imagico has commented on the following diary entries

Post When Comment
Adding unknown roads using ImproveOSM about 1 month ago

It should be noted that what Bing offer is by far not the best available imagery here - even when ignoring the better black-and-white image offered by Mapbox in this case. Even though the classification of the road is not really identifiable from images, often not even when they are very high resolution, it is generally advisable to have some image as context when you map in an area where you lack recent first hand on-the-ground knowledge.

This is especially important for parts of the world where roads might be seasonal - like winter roads over a frozen lake - not really an issue in southern Sweden but definitely a possibility elsewhere in Scandinavia.

Fixing broken riverbanks about 1 month ago

There is no universal recipe to fix this kind of problem. There are many different kinds of mistakes that occur in broken multipolygons and each of them requires a different approach.

Things that are helpful usually with with riverbank polygons:

  • make the individual polygons small - don't make the mistake of fixing things by merging everything into one big polygon.
  • decide on which area your polygon is meant to cover along the river
  • throw out all members outside this area
  • throw out duplicate members (which are in the relation several times)
  • assemble a closed outer ring, if necessary drawing new ways or splitting ways
  • add all islands within as inner rings
  • make sure all tags are on the relation and none are left on the member ways (unless they apply only to them)
  • create additional polygons using members previously removed as necessary

This is all a matter of exercise and training but even with a lot of experience this is still tedious work.

When you 'discover' a city that is not on OSM about 2 months ago

But that is not a city in terms of the meaning of place=city in OSM.

In general having settlements of this size not mapped beyond a node is nothing rare in OSM. It was not rare in Europe 5-7 years back either.

Entwicklerteam der OpenTopoMap wächst about 2 months ago

Ich meinte das mit der Konvergenz für kurze Distanzen und nicht für lange. Das hängt natürlich dann auch davon ab, was für eine Interpolationsvorschrift man für die Höhenwerte verwendet.

Anstatt den Radius zu variieren könnte man übrigens auch die Höhendaten vorher filtern, hierdurch ließe sich der Einfluss von Rauschen und anderen Artefakten in den Daten vermutlich besser reduzieren als durch einen größeren Radius.

Ein anderer Ansatz wäre, nicht mit einem festen Radius zu arbeiten, sondern für jede Richtung so weit weg zu gehen, wie notwendig ist, um eine bestimmte Mindest-Höhendifferenz zu erreichen.

Entwicklerteam der OpenTopoMap wächst about 2 months ago


Zwei Überlegungen dazu:

  • man könnte versuchen, bei der Berechnung als Nebenprodukt auszugeben, welche Sattelpunkte vermutlich falsch platziert sind, also wo der Punkt im Relief deutlich erkennbar nicht an einem Sattelpunkt liegt.

  • zur Maßstabsabhängigkeit: Wenn man die Richtung eines Sattels als Funktion des Abstands, in welchem die Richtung ermittelt wird, sieht, konvergiert diese Funktion bei eindeutigen Fällen (korrekt platzierter Punkt in Gegend mit starkem Relief und guten Daten) wohl im Allgemeinen für kleine Abstände gegen einen bestimmten Wert. Bei hohen Zoomstufen wäre es vermutlich sinnvoll, diesen Wert zu ermitteln und zu verwenden. Andersherum könnte man auf diese Weise auch erkennen, wo die Richtungsbestimmung sehr unzuverlässig ist (Wert der Funktion schwankt stark oder weist Diskontinuitäten auf).

The 2016 Board Elections Statistics about 2 months ago

Thanks for the writeup and the data.

To me with the two seats that were up for election an interesting observation is the frequency of the different combinations for first and second positions. Even though this is not how STV works (i.e. you do not have two votes for two seats) it still gives an impression what voters would consider the most desirable candidate pairs - or in other words: who would be elected if everyone voted like you.

Obviously Frederik-Kate and Kate-Frederik were the most common combinations. The next most frequent combinations were

  • Frederik-Guillaume (not far behind the two top combinations)
  • Kate-Darafei (much less frequent)
  • Frederik-Darafei
  • Kate-Guillaume

The all new candidate combinations (Guillaume-Darafei and Darafei-Guillaume) were equally popular and together as frequent as the last two in the list above.

Preamble about me 2 months ago

What you describe is probably in large parts an experience many people with a background in classic cartography share.

Regarding the purple boundaries - this is something a lot of people are dissatisfied with in the standard style, but it is not easy to change. But the good thing is of course that not all OSM based maps have purple boundaries so anyone who does not like them can easily find a map without these.

I don't think in map design there is a clear division between conventional cartography and OSM regarding what works well design wise and what does not. The main difference are the circumstances - the lack of a central authority, the world wide coverage, the heterogeneity and generic nature of the data and the need for real time updates in rendering for example. In my experience much of the difficulty of conventional cartographers in 'getting' OSM is related to the difficulty to comprehend these differences and what they mean.

