Sign in to follow this  
Guest Richard Hill

Global Land Cover Characterization

Recommended Posts

Help AVSIM continue to serve you!
Please donate today!

Thanks, Dick. Quite an interesting find.It should be pointed out that:- the projections are not to the Microsoft standard. They probably need to be reprojected,- downloads seem to be pretty large - close to 90 MB for North America (including Caribbean) and 45 MB for South America, for example,- values are different for each region, so it does not seem possible to create one unique equivalency table for the whole world.Wouldn't it be nice if all these people could take our needs into account when producing these wonderful databases?Best regards.Luis

Share this post


Link to post
Share on other sites

That's exactly the data source I was using when I was making all those questions here about reprojecting.This seems to be the best data source for worldwide landclass data on the web, but, as it was just said, it's not in lat/lon projection. And if I am not mistaken it's not even referenced to the WGS84 datum.I managed to reproject the lambert azymuthal file to the way I needed with a program called Geographic Transformer, which allows processing one image with the trial version. I remember promising a 'how-to' for some guys who asked me here, but I didn't have time to do it those days and now it's wiped off from my memory.There is another source of landclass data available in lat/lon WGS84 projection, but it classifies the data in only around 10 types.Rafa

Share this post


Link to post
Share on other sites

Hi Rafa.There is a ( Olson ) 'Global Ecosystem Legend', that is almost exactly the same as FS2002's Landclass numbers... so I assume that is their source of data for both the lclookup.bgl and worldlc.bglGeographic Projection Global Ecosystems is a huge download that uses Plate Carre projection... and appears to be the same as WGS84.The uncompressed file is a RAW image of 43200x21600 of the earth. It looks to be a valid source for landclass, but is not in multiples of 256x256... I don't know what the effect of trying to use this as an INF source would be, but at 933,122,048 bytes, it's a bit more than my PC's memory would be able to handle. :)As fas as reprojection, the separate world areas ( eurasia, for example ) might be better using the 'Interrupted Goode Homolosine Projection' and trying to reproject that as WGS84. The 'Seasonal Land Cover Regions' gives a huge amount of different classifications, but these would all have to be translated to the FS2002 landclass legend... so there might be no advantage. The resolution of all the files is the same ( 1km/pixel )... so there is no more detail in any set.The regional sets also have DEMs... but there is no info on the projection or content of those files... I'll download one and see what is what... I assume it's 1km resolution, so it may just be GOTOPO30 data.Dick

Share this post


Link to post
Share on other sites

Hi all.Using the Version 2 global Landclass dataset, from the above source, and the ( Olson ) 'Global Ecosystem Legend', I'm able to make landclasses from the huge 933,122,048 byte file ( goge2_0ll.img ) with the following INF:; TestLCHuge.inf; Huge source[source]Type = ClassU8SourceDir ="."SourceFile = "goge2_0ll.img"Lon = -180.00000Lat = 89.995835NumOfCellsPerLine = 43200NumOfLines = 21600CellXdimensionDeg=0.0083333333333333333333333333333333CellYdimensionDeg=0.0083329476851851851851851851848519[Destination]DestDir = "."DestBaseFileName = "TestLCHuge"UseSourceDimensions = 0NorthLat = 45.0000SouthLat = 42.1875EastLong = -86.2500WestLong = -90.0000I used the OLD resample for my testing, and the following as a batch file:@echo offresample TestLCHuge.inftmf2bgl TestLCHuge.tmf TestLCHuge.bgldel TestLCHuge.tmfSurprisingly, my PC handled the filesize without problems, and the BGL file creation was nearly instantaneous. I suspected something was wrong, but no. The LC BGL is valid... which confirms my belief that the Plate Carre projection is identical to WGS84, and so can be used by resample. There are some numbers used in the source that have no equivalent in the FS2002 or CFS2 landclass sets.. so either the source must be edited, or the actual landclass must be extracted from the uncompressed TMF file... and the unused numbers changed to either a valid number or #254 ( transparent ).The ideal size of a world set would probably be 24576x16384... this set is actually 43200x21600... so some decision making is being done by resample. This leads to the set being altered as you play with the Cell dimensions, or the Source Lat & Lon. For example, slightly changing the variables will lead to a change in small city size, location, or even existance. That could be cured by making a separate "zzzPopulationlc.bgl" that would consist of only of cities, towns and villages, with the rest as #254, and given a higher datastream priority in the scenery folder, by naming it with the "zzz" preface. I'm sure there are other solutions.--------------------------Using this data set as a guide, it should be possible to invent a VB or Delphi program to extract the data from a small area's uncompressed TMF. That raw data could then be altered with a paint program, to place class numbers with more accuracy... then the altered data stuffed back into the TMF for compression and BGL making.Alternately, a hex editor could be used for data manipulation.--------------------------One recommended data alteration:The edges of a landclass file are usually 'squared', or unblended to the default areas. This is cured by adding a replacement border of #254 ( transparent value ) in the LC's edges. That could be done to the extracted TMF data, then replaced into the TMF file, as long as the replaced data is the same size as the original extraction.Christian Stock is going to release some detailed TMF documentation soon, so that may shed more light on the Landclass TMF file structure, and how we may replace/alter data in these files.-------------------------A more laborious approach would be to create a new world data set that actually is 24576x16384 ( or perhaps 24577x16385 ). Then we should have a reference set that has a 1:1 ratio with the TMF data, and should pass through resample without further "interpolation".Dick

