices about new releases for the MapLibre GL JS and the Martin Tile Server.
• Version 2.1.0 of osm2pgsql has been released, bringing several enhancements aimed at better integration with connection poolers and other database middleware.

Did you know that …

• … the OSM welcome tool can be used to monitor and message new contributors in your region? It’s a great help to send a welcome message to new OSM members and inform them about local chat groups and such. Messages like this can increase the likelihood of those users sticking around and continuing to contribute to OSM.

Other “geo” things

• Between science and art, the ‘Forest Cartographies’ exhibition is showcasing ► HD LiDAR data, from the Marne forests, from 8 April to 28 June in Reims (France).
• Several car accidents linked to Google Maps have occurred ► during Indonesia’s recent Mudik season, a massive annual homecoming ahead of Eid al-Fitr, involving millions of travellers. A driver in Ponorogo ended up stranded in a rice field after following Google Maps directions onto what turned out to be a non-public road. In a separate incident in Gresik, a BMW sedan drove off the edge of an unfinished ► toll road.
• EOS Landviewer offers users free access to 10 satellite imagery scenes per day, including recent high-resolution data from Sentinel-2 L2A and Landsat 8 RT.
• Christopher Beddow discussed the distinction between discovery and rediscovery, arguing that the act of digitising elements of the physical world constitutes a form of rediscovery. By transforming tangible realities into digital formats, knowledge once confined to a select few becomes accessible to all, unlocking the potential for instant, widespread awareness.
• Texty.org.ua has published ► an analysis on the disappearing snow in Ukraine, noting that the last snowy winter across much of the country occurred in 2020–2021, with the most widespread and heavy snowfall recorded in 2016–2017. Citing data from MODIS satellite imagery, they highlight a cyclical trend; snowier winters tend to follow several years with little to no snow.
• The BnF (French National Library) has been backing up ► French websites since 2006 by automated collection, in a similar way to the Internet Archive. Pierre Bonneau and Dorothée Benhamou-Suesser (from the BnF) noted that this method is very effective for many sites but not for dynamic sites such as OSM. Even when queries are programmed, the collection and output are generally of poor quality or even non-existent. Christian Quest replied that it has been proposed in the past to provide VMs (virtual machines) of the tile generator, with the data frozen at 1 January and using the map style of the time, as demonstrated in this prototype. The rest of the discussion identified the most suitable solution as a VM for each past and future year, using an LXC container.
• Charlie Plett has created an interactive historical map of Mennonite villages powered by OpenHistoricalMap and MapLibre.

Upcoming Events

Where	What	Online	When	Country
København	OSMmapperCPH		2025-04-13
Delhi	15th OSM Delhi Mapping Party		2025-04-13
Grenoble	Atelier adressage dans OpenStreetMap		2025-04-14
中正區	OpenStreetMap x Wikidata Taipei #75		2025-04-14
	Missing Maps London: (Online) Mid-Month Mapathon [eng]		2025-04-15
Lyon	Réunion du groupe local de Lyon		2025-04-15
Bonn	187. OSM-Stammtisch Bonn		2025-04-15
San Jose	South Bay Map Night	✓	2025-04-16
Lubbock	A Synesthete’s Atlas: Cartographic Improvisations		2025-04-16
Lüneburg	Lüneburger Mappertreffen	✓	2025-04-15
Amsterdam	Maptime Amsterdam: Springtime Mapping Party		2025-04-16
Marche-en-Famenne	OpenStreetMap Belgium au salon Municipalia		2025-04-17 – 2025-04-18
	MapRVA Map & Yap		2025-04-18
	Workshop – mapathon: post-cyclone assessment of building damage		2025-04-17
	OSMF Engineering Working Group meeting		2025-04-18
Mumbai City	2nd Mumbai Mapping Party		2025-04-20
Aketi	Mapathon for Kinshasa City,		2025-04-21 – 2025-04-25
Hannover	OSM-Stammtisch Hannover		2025-04-22
Kiel	Kieler Mapper*innentreffen		2025-04-22
Derby	East Midlands pub meet-up		2025-04-22
	Missing Maps – DRK & MSF Online Mapathon		2025-04-23
	[Online] OpenStreetMap Foundation board of Directors – public videomeeting		2025-04-24
Zaragoza	Mapatón Humanitario Universidad de Zaragoza		2025-04-24
	Atelier – mapathon: évaluation post-cyclonique des dommages aux bâtiments		2025-04-24
	SOSM Annual General Assembly		2025-04-25
	OSMF Affiliation Focus Group Discussion: Thematic and non-geographical groups		2025-04-26

Note:
If you like to see your event here, please put it into the OSM calendar. Only data which is there, will appear in weeklyOSM.

