Google Maps Elevation Web Service | Hacker News Reader

Google Maps Elevation Web Service

code.google.com

23 pointsdrp11 comments

Cool service, albeit it's for smaller uses. The limit is 25,000 requests a day. This sounds like alot, until you realize each point is a request. So, you want elevations for a 100 mile route drawn on a map? That's about 2-3000 points, so you only get 10-20 of such routes before hitting your daily limit.

Other issue is the accuracy: I am not sure the base DEM resolution they are using, but my guess is it's the standard 10meter spacing between points inside US and territories, and 30meters outside. 10meter grid spacing isn't bad at all, it's actually pretty good, but their algorithm for interpolating is non-existent. For any point not directly on a grid point, they just average the four surrounding elevations and return that. I have played with several interpolation schemes, and I have found either a bilinear or a cubic interpolation for interior points to be vastly superior to a nearest neighbor or averaging scheme. Differences can be easily up to 10 meters for a 10meter grid spacing.

I had been hoping they'd release this, but I'll stick to my home-brewed service with the 500gig NED at 10 meter resolution (1/3 arcsecond). No API limits :)

Might also be useful for augmented reality apps, to the elevation of virtual objects right.

[Edit: as pointed out below it seems like it really is 25,000 unique locations, my mistake.]

so you only get 10-20 of such routes before hitting your daily limit.

Not true, from the usage docs: Up to 1,024 locations may be queried within a single request.

So that's really a theoretical max of 25,600,000 locations per day if you maxed out every request.

Not true, from the usage docs: Up to 1,024 locations may be queried within a single request.

"Note that on any given query, you may request elevation data for multiple locations, which count as separate queries."

Approximate! No need to do a lot of requests for flat land!

Approximations only get you so far. Try telling picky cyclists, who compare your calculated elevation gain/loss to ten other sources out there, that you just want to approximate :)

The NASA Shuttle Radar Topographic Mission (SRTM) has provided digital elevation data (DEMs) for over 80% of the globe. This data is currently distributed free of charge by USGS and is available for download from the National Map Seamless Data Distribution System, or the USGS ftp site. The SRTM data is available as 3 arc second (approx. 90m resolution) DEMs.

http://srtm.csi.cgiar.org/SELECTION/inputCoord.asp

I guess it's the same enhanced SRTM data that Google uses?

Wonder if this would be accurate enough for a GPS -> bike route translator. Like say you want to see the altitude throughout your bike right from the gps you recorded.

The RideWithGPS people appear to do this: http://ridewithgps.com/trips/22776

Though if you're somewhere very hilly (ahem Pittsburgh) it appears that slight issues with the GPS track will throw in a lot of extraneous elevation data.

We are working on some cool ways to fix GPS tracks that have drop-outs in signal/accuracy. I am going to use some signal processing tricks to smoothout the points along the route. I am using a Gaussian filter on the elevation data, and I think this will also work good on the actual lat/lng points, which should also help with total gain/loss accuracy.

Would be decently accurate, though their approximations would lead to a misrepresentation of gain/loss. However, their daily limits would really inhibit use of this service for said purpose, since my Garmin bike computer records easily 2000 points in a 30 mile ride, even on smart recording (every few seconds)

Reply on news.ycombinator.com