Slashdot Mirror
Ideas For Exploiting NASA's SRTM Data
MaxTardiveau writes with an excerpt from an article where the pictures are worth clicking through for: "Ten years ago, in February 2000, NASA mapped the entire world in eleven days. It's true: the mission was called the Shuttle Radar Topography Mission (SRTM), and over the course of eleven days, it used a big radar attached to the space shuttle to get elevation data from the vast majority of solid Earth; practically all land between 60 degrees North and 56 degrees South was included, with a resolution of 30 meters (90 feet). Over 9 terabytes of data were captured. It then took two years to process that data and make it usable (and it is still being refined to this day). This data is freely available to anyone, and the number of possible applications is almost infinite. It's been used in GIS, cartography, environmental planning, weather modeling (weather patterns are enormously influenced by the topography), flight simulators, Google Earth, and the list goes on. In this short article, I would like to give you a quick tour of the kinds of things this data can reveal. My hope is to get you thinking about what else could be done with this incredible resource."
It's just perfect use for games, from flight simulators to city building and civilization series. It's a lot more fun to play on real terrain.
I live at 60 deg North, you insensitive clod! (Ok, right now it is 59 deg 51' 7"...) I don't want to learn that I'll fall off a cliff if I take a step in the wrong direction...
It's every programmer's wank-fodder... a solution in search of a problem!
How is this different than the other elevation datasets which are already available? It doesn't seem to be higher resolution, nor more complete coverage.
The raw data, as well as data from multiple other sources can be downloaded from NASA's Earth Explorer http://edcsns17.cr.usgs.gov/EarthExplorer/. The article doesn't really address the fact that the Google data has been cleaned up a lot. SRTM has a lot of voids and areas of poor quality, especially over mountains. The resolution of the data is worse for anywhere outside of the USA.
According to Wikipedia, the 30-meter data is only available for the United States. The rest of the world will have to do with 90-meter data.
1. mapped the entire world in eleven days.
2. vast majority of solid Earth
3. practically all land between 60 degrees North and 56 degrees South was included, with a resolution of 30 meters (90 feet).
Mmm, within ONE paragraph the writer sure lowers his tune. ALL the earth vs a part of it and then that part of it that is land and not sea? Anyone remember why it is called the Blue Planet at times? Because the majority of the surface is water?
Interesting news of course, but come on, can we at least on a tech site skip the hyperbole.
MMO Quests are like orgasms:
You may solo them, I prefer them in a group.
Finally, we might actually be able to find Osama Bin Laden.
I want to know if Plato was just BSing us. Perhaps other lost civilizations could be found?
Perhaps 30 meters isn't enough?
All rites reversed 2010
I had the same thought. For most locations, there is data available which is even more precise than the SRTM data.
But the difference is that the SRTM data is freely available and has the same format all over the world, whereas the other data usually isn't cheap and has a different format in each country. Older data may even be non-structured (i.e. in a simple image format). It would be a PITA to get all the information together in a usable form.
I know Americans like to equate meters with yards, and when dealing with a small number, this is a close enough approximation for most purposes. However, 30 meters is 98.4 feet, so a better approximation for the purposes of this post would have been 100 feet.
Didn't we learn our lesson regarding sloppy unit conversions during one of our recent multi-million dollar collisions with Mars? ;)
Proud member of the Weirdo-American community.
The Canadian RADARSAT-I and RADARSAT-2 satellites have better data. Resolution goes down to 3 meters if desired, and is 25 meters normally. That's much better than what NASA has. Here's Ottawa seen by RADARSAT-II. Here's Paris. They did it first, too; here's RADARSAT-I's first image from 1995. RADARSAT-I was launched from the US on a Delta booster back in 1995, but RADARSAT-II was launched from Kazakhstan on a Soyuz booster
They collect amplitude, phase, and range data, so they can do processing to get false-color images which bring out terrain features. Here's Washington after processing.
RADARSAT is a commercial service. You can order images. The base price for a custom image (taken at your request, not from the archive) is $5400CN. Wait time is a week or two. If you're in a real hurry, an additional $4,800CN rush charge gets your picture taken within about 12 hours. Archival data is much cheaper, and is available from MDA Corporation. MDA also has data from Ikonos, Quickbird, Landsat, etc. Much topo data comes from those archives already.
