OpenStreetMap
Some tricks to map Streams in Mountain Area.
Mapping streams/rivers in mountain region could be very tricky. Sometimes it’s difficult to differentiate between the valley and the peaks. Other times the imagery is covered with cloud, snow or not visible due to the shadow.
Satellite Image showing parts of Gangotri National Park, India.
A few steps can be very helpful to deal with such problems.
Overview of the area
Before jumping into the task of mapping, take some time to have an overview of the whole area. This helps to have a better understanding of the region.
Switching between layers
Better to select the layer that best shows your interest features clearly. A better resolution imagery is not always the best one.
Using terrain layer
Basic knowledge contours could be very helpful while tracing rivers and streams, specially when hill shadow makes it difficult to see the riverbed. It helps to differentiate between peaks and valley. Also, you can easily find the way of the stream. Try using a combination of both terrain data and satellite imagery, works best.
Bottom to top approach
Water always flows from higher ground to lower(into the sea). Better to mark the big streams first, one which are major rivers and easily visible. Now try to upstream connecting small streams. Won’t be a good thing to leave a stream going nowhere
What not to mark?
Mountain regions are filled with landslides and very small seasonal streams. It’s very difficult to defferntiate between them. What is the correct way to map them? natural=scree,hazard_type=landslide or waterway=stream & intermittent=yes?
What do you think?
Happy Mapping.
토론
2015년 8월 17일 13:48에 SK53님의 의견
One trick I find is that if using JOSM the basic structure of the terrain tends to standout in Landsat imagery. Sketching the basic pattern of the streams and rivers at low zooms can really help to maintain orientation once refining the detail at higher zooms. I often trace ridge lines for the same reason. Using Overpass-turbo can help keep an overall prespective on ones work.
2015년 8월 17일 13:48에 SK53님의 의견
One trick I find is that if using JOSM the basic structure of the terrain tends to standout in Landsat imagery. Sketching the basic pattern of the streams and rivers at low zooms can really help to maintain orientation once refining the detail at higher zooms. I often trace ridge lines for the same reason. Using Overpass-turbo can help keep an overall prespective on ones work.
2015년 8월 17일 16:27에 yvecai님의 의견
Also, on JOSM, you can overlay a relief layer over imagery. Tiles from opensnowmap.org are transparent, show contours lines and altitude that help to distinguish peak from valley.
2015년 8월 17일 21:09에 Alan Trick님의 의견
It’s hard to tell unless you’ve been there, and even there’s no clear rules. If creek-bed is dry during the summer, I will usually go with natural=scree for the creek-bed and waterway=stream & intermittent=yes for the line where the water would go.
I usually don’t map streams unless I’ve:
There’s a fair bit that is difficult to tell from satellite imagery. How intermittent it is is the main one, as you mentioned. It’s also possible that the stream might go under a boulder field, and be essentially “underground”.
In the Himalayas, you often have lots of loose soil, which results in very distinct angles in the landscape, which makes it easy to see where the water would go. In many places, the valley bottoms are a lot harder to see.
2015년 8월 18일 09:14에 pratikyadav님의 의견
@all Thank you for your suggestions.
2015년 8월 19일 13:03에 PlaneMad님의 의견
A nice relief shaded map used as an overlay can instantly help differentiate valleys from streams. The cyclemap layer is pretty ok for this for this purpose, but a specialized layer could really help.
2015년 8월 21일 05:05에 BushmanK님의 의견
If there is any digital elevation model or contours available for this area, it’s always possible to use special tools such as SAGA GIS to automatically analyze terrain and create hypothetical watershed lines and so on. Technique described above is only for cases when you have nothing like that.