The Sentinel-2 satellite finally captured a cloud free coverage of the 2018 Laos Dam Collapse site. Over a distance of more than 50km, everything near the river has been washed away, trees being stripped bare from the river banks.

Comparing it to the river pre-disaster shows how much energy was in the water: https://gfycat.com/LeadingAbsoluteConure

The map below shows what happened, one of the secondary dams overflowed, causing the reservoir to drain into another river.

source: http://www.gdacs.org/Public/download.aspx?type=DC&id=28

On 24 June 2017, a big landslide occured in China’s Sichuan province, burying dozens of villagers alive. Although pictures from the ground clearly showed that most victims didn’t have a chance for survival, satellite imagery shows the scale in one glance: (source: ESA Copernicus Sentinel 2. 19 February 2017 and 7 September 2017.)

Some OSM users started mapping buildings in the area immediately afterwards. Pretty much the entire built-up area of one village was buried.

Thanks to the freely available Sentinel-2 data, you can map objects as recent as a month old. It’s a shame not much of the data is available in a readily usable image format, for now, processing it can be a time consuming job.

I’m tempted to replace all those named as clinic by the tag amenity=clinic, but maybe someone who has lived in Korea can confirm if that’s appropriate?

Use this to find them: http://overpass-turbo.eu/s/rRX

To name a few things in the above area:

• residential roads tagged as motorways, primary or secondary

• the same road went from motorway to service to residential to primary

• island roads (not connected to rest of network)

• roads ending just before the crossing

• roads split in arbitrary segments all over the place

• ways with a node for every half a metre length

And how about this?

It’s not exactly a secret that the official Chinese government statistics are often made up or obfuscated. The official stats for road network length say 4,696,300 km of roads. According to the CIA world factbook, China has 4,046,300 km of paved roads. In 2016, on OSM 1,635,931 km of roads was mapped in China. Update: in June 2017 we’re already at 2,022,047 km!

Now compare the OSM data of a typical countryside area to the satellite imagery.

Pretty much all unclassified roads are still missing from OSM, and this is the case for most of rural China. Even for cities, there are only a handful of cities that have their residential roads mapped well. So it leaves me wondering, is the 4 million km of roads in total a realistic statistic?

Looking at the satellite imagery again, there are (minor) roads in a grid of about 300 meters distance. The densely populated North China Plain alone has an area of 409,500 km². Assuming every squared km of this area is covered by 6 km of roads, then there is a total of almost 2,500,000 km of roads in the Northeast China Plain alone. Within villages, the road grid is much more dense. And of course there are plenty of other densely populated areas in China, such as the Sichuan Basin, the Loess Plateau, the Northeast China Plain, the Pearl River Delta etc. etc.

Doing some very rough calculations (area of above regions times 2 km of road per square km):

Is it doubtful that the rest of the country has only 1.3 million km of roads? And this is with a conservative estimate of 2 km of roads per square km (although the Loess Plateau is not that densely populated).

There is also the number of road length per capita. It seems Germany has a comparable population density to China (the population of Western China is neglible). Using the number from Germany (2.81) times the population of China gives around 4,000,000 km of roads. Comparable to the official stats.

The only way to get the answer is by waiting some years for the OSM data to grow further. It’s actually likely that even the Chinese government doesn’t know the real number, as maps used in China (Baidu, Tencent) lack most of the minor roads. It could be that they censor it, but the purpose seems unclear in that case.

USGS has published tree cover data based on 2010 Landsat captures. I wonder if this data would be suitable for importing. Especially outside of Europe, forest cover is largely incomplete now, and 2010 is fairly recent for this kind of data.

https://landcover.usgs.gov/glc/TreeCoverDescriptionAndDownloads.php

In the licence requirement it says: University of Maryland, Department of Geographical Sciences and USGS; use is free to all if acknowledgement is made. So it is not obvious if using this data is allowed.

The data is provided asgreyscale geoTIFFs, i have uploaded one tile as a preview here:

https://api.mapbox.com/styles/v1/hindbaer/cj19m696v009v2ro4rwlq18k3.html?title=true&access_token=pk.eyJ1IjoiaGluZGJhZXIiLCJhIjoiY2lqOGt4bG1wMDAweXR0a25vcGx4ZmUzNCJ9.R55CveYJBChT-Olntza9Mg#7.21/23.611/121.796

Is the licence ok to use for adding data to OSM? If not, we could seek explicit permission from USGS, this data was directly derived from Landsat anyway, so they may be able to allow less strict attribution requirements.

Should the licence be suitable, then of course there will be many issues. For starters, the data is about ‘tree cover’ which could be ‘landuse=forest’, ‘natural=wood’, ‘landuse=orchard’, etc. Then there is the difference between OSM, where an area can only be forest or not forest, and this 1-100 scaled data. At which threshold is it a wooded area?

Worst of all, this data was uploaded by someone using JOSM…

If you map in the West you may come to think that the major stuff in OSM is done, but in developing countries, you may discover major unmapped features.

天地图+OSM data Sentinel-2 (8 December 2016)

I’ve been mapping lots of countryside in Gansu, China, and the number of temples is quite high.

In the predominantly muslim areas, mosques are usually very easy to recognize (although I hope I didn’t miss too many because of tunnel-vision), due to the usually square main building with a central dome, surrounded my 2 or 4 minarets.

Buddhist and Taoist temples however, are impossible to distinguish for me, as someone with only basic knowledge of Chinese architecture and these two religions. On the ground I would probably be able to, but with the number of local mappers in rural Gansu and my travel budget not being impressively high, all there is left is studying Bing imagery.