Unlike the NASA data, this data is good enough to easily tell land from water. Better radar systems return "first and last" returns, which, over wooded areas, return both ground height and tree height, so areas of vegetation can be detected. The Washington DC false-color image shows all this.
It doesn't take all the NASA overhead of putting people in space to do this. The private sector is doing it just fine.
I've been trying to play with GIS data for a few months now, and always seem to get hung up on what to do with said data once I've downloaded it. I've attempted to find some open-source visualization software, and had some limited success with SAGA GIS ... but the interface is pretty clunky, and the documentation is either outdated or for previous versions.
Anyone have any suggestions for visualization of terrain data ? Frankly, whatever was used in the article would be useful, but I didn't see the author specify what software he was using.
The 30m SRTM data is trash, the mission totally fucked up collecting the data. We've been trying to use it for years. It isn't still being 'refined', it's being gap filled with various interpolation algorithms and other data sets. The best NASA DEM is the one from the ASTER instruments. Not only is it also 30m horizontal resolution, but it goes up to 82 degrees north and south.
I got access to the data through my NASA contacts early on and download quite a bit for some western states. We used/and still use the GPS position and altitude of our aircraft to extract elevation from SRTM and then compute the height over ground for our remote sensing data to aid in geo-rectification of our images of wildfires.
Nate
OpenStreetMap uses this data to give the elevation contour lines on the cycle map rendering. Eventually, it'll be used to guide cyclists on a flatter, faster (but possibly slightly longer) route to avoid the steep stuff.
Furries make the internet go.
Hopefully this will spur more interest in some of the many tools that NASA provides for free on its website. There are many free Java applications (standalone or jnlp) to view the data or embed it within your own application. Though the documentation is not always the greatest, with a little tinkering you can make interactive websites for anything from planning your camping trip to searching for ancient meteor craters.
http://worldwind.arc.nasa.gov/java/
Though a lot of the sources are availble, many of the Linux distributions don't have an easy way of building them. It's a real pain to build, but the results are spectacular.
We have no idea where many photographs were taken. Also, there are many old paintings with poor titles such as "Landscape at Dawn/Sunset".
By scanning images with definable features (e.g. two hills, river in the foreground, somewhere in France), it shouldn't be that hard to exhaustively search the SRTM data for possible location matches- especially if individuals could enrich the scanned data by providing height or distance estimates?
I used the SRTM data extensively last year developing a program to identify set gradients for intelligent transport.
One liter of fuel passed through a small engine will provide enough energy to raise you to the height of Mount Everest.
From the top of Mt Everest a consistent 0.7% gradient will travel for over 1,000 km
Current technology, like an aero-dynamic coffin on rails, will roll down a 0.7% gradient at over 65 km/hr
That's 1,000 km/litre, or over 2,500 mpg, and the efficiency we need to be aiming for with our transport system
check it at http//www.megametrelitre.com
The memories of a man in his old age are the deeds of a man in his prime - Floyd, Pink
That's not what I thought of when I read downloading the data. I thought I'd be pointed to an FTP server of huge files or something similar. Instead I've found my way to a map where I can use various graphical means to select which subset of the data I want. There are many limitations with this interface.
Why isn't this data posted to a large archive site where people can freely download more than that GUI allows (or get data in a differently controlled way that facilitates big downloads). Perhaps the Internet Archive can host all of the data and allow such downloading.
Digital Citizen
Throughout most of art history, exacting realism in paintings is pretty much a small blip - an artist will alter the shape of anything, including landscape, to be more aesthetically pleasing. I am really dubious you could get anything out of trying matches with exacting terrain data...
It would work better with photographs but then you also have to figure out a feature distance map from a flat image.
"There is more worth loving than we have strength to love." - Brian Jay Stanley
Big thanks to NASA for this data, I've used it with the Splat! tool for radio path profiling to generate maps and estimate height above average terrain (HAAT). I've been using these maps to show what the approximate footprint is for amateur radio repeaters. See this example for a 90W 440Mhz repeater on top of the Empire State Building.
-molo
Using your sig line to advertise for friends is lame.
"(Ok, right now it is 59 deg 51' 7"...) "
Are you slowly moving south on the ice sheet/glacier?