Реки Сахалина 3 months ago

That looks interesting but if i understand it correctly you are inferring the hydrographic network from relief data at least initially without any confirmation that a waterway actually exists there.

May i advise a bit of caution here. If you cross check with imagery afterwards that is fine (and having a pre-generated network to start with is probably very useful). But without any visual confirmation this can lead to pretty large errors like here:

Even in areas where no high resolution imagery exists this can be avoided by using available images.

Global heatmap of HOT contributions, Sept 2016 (with high-res download) 4 months ago

World map projections with a split not at 180 degrees longitude - as BushmanK suggested - are a special difficulty - better to start with the basic version here. That should not be a problem - unless your rendering setup really sucks.

Regarding the map itself - looks like quite a clear case of regional bias. The extreme emphasis on sub-saharan Africa is most obvious - but even apart from that this is clearly not a map showing the most serious objective needs for mapping across the world in the past years.

highway=service – die unverstandene Wegeklasse? 5 months ago


Der Normalfall im ländlichen ist halt, dass ein Zufahrtsweg zu einem Hof, oft asphaltiert und primär nicht landwirtschaftlich genutzt, daneben auch als Zugang zu angrenzenden Feldern und Wiesen dient. Das ist dann m.E. highway=service aber definitiv kein service=driveway.

Zu Flächen - nur weil eine Fläche asphaltiert ist und an eine Straße grenzt wird sie damit nicht zu einem highway=service. highway=* in OSM ist allgemein ein an der tatsächlichen Nutzung orientiertes Tag und keine Beschreibung des Oberflächenzustands.

Das liegt natürlich auch ein bisschen daran, dass es für solche Mehrzweck-/Abstell-/Lagerflächen keine etablierten Tags in OSM gibt - insbesondere für das Mikro-Mapping von Fabrik-/Betriebshöfen und Ähnlichem fehlt da ein strukturiertes Konzept.

highway=service – die unverstandene Wegeklasse? 5 months ago

Gute Punkte.

Bei service=driveway wäre daneben auch ein Ausschluss-Kriterium, wenn der Weg neben der Funktion als Zufahrt zu einem Grundstück auch als Zufahrt zu angrenzenden Nutzflächen wie Acker, Wiesen oder Waldstücke dient - selbst wenn es dafür keine weiteren Abzweigungen gibt.

Ein weiterer häufiger Missbrauch von highway=service liegt daneben bei der Verwendung auf Flächen mit area=yes vor. Dies wird verbreitet als allgemeines Tag für asphaltierte Flächen verwendet, ist dafür aber nicht da, sondern nur für Bereiche, die primär dem bewegten Verkehr dienen und nicht für Flächen, die hauptsächlich zum Parken, Abstellen oder zur Lagerung von Dingen dienen.

The number of the month: 74.9 percent 5 months ago

One thing that might make sense to consider in the long term how it can be better communicated that intensive organized users of the Overpass API need to set up their own instance or buy commercially run services for this - in a similar way as with the tile servers. You indicate that a large part of the load comes from small distributed users but likely lightening the load from volume users would probably improve the overall situations none the less.

One thing i noticed recently that a common use task i have for overpass (usually via overpass turbo) is querying occurrence of fairly rare tag combinations of tags with individually widespread use. This seems to be a fairly hard thing for overpass to do - i frequently get errors (something like out of memory IIRC).

And if there are limits to the growth of OpenStreetMap in sight, then these are most likley the size of mankind.

This is probably right as far as growth is concerned but i think maintaining the data and keeping it up-to-date is a different story because it does not necessarily scale with the number of people involved but more with the intensity of their involvement. See also the musings of Alan McConchie on the matter.

Some comments about the OpenStreetMap Awards 6 months ago

As indicated the language problem is hard, esp. for subjects like writing, community building etc. It might be an option to include local chapters in it - like having a global nomination process but also allow the local chapters to additionally nominate a number of candidates.

If selection is made by a committee it would likewise be an option to put together this committee at least partially from local representatives.

Since the problem of comparing activities in different languages and from different cultural backgrounds is often inherently unsolvable it might also be an option to make the awards non-competitive across languages - meaning each category would have a winner for every language there are nominations in. This of course would only work for categories where a language can be clearly identified for each nominee like writing - mapping for example can be but does not need to be language specific.

Visualizing's Map Views 7 months ago

The 'comet tail' effect - could it be that this has something to do with the order in which map frameworks requests the tiles as you change the view?

Status der OpenTopoMap 7 months ago

Übrigens enden Höhenlinien auf professionellen Karten durchaus auch mal im Nichts, wenn zwischen Detailgraden gewechselt wird.

Um da mal wieder Eduard Imhof zu zitieren:

Trotz dieser Vorzüge sind solch ineinandergeschachtelte Kurvensysteme nicht zu empfehlen. Sie sind zu schwer lesbar, zu wenig anschaulich. Alle Geländeteile, ob steil oder flach, erscheinen in solchen Karten fast gleichmäßig mit Höhenkurven gefüllt.

