I’m mapping geographic features in a remote region of Alaska and am looking for the best way to ensure that the satellite imagery is correctly aligned, that is, has the proper offset. There are no GPS traces or man-made features to help decide this. I’m using satellite imagery from several providers, Bing, Mapbox, and ESRI Clarity, depending on the quality, cloud cover, snow cover, of the particular layer in the area I’m mapping. Unfortunately, they do not align with one another. In fact, as I move the viewport along the Noatak River from east to west the relative offset differences among those layers seem to change, perhaps because the area is so close to the polar regions (latitude 68 N).
My question is, which one is closest to reality?
I’m using the USGS Topo maps (layer: USA/Mexico/Canada/Scandinavia Topo Maps) for names of lakes, rivers, and mountains, and I’m wondering if that layer might be geolocated with enough accuracy within JOSM to use to align the other layers? Many geographical features have changed over the years since those maps were compiled but, IIRC, they were very accurately made.
Does anybody have any information about the positional accuracy of the USGS Topo maps? Or any recommendations about how to solve my problem?
I don’t think the map you suggest using is really aligned that well.
I’ve been mapping using the National Geodetic Survey markers. I tried GPS traces, and I was taking at least 30 minutes to get a good offset from them. If I can get a survey marker that shows up well on at least one imagery source, I can adjust the offset using that and a ruler on my monitor in only about 10 minutes or less. Then find some manhole covers or something very nearby that show up well on all imagery sources and use them to align the other imagery sources to the first one.
I’ve found that in my local area, they’re off from GPS traces by about 6 feet, but it’s easier and much more repeatable to use them than the GPS traces, so I just map everything to that offset even if it is off by about 6 feet.
If you want to look for the NGS markers, go to the site NGS Survey Map | National Geodetic Survey and click the link for the web map. You want to look for markers that have position order 1, they’re the most accurate. They show up as a white triangle on the map. Most of the survey markers were placed in the 1930s, so some don’t exist anymore, but I’ve found that my local city still maintains some of them in the city. One problem is they can be difficult to find and distinguish on the imagery. Some of the primary ones I use are only 5" in diameter, so unless your imagery is very good, you might not find them. If you don’t find anything with one, try a few others and maybe you can find one that shows up. Or visit the site to see what it looks like in person before trying to find it on the imagery.
USGS 3DEP contours are available in JOSM. The horizontal accuracy is supposed to be very good, although there are a few obvious glitches. Since the USGS topo’s also have contours, you could align them to the 3DEP data.
This is normal anywhere in the world. The position where a pixel is displayed depends on the collection angles and the terrain. Theoretically that all can be corrected for with orthorectification, but in practice the correction isn’t perfect. Offsets from one area cannot be applied to another, even if the two areas are only a few hundred meters apart - especially in areas of high relief. Also, when providers update their imagery, the offsets will most likely change as the collection anges most likely will have changed.
As I mentioned in my post, there are no GPS traces in the roadless, uninhabited areas I’m working with. I will investigate the NGS survey marker database for help in alignments. I have used it before but it was a long time ago. The imagery in remote areas is not good enough to actually see any NGS markers, but IIRC they do appear on the USGS Topos and if I correctly position those, I could use them to align the satellite imagery layers.
I checked the USGS 3DEP layer and it might help however, the shapes of the contours on the two Topo maps are far from identical so I think that technique might be of only limited usefulness. But I’ll work with it a bit more to see what I can come up with.
Can you provide an example location where this is the case? My experience has been that you have to shift the topo by hundreds of meters to get them to line up.
That is not the case in my area of interest. The contours line up reasonably well but there is “more detail” in the DEM than on the Topos so you can’t be positive about the alignment but it’s definitely not a hundred meters off. I’m using the NGS markers to align my imagery at the moment. It might not be precise, as ArgyleZombie indicated, but it’s better than what I had before.
Along the Noatak River (Alaska) there are only a few NGS Survey Markers near the lower reaches of the river and none in the upper portions, Luckily, absolute precision isn’t required for people rafting the river. I might be rafting it this summer and that’s why I decided to enhance the river mapping in the area.
At least now when I switch from Bing to ESRI to Mapbox to USGS Topos and back again, I can be reasonably sure my objects are not massively mispositioned.
Check this screenshot - the fine black lines are the DEM contours, the brown are the USGS Topo contours. I’ve adjusted the transparency so both images are visible. They sort of match but not quite.
Individual contours are not going to exactly align. 3DEP is a lot more detailed, and the contour interval varies with zoom level, so it is most certainly different from the topos. You have to look for individual features in both sources that you can match, such as hills/peaks, ridges, valleys and bodies of water (contour lines generally shouldn’t cross bodies of non flowing water).
The example you provide looks very well aligned. Note the hill in the NE quadrant of your screen shot. The 3DEP contours clearly show a hill, and the topo has a spot elevation point there (998’).
In your example it is not, but if you get into areas of higher relief, it may be.
@tekim
Thanks for all the help. I will continue to experiment with the 3DEP data and the NGS markers. I did not know about the 3DEP layer before. Aside from the alignment issue, I’m using it constantly as a semi-transparent overlay on top of whatever sat imagery I’m using and it really helps illustrate the terrain. I find it especially useful to trace the shorelines of lakes when the imagery is shadowed or obscured by cloud cover.
@AlaskaDave You are welcome! There is also a 3DEP hill shade layer available in JOSM which you may have discovered. This is helpful for locating tracks, and sometimes even trails, that are under tree canopy. It is also a help in aligning streams.
It is my understanding that 3DEP is not artificially “flattened”, so as you have discovered, it is possible to pick out the actual outlines of inland water bodies with a great deal of precision using it (particularly the hill shade). The 3DEP data may not be flat over the ocean however, as it may show the surf.
USGS states that the horizontal accuracy of 3DEP is “better than one meter”, but I am not sure if that applies to Alaska, where it is IFSAR derived rather tha lidar derived. Nevertheless, it should be one of the best sources for horizontal position.