I only live at 46.3 degrees north, and it just now got up to 0F for the first time in 3 days.)
First, a quick reaction to your post. Radarsat-1 and 2 data, in regards to DEMs, is far from being comparable. The SRTM mission was dual-antenna interferometry, with Radarsat (or Envisat), you must use two images at different times. DEMs from Radarsat can be good and better than SRTM, but it's pretty expensive and there are alternatives (in Canada: CDED1 data is free and in many cases much more reliable than Radarsat data).
If you ever want to use SRTM-DEM data, check the CSI-CGIAR version 4 version. It's the best out there, it's a *major* improvement over the original and previous versions. If you're in hydrography, look at HydroSHEDS SRTM-DEM data.
This year, the advent of the ASTER-GDEM (global DEM) diminished the interest of SRTM-DEM. ASTER-GDEM is still "research-grade", but offers higher spatial resolution, and most important, cover much more land than SRTM (northerm Canada and URSS).
On top of my mind, don't forget SRTM-DEM is available at a higher spatial resolution over the USA than elsewhere (1 arc-second vs 3). The 1-arc-second for the whole world is suppose to become available some time in the future, but that has not happened yet. Also, the X-band (the actual SRTM-DEM comes from the C-band data IIRC) is in the hands of the Germans, but to my knowledge, no public DEMs has come out of it yet (even after all those years). Still relying on my memory (I can be mistaken, see next paragraph), the TerraSAR-X in orbit should be able to give us an even better near global DEM than what's available at the moment.
Sorry for the lack of links. I'm still in my holiday break and you can simply google your way in. Or search SRTM on the site in my sig! have fun -
Animoog.org
Quantum GIS might be intereting to you: qgis.org and so may gvSIG. If you need raw power, look at GRASS GIS (the interface isn't up to par if you ask me and you can access many GRASS features directly from QGIS). (if you need a spatial database, there's nothing better than PostGIS)
The OSGeo.org (the open source geospatial foundation) is a great resource in that regards. Depending on your needs (geospatial is vast, see my sig), you may "need" ArcGIS from ESRI - it's commercial, it's expensive and it crashes regularly, but it does things no other GIS can easily do.
Animoog.org
If you have access to MATLAB, give the following project a shot:
http://sourceforge.net/projects/srtm-matlab/
Unfortunately, it only supports the worldwide 90m dataset, not the 30m dataset exclusive to the USA.
No idea if it works in Octave or not.
SRTM 3sec DEM is also available via anon FTP in GeoTIFF format from http://srtm.csi.cgiar.org/
As a small indie video game developer NASA's srtm data was an invaluable resource for assembling an accurate map of ancient greece for our upcoming rts Hegemony: Philip of Macedon.
My company makes use of SRTM data in our Road2electric vehicle range calculator. To tell how far an EV will *actually* go, we need to do a full physics simulation, including driver modeling, real-time weather, real-time-traffic, and of course, terrain. We first rely on NED data and use SRTM as a fallback, since NED has both higher vertical and horizontal resolution. There's a new dataset out that goes above 60 latitude, so we may be replacing SRTM data with that soon, though.
For those curious with some real-world downsides you get using such data in applications like this:
* Bridges and tunnels don't show up, so we have to hack around them by recognizing such situations and accounting for them. ;)
* The altitude data doesn't precisely match up with the locations of the roads, and in rugged areas, it can make a big difference.
* Random noise can introduce relevant artifacts into your simulation, so you need some smoothing.
* Obviously, you have to interpolate between datapoints, although it's pretty trivial.
* There are a few major errors in the dataset, places where you get huge vertical spikes (positive or negative) for one or more datapoints, then everything returns to normal. We actually make use of them to help line up our roads with the known points of the errors.
Present day. Present time.
You can find the MOLA MEGDR DEM on FTP here, or browse the datasets here.
You can get high resolution data for free through GeoBase in Canada. Your tax dollars at work..
http://www.geobase.ca/
..don't panic
You win the award for completely missing the point, and doubling down on explaining basic math to an audience that mostly can be expected to know that. (I know, I'm being too generous.) Hey, did you ever notice that a set of things that don't contain themselves looks a little funny? Someone should write congress! Or maybe Glenn Beck.
I forget what 8 was for.