New SRTM_3ArcSec_To_BGL - Fixes holes in SRTM data

[Guest]

Just uploaded v1.02 to the library.It now interpolates any missing SRTM data. So that means no more holes in your SRTM mesh files :-jumpyFor those who don't know what this program does... all you have to do is place this program in a directory with some SRTM 3arc second (ie 90m) files, and it will create BAT and INF files to automate the production of FS2002 mesh from the SRTM data.NASA SRTM mesh data can be downloaded herehttp://edcsgs9.cr.usgs.gov/pub/data/srtm/ Matthew.

[Guest moomox]

Freakin' Awesome man!I've got form N43 to N47 done.Now I can complete all of North America today with your update!Woooo Hoooo!

[Guest]

Glad you like it...The interpolation (ie hole fixing) could be better. I'll improve it later.Also you may want to build LOD8 mesh as well... Since the LOD8 mesh is used at longer distances than the LOD9 mesh can be displayed. So this means your mesh will look better from a distance. Just edit the .INF file, and change the LOD = 9 to LOD = 8 and rerun the batch file. Make sure you rename the existing LOD9 .BGL first.Also working on a SRTM_1ARCSEC_TO_BGL utility for the 30 meter SRTM data of the USA.BTW, build some mesh for the Caribbean... makes a big difference! Now all those islands actually look rugged (volcanic even) :)Matthew

[Guest moomox]

Might just try the LOD8. Any iea if it's a higher detail terrain? I'm curious about whether any of you have made a complete freeware mesh of north america/south america yet? I'm currently working toward that goal. Each BGL is compiled by degrees latitude. (ie N43W060 to N43W126) Each BGL is in the neighborhood of 50MB each.So it's gonna be a space hog :-lolSo far the mesh looks awesome! I'm trying to complete Canada and northern USA first. Then on to soutern USA. etc etc.It's a bit time consuming, but it's well worth it. Any one else interested in helping to compile data together?

[Holger 0]

Howdy:LOD8 is lower resolution than LOD9 (153 vs 76m). However, since the SRTM resolution is somewhere in between (depending on latitude) there's little or no noticable difference with respect to display accuracy. The big advantage is that LOD8 tiles will save you a lot of disk space.Maybe you know this already...Be aware that your tiles have to overlap somewhat to fit with the underlying LOD grid (and the LOD8 grid is different from the LOD9 grid). If you're not aware of this, check out Steve Greenwood's website; it also provides a spreadsheet tool for doing the LOD grid calculations: http://www.fs-traveler.com/fs-traveler.htmlInterpolation will NOT replace the missing data. For small holes it will do a good enough job but for larger ones expect disappointing results. For example, most of the steep canyons in Utah and Arizona are AWOL in SRTM (including the Grand Canyon), as are some of the fjords and mountains along the BC and Alaska shoreline. Simple interpolation will make a mess out of these places. For the US areas, your best bet is to download the USGS tiles of the worst areas and merge the two DEMs in MicroDEM to fill the holes.One thread to check out is this (also, do a search for "SRTM" in this forum):http://forums.avsim.com/dcboard.php?az=sho...g_id=626&page=2Finally, like all add-on meshes, there will be problems with lake elevations, rivers, and shorelines due to a horizontal offset between the FS default hydrography and the SRTM data. This often leads to sunken or raised lakes (and airports) and ugly cliffs on shorelines. Fixing it requires either tons of flatten switches (as I did for my BC meshes) or changing and recompiling the FS hydrography .bgl files, like Gilles Gauthier did for his Quebec and Maritimes meshes (search under his name in the AVSIM library).These issues are meant to be helpful, not to discourage you ;-)Cheers, Holger

[Guest sgreenwood]