This weeklyOSM was produced by MatthiasMatthias, Raquel Dezidério Souto, SeverinGeo, Strubbl, TheSwavu, YoViajo, barefootstache, derFred.
We welcome link suggestions for the next issue via this form and look forward to your contributions.

I’ve suggested an updated version of the diagrams explaining the tagging of railway stations. The goal is to make them simpler to understand, especially for newcomers, without losing the core meaning. There have already been some feedback and interesting ideas for further improvements.

I invite you to join the discussion — your input would be highly appreciated!

Yanzi (燕子镇) Mapping Project

Overview

This is the first entry in a series of Chineese towns I plan to map out.

Considering the lack of mapping detail in the vast majority of China, I plan to add detail to those places which lack even the basics, such as Yanzi. At first, only the location of the town was mapped as well as a VERY rough outline of the residential landuse.

If there are any mistakes, please tell me! I tried my best to overcome the language barrier

Changes Made

1. Fixed the town name
2. Added road network
3. Added buildings
4. Added nearby villages/hamlets
5. Added bodies of water and the Yanzixi Reservouir
6. Various amenities (W.I.P.)

To-Do List

1. Continue adding amenities
2. Add nature landuse

Before

After

Stats (12.04.25)

• 19 Changesets
• 13 175 Changes
• 9 Mapping days
• 693,4 Changes/Changeset (on Avg.)
• 2,1 Changesets/Day (on Avg.)
• 1463,8 Changes/Day (on. Avg.)

Greetings OSM community,

The Humanitarian OpenStreetMap Team (HOT) adapted its governance structure last year to open seats on the Board of Directors for non-members. We are now recruiting two (2) Appointed Directors. Applications are open until April 24, 2025 — visit HOT Careers for more information and to apply.

The main reason for introducing Appointed (as opposed to solely Elected) Directors is to bring in people—and skill sets—that are not currently in abundance within our voting membership. Of course, a passion for geospatial technology and data in humanitarian, disaster, and climate response is still desirable. However, we are particularly seeking individuals with expertise in financial sustainability, diversified fundraising, and innovative hybrid business models. Ideal candidates will have a proven track record of leading impact-driven organizations through strategic growth and transformation.

As the current Secretary of the HOT Board of Directors, I’m happy to answer any questions about this opportunity. We’re excited to welcome new perspectives and strengths that complement our existing Board, which brings deep OSM/HOT history and expertise.

Best regards, =Russ

Use Russell.Deffner (at) hotosm (dot) org if you’d like to email inquiries related to this position.

Cover

On Tuesday we had a London pub meet-up. We are still doing this every couple of months. I stopped writing diary entries about them, but for some reason (for your enjoyment!) I’ve decided to write about this one. So here goes:

We were at the Jack Horner Pub. It’s becoming the default choice in this bit of London where previously I would’ve chosen the Blue Posts (Sam Smiths pub. I still quite like them but there are reasons not to). Meanwhile Fullers make it quite easy to book a table on their website(s), so in uncharacteristically organised fashion, I booked table! but characteristically I arrived quite late. I only booked for 8 people, and that turned out to be about right on this occasion. A small intimate gathering.

We had along Json Singh from india, who just moved to London. I like the way he’s named himself after a file format (I was forgetting to ask him about that). So we talked to him about the state of mapping in India. It sounds like it’s progressing well with lively communities in the cities, although sometimes struggling to build formal organisations, partly because it’s a little bit illegal there, so has to remain a bit underground. That’s mostly because of tensions around showing of borders. I wonder whether vector tiles will soon allow us to do the same slight-of-hand trick that other map providers do, showing different border positions depending on your IP address location. It’s kind of comical how well that trick seems to work to take the heat off. Another thing which is daft, is that there’s no official openly published data on the border position which officially must be shown on maps of india. That would be a basic way of helping assert a particular desired border position. But no.

We talked about some mapping apps. Derick did a little demo of everydoor, including how to add a new POI. I’ve never done that. Personally I didn’t really adopt everydoor yet. Initially I imagined it didn’t add much value for experienced mappers who know how to use the Vespucci app. But others started raving about it. I think I’m starting to get why. A while ago I installed the OSMfocus android app, and recently found myself using that while looking along dense shopping street a bit like in this video. To my surprise I realised that this was better for checking shops in maintenance-mode. Where previously I was fiddling around with Vespucci, going in and out of each POIs/shop outline to check the tags on it. The OSMFocus app has a “show all the data, all at once” approach. It has crosshairs in the middle of the screen, and the data gets listed which is nearest to that. Everydoor is the same! And it looks like Vespucci recently redesigned the default interface to have an everydoor-like list. In a way though OSMfocus takes the idea to the logical extreme. It’s kind of crude, but it uses a rainbow of colour and even connecting lines (easier than numbers) to help you tell which list of tags is for which shop.

