The author is senior engineer at Cavell, Mertz and Associates Inc.
When we say “propagation software,” we refer to models that run pathloss (signal propagation) studies, as distinguished from the software platform that manages the process and plots the results geographically over a map. This software model is vital to the result, but most models were developed decades ago and have remained unchanged, despite the advancement of computers and digitized earth data.
The most familiar model, and the most frequently used for FM, is the FCC’s F(50,50) field strength curves. These curves are based on empirical data going back to the beginning of VHF television in the 1940s. For a given transmitting antenna height above terrain along the first 2 to 10 miles (3.2 to 16.1 km) of a path, they provide a statistical estimate of field strength (the “F”) where half of locations have a lower field and half have a higher field (the first “50”). Atmospheric fading at the 50th percentile is considered as well (the second “50”). Because early television reception often required elevated outdoor antennas, the curves were based on the fields at 30 feet (9.1m) above ground. The 88–108 MHz FM band shares the same curves with low-band VHF TV, covering 54–88 MHz.
|Fig. 1: This coverage map shows the FCC F(50,50)
predicted 60 dBu contour of hypothetical station with omnidirectional antenna.
Green shading of parkland generally indicates the Front Range of the Rocky
Click To Enlarge
These curves are considered an “area” model because they represent the average case for large areas only. Other than the effects on average terrain elevation from 2–10 miles, the curves do not consider the effects of terrain obstructions. When mapped as iso-contours representing a single field strength, the curve predictions tend to produce a smoothed, single line, such as the 60 dBuV field strength for the hypothetical station in the foothills of the Rocky Mountains shown in the map in Fig. 1. This transmitter’s height above average terrain to the west is low, due to the surrounding mountains. But to the east, the land drops to relatively low and smooth plains, producing a much larger and more even radius around the transmitter.
RF engineers working in VHF and higher frequency bands have long sought better predictions than area mode contours, using terrain-sensitive pathloss models. The granddaddy of models is the Irregular Terrain Model, aka Longley-Rice, which was developed in 1967 at the Institute for Telecommunications Science, part of the U.S. Dept. of Commerce, in Boulder, Colo.
It is widely used because it’s free and open-sourced. For example, the FCC uses it officially for the DTV station service and interference studies. Other than a correction to the FORTRAN code 30 years ago, it is unchanged.
TERRAIN REVEALS DETAIL
|Fig. 2: This shows the FCC contour overlaid with
Longley-Rice predicted signal strength values. Red, green and yellow overlays are
ITM’s prediction of in-car service at 33 dBu, and 60 dBu and 80 dBu field strengths,
all at 1.5 m above ground level.
Click To Enlarge
The effects of terrain are striking for our Rocky Mountain foothills example, shown as a three-color underlay in Fig. 2. The predicted 60 dBu field strength is shown as the green shading, although it should be noted that the field strength is shown at a height of only 5 feet (1.5m) above ground, as is more typical of FM radio antennas on cars and portables. If predicted at 30 feet, ITM would have produced a much larger area than the FCC 60 dBu curves would show.
Digitized terrain databases used by ITM have improved tremendously over the years, from 1,000-meter gridding in the 1970s to less than 30 meters today, thanks to NASA’s Shuttle Radar Topography Mission data. This increased resolution by greater than 1000 to 1, which could hardly have been anticipated in 1967. Little scientific research has been done since to know how much this improves the accuracy of pathloss predictions, if at all. My experience is that it increases the “noisiness” of pathloss on a point-to-point basis, but the reduction in standard deviation of error relative to measured data is rather limited when viewed on the scale of broadcast station coverage.
Another development in terrain data is in morphology, better known as “clutter data” or “land cover data.” This data didn’t exist when ITM was written as a “bare earth” model, lacking any input for and correction of clutter loss. It stands to reason that a grove of trees or a built-up urban area would have more pathloss than open land, but there are no studies that I am aware of that thoroughly determined the optimal correction values with current morphology data. Consequently those who use it as a “correction” to ITM pathloss are really just guessing at the supposed improvement. The corrections I’ve seen assume a loss adjustment for the “bin you’re in” (the terminal point of a path, usually the receive point). However, the length of the path through multiple types of land cover has not been studied.
Pathloss correction in built-up areas is especially challenging because the signal diffracts over local building clutter and reflects from large structures. Land classification data lacks individual structures for this calculation. There are 3D building databases (often used by cellular network designers), but they are expensive and computationally intensive.
There are also errors with the classification of land cover. The classifications were developed for land uses such as agriculture, not RF engineering; so we get divisions according to types of grass and brush, for example, but little identification of how built-up areas affect pathloss at VHF or UHF frequencies. There are also some outright mis-classifications, such as designating open roadways as “urban” clutter. An eight-lane freeway is really an open environment.
A lot of work is needed to better correlate land classification data with signal propagation models.
Despite its popularity, there are other known shortcomings to the accuracy of the Irregular Terrain Model.
One notable effort to improve ITM was the development of SPLAT! — short for an RF Signal Propagation, Loss, And Terrain analysis tool — in 2005 and later. This unofficial revision to ITM addressed the use of higher-resolution terrain data and other issues. These changes may have improved ITM’s accuracy in various ways, but the broadcast industry, along with other users, needs comprehensive scientific field studies to verify these revisions.
The International Telecommunications Union developed Recommendation ITU-R P.1546, a well-established model for point-to-area pathloss. However, it is based on statistical methods (hence the point-to-area designation) and field strength curves measured a long time ago. The ITU-R P.1812 is a commendable model for path-specific prediction and continues to be refined. However, its conversion to a software implementation is generally available only in high-end RF planning tools.
Because the profile between the origination (transmitter) and each point along the path changes constantly, a PTP model can show large variations in field strength over small changes in distance. These larger-scale effects are why PTP models can show rather striking “holes” and “islands” in the fabric of coverage that is draped over the map.
This geographic variation is separate from the sub-wavelength changes called Rayleigh fading, which are caused by local multipath propagation. This fading is treated by most PTP models as a statistical variation around the local mean field strength. Like the FCC curves, PTP models express their predictions by a percentage of locations meeting a given pathloss. There is also a specification for a percentage of the time that the pathloss is met, to account for atmospheric fading.
This story is excerpted from the Radio World eBook “Propagation Analysis for Profit,” one of the growing Radio World library of free eBooks.
PTP models are really a collection of models, representing losses from diffraction over major and minor terrain obstructions, Fresnel zone attenuation, ground reflection effects, atmospheric scatter and refraction, height-gain and more. They may switch from one mode to another, or combine some of them at each point along a path profile, as needed. Depending on conditions, one model may be more accurate than others, and there are no comprehensive comparisons to rely upon.
The difficulty with improving or developing new signal propagation models is the exhaustive testing required to validate a model over a vast range of conditions. The development of National Bureau of Standards “Tech. Note 101” and the ITM were based on more than a decade of work at the Institute for Telecommunication Sciences in Boulder more than a half-century ago. Work to improve and validate new models takes time and money, which has not been supported by users in the VHF and UHF bands. This is unfortunate since reduced statistical confidence from these models requires over-building of radio frequency networks, which increases their cost.