Hi David,Expanding on Holger's comments, I'd like to suggest a different approach.> Each BGL is compiled by degrees latitude.The height of an LOD8 quadrant is 0.35 degrees. This means that about 35% of the area between your nominal one-degree longitudinal bands will still be default mesh. You can confirm this using the TmfViewer, which should be used to check all mesh.If you construct your mesh in smaller blocks you can avoid this problem, produce smaller files, and make it easier for folks to download smaller areas of interest. It will also make it easier for you to fill large areas of missing data with lower resolution source data, if desired. (However, this will be VERY difficult to do if the missing data has been interpolated - you may want to use the earlier version of SRTM_3ArcSec_To_BGL if you decide to fill in this data.)For example, a nominal 5-degree square block can be constructed from 6x6 1-degree hgt files (36 total). You will lose some data along 2 sides and have excess on the other two, but when combined with adjacent mesh files, the coverage will be complete. This crude approach will create some redundant mesh, but avoiding the overlapping mesh will require a great deal more effort.E.g.:[pre] N65 N65 N65 +---+ +---+ +---+ W075 | | W070 W070 | | W065 W065 | | W060 ... +---+ +---+ +---+ N60 N60 N60 [/pre]And the next row would include hgt files N65 to N70, ....Good luck!Stevewww.fs-traveler.com

[Guest]

(However, this will>be VERY difficult to do if the missing data has been>interpolated - you may want to use the earlier version of>SRTM_3ArcSec_To_BGL if you decide to fill in this data.)>I'll add a command line switch in the next version to disable my interpolation. Matthew.

[Guest]

I think we should ask AVSIM to let us host a "Project Mesh" site (eg www.avsim.com/projectmesh) on the AVSIM site, just like they do with ProjectOpenSky, TerraFish and others."Project Mesh" could be a colaborative effort to build high quality mesh for the entire world from SRTM 3 and 1 arc second data, and other data.We could break the world up into various areas and each person compile FS2002 / FS2004 mesh for their area and then upload it to Project Mesh webiteWe would continue to use the current Mesh forum for our discussions and load our work into the AVSIM file library. The Project Mesh site would link to the file library and forum.We could have a clickable map of the world that would allow users to easily download mesh files (from the AVSIM library) that interest them.I could continue to enhance the SRTM_3ARCSEC_TO_BGL and SRTM_3ARCSEC_INTERPOLATE utilities to make our jobs easier and to add new features to improve the mesh quality... including merge missing data with GTOP data instead of using interpolation for large holes.What do you think?Matthew.

[Guest sgreenwood]

Hi Matthew,>I'll add a command line switch in the next version>to disable my interpolation. Excellent solution. Apparently I've spent too much time working with the data in MicroDem today! At the moment, my approach is to auto-fill holes up to a specified radius and then fill the larger holes with GTOPO30 data. I don't know how difficult this would be for you, but an alternative might be to allow the user to specify the radius/size of the gap to auto-fill. A value of 0 could disable the interpolation, while other values would provide more control over the process. (And perhaps something like -1 to replace all missing data, as I assume your program does now.)You've created a promising program; I look forward to seeing how far you can go with it. Cheers,Steve

[Guest]

I've now made a change that allows the interpolation to fill holes from the outside of the hole to inside of hole and upto a specific radius.So small holes are filled nicely while larger holes have their middle left unchanged.What I should do next is fill those remaining (large) holes with GTOP30 or sample resampled lower LOD SRTM dataI'll try and upload something today/tomorrow. Right now though I better get back to work :)Matthew.

[Guest Richard Hill]

Matthew;Is the data derived from STRM only used for Mesh? Is their no way to use the Water, Urban, and Transportation data?Richard;

[Guest moomox]

I don't think that SRTM has any other information included in it other than elevation data. But i'm a newbie at it so I could be wrong.

[Guest]

Only elevation data in the SRTM files.However I guess you could derive the coast line since that would be where elevation is 0. But from what Ive seen of the initial SRTM data is would be quite inaccurate.Matthew.

[Holger 0]

Hi Matthew:yup, coast line extraction wouldn't work because the darned radar is so accurate that it measures the exact water elevation, which is anything but zero. I noticed in some of the tiles along the BC coast that if you change your color scheme to highlight small differences at low elevations you can actually see the wave patterns in the ocean and fjords - pretty cool. In fact, ship spotting (by their wakes) is one of the applications of radar satellites, though not of this particular Shuttle system.Cheers, Holger

[Guest rafaelds]

1. If there are already some .HGT_INTERPOLATED filed in the directory, they become .HGT_INTERPOLATED_INTERPOLATED and so on... Perhaps the program should process only .HGT files2. [MultiSource] with only one source does not work, resample's fault... The workaround is to remove the MultiSource section and rename [source1] to SourceOtherwise your software is great!Interpolation works like a charm for me (south Brasil), where all voids are under lakes/rivers or are very small dots.Thanks a lot!Rafa