Hi all:there were a couple of posts about small bumps or ditches found in some of the recently released add-on meshes based on Shuttle Radar (SRTM) data. Strangely, those problem spots seem to be most common in metropolitan areas.In the absence of a suitable "conspiracy theory" I wanted to quickly summarize why this occurs and what users or mesh designer can do about it (it's a rehash of a post of mine from last September):SRTM data were collected by a special Space Shuttle mission, with a radar "satellite" as payload. The main difference to the usual way of generating digital elevation data (with stereo airphotos) is that the radar "faithfully" scans all features of the Earth's surface, including human-made structures.Tall buildings introduce "noise" in the radar signal, depending on the angle of "illumination" (by the radar beam) and whether it's a single structure or an entire downtown core. Also, if the radar beam hits a smooth surface it might get deflected in different directions rather than reflected back to the Shuttle's emitter. Below is part of a SRTM 1-arcsec (30-m) image of downtown Chicago (N41W088.hgt). I used MicroDEM for the display and then enhanced the image in Photoshop to highlight the problem areas. http://forums.avsim.net/user_files/64499.jpgNotice the elevated Interstate ramps radiating from the downtown core. The core itself is largely missing, due to the loss of a suitable signal as mentioned above. The areas that are there are all bumpy and obviously don't reflect the true ground elevation of Chicago. In addition, outside of the city center, smaller clusters of buildings or individual structures can be spotted as reddish 'bumps'. Pretty interesting but not very desirable for FS mesh making because the compiler will read the source data as is, that is, it keeps the bumps and fill in the holes based on the surrounding elevations.So what to do about this? First, for the US, mesh designers can use other free source data of better quality, the seamless NED 30-m data. They are not more difficult to compile than SRTM data, as Dean has pointed out here: http://forums.avsim.net/dcboard.php?az=sho...g_id=1485&page= However, if one choses to use SRTM data, the compiled mesh can be easily fixed after the fact by external "flatten switch" files. The best and most versatile tool for this task is Steve Greenwood's FSTflatten, availabe here: http://www.fs-traveler.com/flatten.shtmlHope this is useful information.Cheers, Holger

Shows that just running the data through the compiler and publishing the results is not producing the best mesh. The person that creates the mesh has a tedious task of "fixing" it before distribution. I find that the FSGenesis mesh seems to be the best in this area since Justin makes obvious attempts to fix odd areas and follows up with added fixes when warranted. I have also used Lago Terramesh and, although not under extensive "flying", found it to be very good as well.Dick Boley @KLBE

Hi Dick:All of the FS Genesis US products are made from NED data rather than SRTM but you're right about the non-US products, which are based on SRTM (except for the new Cordillera Mesh of western Canada, which is based on the excellent CDED1 23-m source data).I should also point out that the image above is a sample of the high-res (1 arcsec) SRTM data. Most SRTM add-on meshes are made from 3-arcsec (90-m) data, which are based on the same source data but "smoothed" down to 90-m resolution by NASA/NGA. It's possible that the reduced detail also makes some of the bumps less pronounced but obviously it doesn't filter them out completely.Anyone interested in more technical details, check here: http://www2.jpl.nasa.gov/srtm/index.htmlCheers, Holger

HolgerAfter reading your post above it seems like SRTM data isn't a great source for creating DEMs. Was this data collected with some other use in mind? ThanksDavid

Hi David:good question! Actually, the main objective of the SRTM mission IS to create a global (N60-S56) seamless, accurate, and consistent digital elevation model. This is what the website states:"There are lots of ways that people make use of topographic data, and for various parts of the world, maps of Earth's topography are limited, inaccurate, or nonexistent. Lack of standardization limits the scope of regional or global studies where precise topography is important." --> http://www2.jpl.nasa.gov/srtm/whymaptheworld.htmlHowever, geopolitics and perhaps technical issues seem to have interferred and delayed the achievement of the goal. NASA/JPL staff are working on a more accurate version, free of the holes and bugs mentioned above, by merging data from two different passes (i.e., different angles). When and if these data will become available to the public I don't know.Keep in mind, though, that for most of the planet the best freely available alternative are GTOPO30/DTED0 data, which have a resolution of 30 arcsec or approx. 1km (most of the FS default mesh is made from these data).For Canada, the US, and a few other countries, free data of better quality than GTOPO30 or SRTM are available but for the "rest" of the world, SRTM-based add-on meshes are the way to go and, despite all the problems, a great improvement over the default mesh.For the mesh designer, there are several tools and methods to improve the quality of SRTM data prior (!) to compilation. "Blackart" by John Childs, is one of them: http://www.terrainmap.com/ I recently ran a comparison of various options for filling in holes and Blackart came out on top of the list: http://forums.avsim.net/dcboard.php?az=sho...ing_type=searchHowever, incorrect elevations, such as the inner-city bumps, can not be easily fixed with available tools, which makes external flatten switches the better choice.End of lecture - wake up everybody ;-)Cheers, Holger

