I. SRTM DEM Acquisition

SRTM 90m (3 arc second) data is distributed by Continent and is available in 1 deg tiles first on the SRTM FTP server (ftp://edcsgs9.cr.usgs.gov/pub/data/srtm/). Eventually it will be available in a web interface on the USGS Seamless Server. Sometime in 2004 all continents will be available at 90m resolution, currently only North America is available as far south a 15 deg N. The address of a given tile is derived from the lat/long coordinates of the SW corner of that tile. 

II. Converting to Grid

• Importing the SRTM tile in ESRI ASCII GRID 
• Unzip the SRTM tile, it is Winzip format. Put the data into a GIS data directory. 
• The contents will be in .HGT format. ESRI products can import .HGT files if they are renamed to be .BIL files and a text header file (.hdr) is created. 
• Change the .HGT format extension name to .BIL. Then create a .hdr file for the file using a text editor like Notepad. Here is an example for N18W077.bil. It should be should be named N18W077.hdr 

BYTEORDER M LAYOUT BIL NROWS 1201 NCOLS 1201 NBANDS 1 NBITS 16 BANDROWBYTES 2402 TOTALROWBYTES 2402 BANDGAPBYTES 0 NODATA -32768 ULXMAP -76.9995833333 <- This must be changed adjusting to each file. Apply this formula to calculate the right parameter. (WYYY - 0.0004166667) for the Western zones OR (EYYY + 0.0004166667) for the Eastern zone. ULYMAP 19.00041666666667 <- This must be changed adjusting to each file. Apply this formula to calculate the right parameter. (NXX + 1.0004166667) for the Northern hemisphere OR (SXX – 1.0004166667) for the Southern hemisphere. XDIM 0.000833333333333 YDIM 0.000833333333333  (note, remove the <- XXXXXXXXXX text)

After creating a header file for the HGT file (now it should be BIL file), convert this file to GRID to use it for analysis. In ArcCatalog browser right- click on the filename and go down to Export to a Different Format... , this is the way to Export to GRID. 

III. Initial processing
Projection / Coordinate System:
ArcCatalog or ArcToolbox can be used to Define Projections for GRID files. To use ArcCatalog, right-click to get the Properties for a GRID file. Next click the "Spatial Reference tab", and click "Edit..." to redefine the projection/coordinate system. 

Three dialog boxes follow, select the following:

• Projection: GEOGRAPHIC 
• Geographic Units: DD (decimal degrees) 
• Datum: WGS84 No offset 

Open the GRID in ArcMap with Spatial Analyst extension turned on. Change the working directory of the Spatial Analyst immediately from the default directory. Path and directory names must be no more than 13 characters (or 8 to be on the safe side), as far as I know of (Chieko). In order to change the working directory, click the Spatial Analyst tab, go to Options, and check the General tab. Change the working directory by browsing.

In Spatial Analyst Raster Calculator apply the following GRID command, substituting your file name for 'in_grid': 

• con ( [in_grid] >= 32768 , [in_grid] - 65536 , [in_grid]) 
• Then apply this command.
• setnull ( [in_grid] < -9998 , [in_grid] ) 

Save the resulting "CALC" raster to a real GRID file with the "Make Permanent..." command.

More information is available on converting these files in Workstation GRID using IMAGEGRID and a related AML script.

(Optional 1)
If you are working on multiple rasters, then it is better to merge them. To merge them apply

• Merge([in_grid1], [in_grid2], [in_grid3],…, [in_gridn])

OR

• Mosaic([in_grid1], [in_grid2], [in_grid3],…, [in_gridn])

(Optional 2)
After merging or mosaicing rasters together, you may find gaps of No Data left between them. Use Raster Calculator to interpolate data into those gaps based on neighboring cell values.

• con(isnull([in_grid]), focalmean([in_grid1], rectangle,4,4), [in_grid1])

(Optional 3)
Hillshade Note: As with the coarser Gtopo30 data, the default Spatial Analyst Hillshade Z factor of 1 results in very exaggerated mountainous terrain when the data is in Geographic projection. This is because the horizontal spatial units are degrees, but the vertical units are meters. Therefore in Geographic project try a Z factor of around .0001 or .00001 with SRTM 90m data. Putting this hillshade layer behind the elevation layer with the elevation layer set to 40% transparency (especially using the more exaggerated .0001 Z value) looks quite good.