This page was updated on 12 February 2006.

The files in the DEM3 folder are based on the 1°x1° files available from the 2000 Shuttle Radar Topography Mission. There are 1201x1201 16-bit cells in each file, giving a resolution of 3", which is about 90 metres. Their format is identical to the data that can be downloaded from here, and for most areas, the data is the same.

But there is one difference: there are fewer mountain no-data areas, and in some areas these have been eliminated altogether. And, unlike the data from other sources, this has not been done merely by extrapolation or even by incorporating 30" DEM data; the best available topographic maps have been used. All of the files on this site are for areas where there is a high incidence of raw SRTM mountain no-data cells.

The data should be regarded as "work in progress". Not all the no-data areas from the tiles have been removed, although the most significant ones, which are hardest to interpolate out accurately, have been given the most attention. There are still some scratches and "terraces", mainly where there are phase errors in the SRTM source data, and where data from different sources has been awkward to merge. These will eventually be removed.

For most of Europe, the main source of data does not come from SRTM at all, even where SRTM is not void. The main source is the Russian 1:100,000 topographic map series, which for Europe is near complete. More detailed topographic mapping, from a wide variety of sources, was used in high relief areas of the Alps, Pyrenees and part of Norway's Jotunheimen. The reason for the absence of SRTM is that, for some areas, I had created adequate digital elevation models before SRTM data became available. For some of these areas there could be a slight loss of accuracy relative to existing SRTM data. This data will eventually be merged with SRTM data to obtain the best from all sources.

For all areas outside Europe, plus the Tatra and Corsica, the starting point was SRTM data. Click here for a description of the infill process.

For the Himalayas, the main source of infill data is the Russian 1:200,000 series, which covers most of the world with remarkable detail. Its accuracy is variable, and not generally up to SRTM standard, but the grain of the contours seldom strays from the truth, allowing contours generated from non-void SRTM data to be extended into void areas to an acceptable degree of accuracy. Where more accurate sources exist, these have been consulted. The excellent 1:50,000 Finnmaps of Nepal were very useful; so were the 1:100,000 Snow maps of the highest Chinese summits. Local 1:150,000 maps in the Guide to Mountaineering in China also helped, notably with Minya Konka. Some older German maps of Gahrwal, Sikkim, the Rakaposhi area, and Nanga Parbat also contributed significantly. Elsewhere in the Himalayas, spot heights from orographic sketch maps were consulted, but the elevations were sometimes rejected for being SRTM incompatible. Elevations in 2003 Japanese Alpine News publications for the Eastern Himalayas also helped, especially in the area between Sepu Kangri (31/93) and Bairiga (29/96), although some of this area has not yet been fixed.

For the Andes, there is some coverage from Russian 200K maps but more detailed maps were easier to find and are more accurate, so they were preferred. Peru and most of Chile have been well mapped but in some other countries, notably Argentina, suitable maps are difficult to obtain and are often of poor quality.

An annoying no-data in the North Japanese Alps (around Hotaka-dake) was filled using Russian 1:100,000 mapping. Much of the Caucasus has been filled using Russian 1:50,000 mapping, and 1:50,000 mapping was also the source of data used to fix the many SRTM voids on the spectacular western side of New Zealand's South Island.

No data has been uploaded for North America. On request, I can supply it on DVD from the following sources: for the USA Rocky Mountains, DEM data from the USGS (1" for lower 48, 2" for Alaska). For BC and Yukon, 3/4" Canadian NTDB data; north of 60°, 3" Canadian NTDB data. But all these data, as well as good data for Mexico, are available directly from governmenmt sites. Elsewhere in the USA and Canada, existing SRTM data are near enough complete. No parts of Northern Russia, Finland, Iceland, Svalbard or Greenland have been fixed. I hope to fix the high relief parts of these areas, and eventually the rest, but 30" resolution coverage of the low relief parts is often adequate.

I have the entire SRTM Finished Product 3" SRTM dataset on 13 DVD's. This set includes defined coastline. Subject to reproduction and handling costs, I am willing to copy these for anyone interested. For my contact details, click here.

Go here for further information about my project to complete the topographical map of the world. Go here for a good example of the application of SRTM data.

Uploaded in Scotland by Jonathan de Ferranti. Last revision 12 February 2006.