In the pub I was floating the idea of a “Notathon” mapping party. I’ve been looking at notes (Notes) in central London recently. Obviously there’s a lot of them (map) (Did anyone make a map that show’s more than 1000 on the screen at once?), but the thing I find interesting, is that the notes which remain open have been weeded out pretty well. The spam gets removed. The nonsensical ones get removed. The ones which can be resolved by knowing how to edit, or by doing a bit of research, tend to get resolved, and so actually what we’re left with is something we need to survey. It’s surprisingly good quality information in a way. Or so my argument went. Derick was not convinced. In the pub we used my Notes KML downloader tool for OrganicMaps to take a look at notes near where he lives in Kilburn. I think this weeding out of useless notes hasn’t been as thorough outside the centre, because he did find some cruft, although it was mostly stuff that he was irritated by because he had re-mapped the thing himself already (So the note was not wrong, but not up-to-date with local mapping he was already on top of). But that’s easy for him to fix quickly, and satisfying no? Derick was not convinced, but personally I’ve been looking for a long time at pushing back my “notes free radius” from where I live, and I think we need to start doing this around the pubs we drink in. Or maybe a good old cake diagram for a bigger note fixing session!

I don’t have a date to announce just yet. If we decide on a notathon, or some other kind of event, I will put details on the London page and on OSMLondon mastodon.

This blog post is part of a series of blog posts about the new OSM file format “OMA”. This is the sixth post. At the end of the article you’ll find links to the other blog entries.

The Oma file format uses a set of bounding boxes to sort the data geographically. The default set used by the converter consists of three grids.

The Default Grids

The first grid uses a mesh size of 1 degree in both directions between -45° and 45° latitude. Towards the poles the mesh size increases and finally the area around the poles is put into a single bounding box.

The grid looks like this:

The second grid uses a mesh size of 10 degrees in both directions between -80° and 80° latitude. It does not include the areas around the poles.

This grid looks like this:

The third grid isn’t really a grid, it’s just the whole world as one big bounding box.¹

Data Distribution in Bounding Boxes

In practice, data isn’t evenly distributed around the globe. Large areas are covered by oceans where there is little data distribution, while other regions are densely covered.

Using identical bounding boxes everywhere results in a few large chunks and many small chunks. It would be better to balance the sizes. This would speed up data access.

As an example, look at the distribution of nodes in Germany:

The Ruhr area (centre, left) contains about 10% of Germany’s nodes. And the area around Berlin contains another 8.6%, while for example the area just north of Berlin contains only 1.4%.

Let’s see, how this affects the search times: I choose the street Bienenweg in Neubrandenburg for comparison. The search for this street takes 2.5 seconds. Compare this with the 4.7 seconds needed for the Viktorstraße in Wuppertal.

Improving Bounding Boxes

But how do you get a better set of bounding boxes? I tried some ad hoc algorithms, but could not find out, what really makes a good set of bounding boxes. Probably, this question could make a good master thesis (or even a PhD thesis).

Anyway, I’d like to take a quick look at what I tried with data distribution: The first idea led to this set of bounding boxes:

The nodes are still not evenly distributed, but the differences are somewhat smaller. Searching for the Bienenweg now takes 2.8 seconds, while searching for Viktorstraße has been reduced to 2.9 seconds. On average, this seems to be at least some improvement.

I wondered if overlapping bounding boxes were a problem. So I tried the same thing with (almost) non-overlapping bounding boxes. The result looks like this:

The distribution looks similar to before. Unfortunately, the search times are worse: still 2.8 seconds for the Bienenweg, but 3.5 seconds for the Viktorstraße.

I also tried to increase the number of bounding boxes. It seems that this improves the results slightly. Unfortunately, the more bounding boxes you use, the more difficult it is to create the Oma files (crashes happen more often). I fear that the algorithm for sorting data into bounding boxes will have to be improved before further experiments can be made with this.

In one of the comments to an earlier post Geonick pointed out a discussion about geoparquet.² In that discussion, GitHub user jiayuasu claims that an algorithm called KDB tree is optimal for geoparquet. I have not had the time to look into this, but it might also be a good choice for OMA files as well.

To summarise: The question of what makes a good set of bounding boxes for OMA files is still open.

See also

• A New File Format for OSM Data
• Using the Oma Library
• Getting Files in OMA File Format
• The OMA File Format
• Dealing with Relations in Oma Files

1. There is also a last chunk without any bounding box. This is used for anything without geometry. In the current implementation, it’s all the collections. ↩

2. Geoparquet is a an approach similar to OMA files, based on parquet files. ↩

Wie lange dauert es bis Anmerkungen und fehlermeldungen bearbeitet werden?