Elevating Accuracy in Longley-Rice Coverage

Using the most accurate data possible when designing a new broadcast facility reduces the chances of real-world coverage problems.
Publish date:
Social count:

In past columns, I've discussed how incorrect assumptions and simplifications result in misleading coverage studies. Accuracy suffers when the actual antenna elevation pattern, including the effects of mechanical beam-tilt, is not used. While these simplifications may be necessary when analyzing interference between a large number of stations, using the most accurate data possible when designing a new broadcast facility reduces the chances of real-world coverage problems.

Suppliers of propagation software offer terrain data, usually in a compressed, proprietary format optimized for use with their software. Most Longley-Rice studies now use data based on the 3 arc-second National Elevation Dataset (NED) terrain data available from U.S. Geological Survey.

USGS datasets based on point spacing as small as 1/9 arc-second are available for more accurate studies, although I haven't seen this resolution used for broadcast studies. The terrain data by the FCC for OET-69 studies is available at http://www.fcc.gov/oet/dtv/dtv_apps.html.


In late May, USGS released the Shuttle Radar Topography Mission (SRTM) "finished" data in 1 and 3 arc-second resolution. This data was collected using dual Spaceborne Imaging Radar and dual X-band Synthetic Aperture Radar on board the Space Shuttle Endeavor in February 2000. The radars acquired two images at the same time, which were combined to produce a 3D image that contained the elevation data. This data, however, was of limited use without some post processing.

As readers know, some materials reflect microwave energy well and some materials absorb it. These variations caused "spikes" and "wells" in the data that were removed as part of the finishing process, which also included delineating and flattening water bodies, better defining coastlines and filling small voids.

The finished SRTM data has replaced the research-grade data in the USGS Seamless Data Distribution System (http://seamless.usgs.gov/). Research-grade data is still available from NASA's Jet Propulsion Laboratory (http://www2.jpl.nasa.gov/srtm/).

The research-grade data in JPL's initial release uses the unedited SRTM data for the 1 arc-second dataset, and averages the elevation over the nine 1 arc-second samples in each 3 arc-second cell to create the research grade 3 arc-second database. The 3 arc-second version of the finished dataset, however, subsamples the 1 arc-second data by using the value of the center cell of the nine samples, discarding the other eight samples.

How accurate is the SRTM data? One arc-second corresponds to about 30 meters in mid-latitude areas like the continental United States.

(click thumbnail)Fig. 1 SRTM finished data image of the Los Angeles area in 1 arc-second resolution.
USGS says that the finished SRTM data meets the horizontal and vertical accuracy requirements of 20 meters (circular error at 90 percent confidence) and 16 meters (linear error at 90 percent confidence), and that vertical accuracy is actually closer to +/-10 meters. To see what the data looks like visually, visit http://seamless.usgs.gov/ and play around with the interactive map (click on the U.S. map on the right side). This map will come up with default datasets, so you will want to modify them using the "Layers" menu on the right side of the page after you zoom into the area you are interested in. It is interesting to compare the SRTM data with the NED data.

If you look closely at Fig. 1 and Fig. 2, you will notice that unlike the NED, the SRTM terrain data includes large buildings that reflected the radars' signals! The SRTM elevation data also shows the height of heavily wooded areas. Dense forests will block UHF signals, but until the SRTM data became available, elevation datasets did not include the height of the forest canopy.

To get a closer look at these maps, view them at http://www.transmitter.com/images/LA_Area_1_sec_SRTM.png and http://photojournal.jpl.nasa.gov/catalog/PIA02783.


(click thumbnail)Fig 2. An SRTM perspective with Landsat overlay of Manhattan courtesy of NASA/JPL/NIMAHow much of an impact do buildings have on coverage? V-Soft showed the SRTM dataset integrated with its popular Probe 3 propagation software at NAB2005. I stopped by the V-Soft booth and working with two of the V-Soft engineers, I tested the software with a transmitter site in New Jersey west of New York City. The resulting map clearly showed the impact of the Manhattan buildings on the signal.
Signal Strength
(click thumbnail)Fig. 3 Signal strength image generated by V-Soft using USGS NED 3 arc-second data.

WHITE: <70 dBuV/m
RED: >70 dBuV/m
GREEN: 80 to 100 dBuV/m
YELLOW: >100 dBuV/m
The blue areas represent water

(click thumbnail)Fig. 4 Signal strength generated by V-Soft using SRTM data
John Gray from V-Soft refined my quick experiment using a site most TV broadcasters in New York know well--the Armstrong tower at Alpine, N.J. Many TV stations used the site after the destruction of the World Trade Center on 9/11 until new facilities were constructed at the Empire State Building. This study is based on a hypothetical DTV station on Channel 40 using an isotropic antenna with its center of radiation at 286.5 meters AMSL and an effective radiated power of 30 kW. To highlight the impact of the buildings, only field strength levels above 70 dBµV/m were plotted.

The maps show signal levels above 70 dBµV/m as red, from 80 to 100 dBµV/m as green, and above 100 dBµV/m as yellow. The blue areas are water. Areas where the signal is less than 70 dBµV/m are white.

You will notice in Fig. 3, the map using the USGS NED 3 arc-second database, shows greater than 80-dBµV/m field strength over most of Manhattan and the western parts of Queens and Brooklyn.

However, when the SRTM data with the buildings included is used, as in Fig. 4, field strength drops below 70 dBµV/m in significant areas on the eastern side of Manhattan and the western part of Brooklyn. If you look closely at the signal strength in New Jersey, you will notice some reduction in coverage, possibly from forested areas.

Visit the V-Soft Web site at www.v-soft.com for more information on its products using the SRTM data. Now that the finished data is available from USGS, I expect other companies to offer SRTM terrain databases with their propagation software.

Since there are holes in the SRTM coverage and a certain amount of "noise" in the data, it is worth asking how the database handles this. A comparison with USGS NED elevations could eliminate elevations that are impossible in nature, but care is needed not to eliminate buildings. Averaging can be used to reduce the noise, but could also have the effect of reducing the height of buildings or other small obstructions.

Although my focus this month has been comparing USGS NED elevations with those from SRTM, SRTM elevation data covers most of the world. In many areas, terrain elevation data is inaccurate. SRTM data will allow engineers designing facilities in these areas to more accurately determine the best transmitter locations, optimum antenna patterns and number of transmitters needed to coverage an area.

One final note--if you downloaded the SPLAT software I described in my column in the Feb. 2, 2005 issue, (also available on this site), you are probably wondering whether you can use the freely downloadable SRTM data in SPLAT.

Unfortunately, I haven't been able to find a utility to convert the formats available on the USGS Seamless Data Distribution System (ArcGrid, GeoTIFF, BIL and GridFloat) into the SPLAT Data Files. If you find a way to convert the SRTM supported formats into the SPLAT SDF format, please let me know.

If you had trouble finding the 3 arc-second, 1 degree, 1:250,000 elevations files used with SPLAT, try http://edc.usgs.gov/geodata/. USGS has indicated distribution of DEM files will change June 30, 2005. See http://edc.usgs.gov/DEM_DLG_discontinued.html for more information. It appears the files will still be available for a fee through the USGS EarthExplorer.

Your comments and questions on anyRF topic are always welcome. Drop me an e-mail at dlung@transmitter.com.