Share this post


Link to post
Share on other sites

Hi all.This data set has some possibilities. :)It is not as good as the default FS2002 landclass, as some values must be made transparent ( water and coastal types ), but by changing only 8 values with a hex editor's replace function, I'm able to generate a world-wide BGL from the data.Here's the INF:[source]Type = ClassU8SourceDir ="."SourceFile = "baseworld.img"Lon = -179.99999999Lat = 89.99999999NumOfCellsPerLine = 43200NumOfLines = 21600CellXdimensionDeg=0.0083335262390333109562721359290724CellYdimensionDeg=0.0083337191536645215056252604287236[Destination]DestDir = "."DestBaseFileName = "BaseWorld"UseSourceDimensions = 0NorthLat = 89.99999999SouthLat = -89.99999999EastLong = 179.99999999WestLong = -179.99999999The cell dimensions are derived as follows:CellXdimensionDeg = ( 360 / 43199 ) ------ cells being 1 less than the number of verticies.CellYdimensionDeg = ( 180 / 21599 ) ------ cells being 1 less than the number of verticies.The landclass fit is very good, with cities located well globally, and no odd gaps or lines. The whole world takes about 25-35 minutes to resample, compress, and convert to BGL, on my system ( 640Mb memory, AthlonXP 1600+, and WindowsXP ).You should be able to use those Cell dimensions with any LOD5 Area.. just alter the NSEW destination:[source]Type = ClassU8SourceDir ="."SourceFile = "baseworld.img"Lon = -179.99999999Lat = 89.99999999NumOfCellsPerLine = 43200NumOfLines = 21600CellXdimensionDeg=0.0083335262390333109562721359290724CellYdimensionDeg=0.0083337191536645215056252604287236[Destination]DestDir = "."DestBaseFileName = "DelavanLC"UseSourceDimensions = 0NorthLat = 45.0000SouthLat = 42.1875EastLong = -86.2500WestLong = -90.0000A single LOD5 area takes just seconds to compile, even from that huge data source!The set you'd want to download for experiments is: Geographic Projection Global EcosystemsThe values I changed are ( in hex ):0E ---> FE ( water data becomes transparent )0F ---> FE ( water data becomes transparent )64 ---> FE ( unavailable data becomes transparent )4F ---> 48 ( unavailable for FS becomes Mangrove (?) )5F ---> 3E ( unavailable for FS becomes Narrow Conifers )60 ---> 06 ( unavailable for FS becomes Evergreen Broadleaf )0C ---> 7A ( unavailable for FS becomes Ice )01 ---> 67 ( Do Not Use for FS becomes City )After changing these values, I saved the file as "baseworld.img".Dick

Share this post


Link to post
Share on other sites

Dick;This is an unrelated question, but here goes.I decided to give TerraScene a try again using 7.5 DEM areas per file.I tried to use TerraBuilder but ended up with grey texture only. Could you try and explain the steps involved. I will attach a TerraScene output file if you need it for an example. I wish we would get some news on programs talked about a while back that would replace TerraScene steps.Richard L Hill.

Share this post


Link to post
Share on other sites

Create an account or sign in to comment

You need to be a member in order to leave a comment

Create an account

Sign up for a new account in our community. It's easy!

Register a new account

Sign in

Already have an account? Sign in here.

Sign In Now
Sign in to follow this