Heightfields (available as .hgt files) from the joint Shuttle Radar Topography Mission (SRTM) can give excellent results in many 3D visualization programs. Unfortunatly however in some areas, particularly mountains, they contain considerable numbers of holes of varying sizes. In extreme cases whole mountains may be missing.

Filling in this missing data can be hard. Most traditional heightfield editors use some form of progressive infilling, which largely just smooths over the holes. This can be an effective solution where patches of missing data are small, but is not really suitable with areas of high relief, such as mountains, as these methods do not make use of all the available information in the landscape.

Fortunatly Landformer Pro's advanced interpolation system can be deployed to considerable success for these difficult areas. Landformer Pro can also make use of the corresponding GTOPO30 data. GTOPO30 contains on 1% of the points that the SRTM data does for any given area, but where the data is accurate this the additional information can be used to guide the interpolation system towards a fully realistic result.

For this tutorial we'll demonstrate how to interploate a missing moutain ridge in the Alps and generate an accuratly interpolated, geocorrected, Terragen terrain.

Patching the .hgt file

You will need Landformer Pro 2.5 or later. Open the program and from the File menu choose Import then SRTM data as the file type. For this demonstration we loaded N46E009.hgt - a file from the Swiss alps with quite large holes in some areas. A section of Landformer Pro is shown below displaying a large hole we will interpolate.

Now we'll import the matching DCW file and patch. Click Load DCW and choose the DCW file suggested by Landformer Pro (download from the USGS). The DCW heightfield is displayed below the HGT Tools buttons - note that Landformer Pro automatically picks out the matching section of the DCW file. Click Patch HGT to load the matching DCW points into the HGT heightfield. The image below shows the results of these actions.

Although the DCW data is nowhere near as details as the SRTM, it does give our interpolation step additional clues. Note that the use of GTOPO30 data is not essential and in some area - such as the Himalayas - the GTOPO30 data may not be of sufficient accuracy and is better ignored.

Now we apply a Nearest Neighbour interpolation to our area of interest in the usual way - here the mountain ridges in the centre of the image. Note as, unlike contour tracing, a large number of points are already defined the interpolation can take several minutes. When we have finished our results should look like this.

Our final step is to choose an area to geocorrect to create our Terragen file (you may wish to save your modified .hgt file before doing this). Click the Sample for .hgt correction tool from the toolbar and click on the map where you wish the top right of the tile to be. A red rectangle is draw indicating the area to be corrected - the dimensions of the rectangle are automatically set by Landformer Pro to be correct for the .hgt file loaded.

Once selected click Resample to generate the geocorrected Terragen heightfield. We're just using a small 257x257 sample here for demonstration.

Finally save the grid as a terragen file and view. It's as easy as that!

Further enhancements and techniques.

Because Landformer Pro was originally developed with a strong emphasis on generating heightfields from scanned contour maps the interpolation methods used above fit neatly into the program's methadologies. You can further enhance the interpolation by using know spot heights or contours from topographic maps, the more information you feed the interpolator the more accurate your results. Of course in less mountanious areas where holes tend to be smaller good results can be obtained without adding DCW data.

Of course, the interpolation system can also be used to edit landscapes to intelligently add quarries, cuttings, mounds etc., but that's another story.

Finally you may wish to compare the results of our intelligent interpolation methods against the infilling used by other programs. The difference should be obvious from the heightfield images below!

Area interpolated using a freeware SRTM fill program from another manufacturer. Note the discontinuities in the data caused by infilling algorithms. This program runs without any parameters so we have not 'doctored' the results in any way.		Area interpolated using Landformer Pro. Even without showing in 3D the quality of the interpolation should be easily apparent. Note neither of these images have been geocorrected as the option was not available in the freeware program.