Was eher geht, ist in flachen Gebieten einzelne Zwischenkurven einzufügen, die man in steileren Bereichen weglässt. Auch das geht aber nur, wenn diese in der Darstellung deutlich abgegrenzt und dadurch auf den ersten Blick identifizierbar sind, zum Beispiel durch Strichelung - Verwendet ist dieser Ansatz zum Beispiel hier.

Eine gute Auswahl der Gebiete, wo man Zwichenkurven darstellt und wo nicht ist aber ziemlich schwierig (zumindest wenn das irgendwie auch performant sein soll). Und das ganze kann auch sehr schnell unübersichtlich werden.

Wenn ich mir die Höhenlinien bei euch anschaue sehe ich Einsparpotential eigentlich eher im Flachland, wo vermutlich mindestens 2/3 der Datenmenge Rauschen ist wie hier. Aber natürlich ist die Hauptmenge der Daten im Gebirge und letztendlich sind die 10m-Intervalle dort bei SRTM-Datenbasis weitgehend redundant (d.h. könnten im Prinzip aus den 20m-Linien fast perfekt interpoliert werden).

Im Grunde ist ja auch das Speichern der Höhenlinien in einer PostGIS-Datenbank ziemlich sinnlos, denn sowohl produziert als auch verwendet werden sie ja immer in Kachelform.

Deriving centerlines from riverbanks without. 8 months ago

Not sure if you do this as a helper tool for mapping or for data use.

For mapping this works nicely - have done this as well a few times for long and winded riverbank polygons because i was too lazy to draw the centerline by hand. Would be useful as a JOSM plugin by the way.

For data use as a preprocessing step for missing data? Forget it if you want to process things globally - unless you have a real lot of computing power to spare. Skeleton algorithms generally scale poorly with polygon complexity, generating skeletons for big polygons is slow.

The map is a fractal 8 months ago

Be careful here not to confuse progressive data representation and mapping strategies with fractals.

Fractals are characterized by self similarity and fractal shapes in physical reality are commonly the results of a self-organization process. This is widespread for natural shapes but is very rare for human generated structures. What you describe above is a progressive approach to mapping things, starting with a crude representation, both semantically and geometrically and refining it step by step. There is nothing fractal about this, there is commonly no similarity between either tags or geometric forms between different levels of refinement.

And it is not the fact that the coastline is longer when measured on a finer scale that makes it fractal-like, it is the fact that the relationship between scale and measured length has a simple form (power law) across a wide range of scales that causes this - see here.

Up-to-date open data imagery - it is available, use it! 8 months ago

I think i made it pretty clear that the primary purpose of this is to increase awareness that such imagery exists and can be used in the hope that this prevents mappers from wasting their time with 15 year old images in Bing and Mapbox. This happens to apply mostly to remote areas. The availability of up-to-date open data satellite imagery however is not limited to those areas.

Another advantage is to have up-to-date imagery for every part of the world, up-to-date here usually meaning a few weeks old, in difficult cases sometimes a few months. Age of high resolution imagery in Bing, Mapbox or from local government sources is usually at least about 2-3 years. I think i demonstrated this quite clearly with the Panama and Vostochny images.

And yes, most use of maps and most mapping takes place in crowded areas but where maps are most needed are usually the less crowded parts of the world. Or in other words - in a European or North American city a map is a luxury, in the remote parts of the world it is frequently a necessity.

Up-to-date open data imagery - it is available, use it! 8 months ago

@BushmanK - yes, there is a lot written about processing satellite data. Science papers are usually only of limited use though since they tend to not mention the practical problems and limitations but concentrate on presenting something as 'nice in theory'. There is a lot of practical material available as well of course, starting with the GDAL documentation and the satellite data specifications:

Most practical tutorials present very specific workflows based on specific tools but can also serve as sources for general information. Like:

@Skippern - my images are in the OSM Editor Layer Index but i do not manage to keep them completely up-to-date there so you either have to enter the layers manually or wait until an update is made.

For JOSM i recommend using

in addition to the Editor Layer Index - this way you have the individual layers and an up-to-date version of the aggregate of all layers.

Experimental publishing of Sentinel 2 satellite data 8 months ago

There is nothing wrong with looking for and asking for instructions but of course for a faceted topic like satellite image processing there is not one single way to do things and what works well for someone processing satellite data professionally rarely works equally well for beginners. So when i write specific instructions (like here) i tend to do this based on a workflow specifically designed for the task in question which is usually not what i use myself routinely. You can try to follow such instructions step by step to do very specific things or you can try to learn more generic matters from them but you cannot expect develop a fundamental understanding that enables you to do new things and make qualified decisions just by reading and following a specific tutorial.

And the accusation of "rude smart ass professionals" is not that far fetched with regards to Sentinel-2 - there is a strong tendency within ESA and related companies towards elitist, non-transparent work and a distinct culture of non-openness, the SNAP/Sentinel toolboxes system is just one example for this. The open data nature of the Sentinel program is very new to most of the people working in that environment.