Hi HolgerThanks for the reply. You've answered my question and given me other interesting info to boot. I seem to be developing an interest in this stuff and I will have a look at the links you've provided. Thanks again David

This is a great demonstration, Holger, and a real eye-opener for most people here who have been using add-ons of all sorts without really understanding the problems and existing flaws. Perhaps this goes to show that accuracy is not important for most people, rather they want greater realism.And you do well to dispel any suppositions as to some kind of government interference in source data, when the inherent flaws in SRTM methods can easily explain all the problems.As you say, though, for those of us in the rest of the world, SRTM data lets us create more detailed altitude mesh quickly and easily.Thanks for the explanation.Best regards.Luis

Hi Luis:"Perhaps this goes to show that accuracy is not important for most people, rather they want greater realism."Very good point! All of us probably have a few places we know well and would like to see rendered in FS as close to reality as possible but for the "other" 99% of the FS world the key is to suspend the disbelief; as long as we don't see any square mountains or high-rises wedged into mountainsides we're pretty happy - Well, I am ;-)Cheers, Holger

Thats very interesting.One thing Ive noticed recenty was with the SRTM sourced Italy country mesh. When the milan VFR scenery came out with this SRTM mesh, right at the point where the san siro stadium is located there is a small hill. This hill is in exactly the same spot as the stadium and also is about the same height so obviously that small hill is what the data saw as the stadium. This then makes the stadium seem to float in mid air so Im trying to find a way for that to be fixed so that it goes flat.Craig

Hi Craig:Bingo! Congrats, that's the first confirmed building "bump" in Europe ;-) I was wondering whether any of the bigger cities covered by the European SRTM meshes would have similar issues.To fix it, just follow the link in the very first post to Steve's FSTFlatten, install, fire it up, slew around the stadium and choose the appropriate ground-level elevation. Compile the flatten switch and place it in SRTM mesh's scenery subfolder. Share with others if you like. Done. Cheers, Holger

Hi Craig.As usual, Holger has explained it right. Good find for the stadium!In another post I refered to "dirty" data. This is a good example of what I meant.The SRTM data is raw... it hasn't been cleaned yet. Cleaning will be done by geographers, scientists, and mathematicians hired by government agencies. It will be a very long process. The USGS has been steadily updating the NED dataset for the US. They will include the SRTM data, as it is cleaned up.What this means is: The newer meshes, based on raw STRM data, will have errors that either require us to suspend our sense of reality ( as Luis indicates ), or we need to fix the spikes, holes, bumps, etc... as we find them. We'd hope the creators of the newer SRTM meshes would then take responsibility to fix the errors... or at least warn the endusers of the problems and let them know they are on their own to fix them. It is very simple to make SRTM based mesh. It can be very difficult to fix problems that arise by it's use.Unfortunately, some newer designers have no clue there is even a problem, much less have the skill or time to fix those problems. The commercial products of Justin Tyme and Steve Greenwood are backed by their commitment to fix errors in their meshes as they are found. Are they earning their money? Absolutely. Their products are reasonably priced, and they have great business integrity. Each of their websites are a treasure chest if information about the process of mesh creation, and how it will appear in the sim.Holger's commitment to update his freeware meshes is a sterling example of the best of freeware for flight simulator... not that there will be no errors, but that he will not abandon the enduser. He will fix problems. And he does it without cost to the enduser. One of the best bargains for Flight Simulator.Dick