Some examples:

1. https://binged.it/2iiI15h
2. https://binged.it/2hKq2YL
3. https://binged.it/2hKjv01
4. https://binged.it/2hKCcAS
5. https://binged.it/2hKuLd0
6. https://binged.it/2hKr86Q
7. https://binged.it/2hKs3nO
8. https://binged.it/2hKllxX
9. https://binged.it/2hKwn6y
10. https://binged.it/2iiP2mS

Without looking, can you tell which one is Buddhist? Of those listed, one was documented, so I could confirm (or rather, correct) that is is Buddhist: http://www.openstreetmap.org/way/454051191

Besides purely looking at the architecture, the location also gives things away of course. In a Hui autonomous county, you expect to find more mosques, in a Tibetan Autonomous Prefecture, you expect more Buddhist temples. Some temples are located on hilltops, others within a residential area and others are surrounded by farmland. I don’t know what significance this has, but surely, there must be some.

If you know the name of the temple, this can apparently also give you a hint. https://en.wikipedia.org/wiki/Chinese_temple

Chinese religion experts, please let yourself be heard in the comments or via a message.

Population numbers turn out to be pretty important for rendering places at the correct zoom level. Since ‘town’ or ‘city’ is the highest possible qualification for any large urban centre (whether a sleepy 100 000 population suburb or a 10 million population metropolis) there may be many of these places very near each other. On lower zoom level, the renderer has to make a choice on which cities to show. Without any population numbers, no decision can be made, thus no names are rendered.

The default Carto layer does take population into account, and it can be seen that it is a major factor in deciding which place is shown at lower zoom levels.

In green are cities I’ve recently added population number to (number is brackets is the urban population). These are all rendered as long as the labels fit nicely without overlapping. It is not strange that the desert town of Hami is labeled at this zoom level. It should be, because it is a major stop along this transport corridor.

However, Hangzhou and Xuzhou are not labeled in this case. For Hangzhou it can be argued that the label would overlap with Shanghai, but for Xuzhou this argument cannot hold.

Between 兰州 (Lanzhou) and 天水 (Tianshui) lies 定西 (Dingxi). As you can see here I’ve also added this city’s population number. And although the label would fit, it is not rendered. I have not looked into the code of the Carto renderer, but my guess would be that either it’s population of under 500 000 would not make it important enough, or that the other cities are too close by (Hami is a lot more isolated). For completeness, Lanzhou and Tianshui have a population of a few million.

Also good to note, if you zoom in to level 6, you get a little dizzy from all the city names. Apparently here the renderer shows anything that is a town, and because almost all lack population numbers, all are shown.

It seems that in India for example, most cities do have a population number included, so if you zoom through the levels there, the map looks much more usable.

在中国駕駛人校让学生来练习在闭路，平素也有练道，可能是相当长。

我的问题是，这就是’highway=raceway’？

我也用’amenity=driving school’，可是，最多呈现没表示。此外，可能办公室是在别的地方。例如如果学校有体育场在别的地方，这个体育场不是’amenity=school’。所以，可能我应该只用’highway=raceway’ tag.

In China, driving schools operate closed circuits where students practice. These may include tracks that can stretch for a kilometer or so. Currently I usually tag the grounds as ‘amenity=driving_school’ and the track with ‘highway=raceway’. However, this may not be correct, as the driving school may have its main office elsewhere. You wouldn’t tag a sports pitch as ‘amenity=school’ just because a school uses it. So maybe I only should use the ‘highway=raceway’ tag.

例子：

• http://www.openstreetmap.org/way/438782365
• http://www.openstreetmap.org/#map=17/36.08822/103.87192&layers=N

自己研究：http://overpass-turbo.eu/s/jbQ

Just started mapping this area. Currently looks uninhabited on the map, but check back later in a few weeks or so, or join me in tracing the area. Mapping areas like this is so much different from mapping in Western Europe where you sometimes wonder if there’s anything to add. It gives a larger sense of accomplishment to ‘discover’ a town or an entire valley compared to drawing in a missing footway.

Edit: Finished mapping most tertiary and higher class roads, and added landuse for the major villages and towns. Now hopefully someone with access to name data can continue improving this area too.

Mapping southern Gansu Province, China, I’ve encountered dozens of closed walls, in shapes usually square or oval. In some areas, every village has it’s own ‘castle’. So far, the only thing I’ve read about them was this:

“After the fall of the Qing-dynasty, when the country was undergoing an epoch of chaos, the farmers in the area were being constantly terrorized by local warlords and bandit hordes. So they hurriedly erected these fortifications on the highest points in the hills, thus having a safe haven to fall back to in case of an attack.” Source (with a few pictures)

Here is another picture up close. It even seems to be made from rammed earth, which could mean that some of them may be older than they may seem at first sight.

Edit: A kind OSM contributor sent me a link to a documentary about these structures. It’s in Chinese, but with English subtitles.

http://tv.cctv.com/2016/03/22/VIDE2oprFGUUcK44tKYNHfGX160322.shtml

I haven’t had the time to watch the entire documentary, but I got as far as that parts of these structures may be many centuries old.

I wonder if some of them may have already been constructed during the muslim rebellions a few decades earlier.

So far I’ve tagged hem as ‘historic=fort’, ‘barrier=wall’. Use this Overpass query to find them. A few that had noticeable decay have been tagged as ‘historic=ruins’.

If you have more information on these structures I’d be happy to hear so (may also be in Chinese).