Longtime readers know that I’m fascinated by RF propagation—how RF travels through the air from the transmitter to receivers and the obstacles it meets on the path. Many models have been developed to predict how much of the transmitted signal, if any, will arrive at a receiver in a given location. Accurate field strength and availability predictions will be important for new ATSC 3.0 business models.
Before the age of personal computers, graphs were used to determine signal level at a distance. The Federal Communications Commission F(50,50) and F(50,10) curves released 60 years ago (Google “FCC Report No. R-6602” for details) were based on field measurements. The curves are combined to determine the F(50,90) contour used today for DTV service-area contours and can be found in Section 73.699 of the FCC’s rules.
When the FCC had to find a second channel for every TV station to allow them to transmit ATSC 1.0 with comparable coverage, FCC contours and the FCC separation distance tables didn’t work, as not enough channels were available.
Modern Coverage Challenges
By this time, computers were available that could calculate field strength inside the coverage contour. The Longley-Rice Irregular Terrain Model was used to calculate the signal level and interference in 1- or 2-kilometer cells in the coverage area and pack stations more closely together.
One early problem was that different computers with different CPU and math coprocessors generated slightly different results. Eventually, computers became more accurate and, with the release of TVStudy in 2013, any computer capable of running the FCC’s software provided the same result.
Today, FCC contours aren’t of much use in determining whether a viewer who says they can’t receive the station is in a bad coverage area or has a problem with their antenna. They also aren’t much use in designing an optimum transmit antenna pattern or maximizing coverage.
Fortunately, there are now tools available that show terrain-sensitive coverage, including websites like Rabbitears.info. TVStudy, as I described in my January column, can be used to create your own studies and coverage maps.
However, as I’ve been reminding readers since my column in the July 5, 2005, issue of TV Tech (“Elevating Accuracy in Longley-Rice Coverage”) and in later articles, there are limits to the accuracy that can be achieved with Longley-Rice, especially using the planning factors in FCC OET Bulletin 69 that are based on a directional outdoor antenna 30 feet above ground. Without considering lower antenna heights and surrounding structures, the actual field strength at the antenna will usually be much lower than predicted.
Businesses or agencies using ATSC 3.0 to deliver critical services will demand more accurate coverage models and broadcasters will need them to design single-frequency networks and position gap fillers. Better models are used in other countries and by the wireless industry, so why not by broadcasters in the United States?
Businesses using ATSC 3.0 to deliver critical services will demand more accurate coverage models, and broadcasters will need them to design single-frequency networks.”
One of the reasons Longley-Rice has endured and remains the FCC model of choice is because the core software that provides repeatable results is available at no cost. The inputs required are reasonable and, at least for the U.S., all required data is available in the FCC’s free TVStudy program.
Software models predict coverage by calculating loss over the different paths and propagation modes the signal can use to reach the receiver. If there is line-of-sight, that’s easy. If there is an obstruction, how high is it and how does the signal diffract over it?
There are different ways to calculate this depending on the type of obstruction, as the refraction index of the air the signal is passing through also matters. The model’s accuracy largely depends on how it handles diffraction; very complex models use ray-tracing to account for reflections from structures but probably aren’t practical for a station’s entire coverage area.
Revisiting ITU P.1812
In previous articles, I discussed the ITU Recommendation P.1812 propagation model. (Google “Doug Lung” ITU P.1812 for a list of them). Because ITU P.1812 is a widely tested and publicly available model, similar to Longley-Rice ITM 1.2.2, it is the model I see as most capable of replacing it. It does a better job handling diffraction than Longley-Rice and, with the addition of a land-use database, does a better job in predicting path loss to antennas below clutter height, where most TV antennas today are positioned and where portable and mobile devices using new ATSC 3.0 services will be located.
The good news is Ivica Stevanovic has developed a Python implementation of ITU P.1812 that can be downloaded from this link. The software is free to use, but requires some restricted files containing location-specific climatological parameters that must be downloaded (at no cost) from the ITU website.
I tested Py1812 and it works! Inputs required include effective radiated power, frequency, location, transmit antenna height above ground, receive antenna height above ground and path data. The path data is an array of points from the transmitter antenna to the receive antenna, with each point containing the distance from the transmitter, the ground elevation (AMSL), the clutter height and the climate zone. The GitHub repository includes a validation program that compares Py1812 output with ITU data taken from real-world propagation paths in different environments and field measurements.
I’ve developed some simple Python scripts to create the path array using a digital elevation model (DEM) and the NLCD land-use database. I’m still working on combining it with a Python program using spherical geometry to calculate ERP from a directional antenna (including one with mechanical beam tilt) at the elevation and azimuth angles to the first obstruction on the path. When complete, I’ll use it to compare results with field measurements I’ve taken and adjust the clutter height table to improve accuracy.
Put to the Test in L.A.
To test the software, I plotted path loss and fields from Mount Wilson on TV Channel 36 at 1,000 kW ERP 160 meters above ground to a location in downtown Los Angeles with the receive antenna 4.3 and 10.0 meters above ground.
Fig. 1 shows calculated field strength from P.1812 and TVStudy’s Longley-Rice, both using USGS 1-second terrain data and NLCD land-cover data. For comparison, I also studied the path using SPLAT!’s Longley-Rice model with the SRTM 1-second DEM. The TVStudy field-strength predictions are higher than P.1812, even when land use is taken into account. There is no difference between 4.3- and 10-meter receive antenna heights, as it sees both as line of sight and the same land-use loss applies to both.
TVStudy/Longley-Rice assigns a fixed dB loss to different land-use categories, while P.1812 assigns a height to them and calculates the diffraction loss over the surrounding structures like trees, open field, suburban, dense urban, etc. I modified the P.1812 land use/height table to use other categories available in the USGS NLCD database.
Note that the field strengths are F(50,50), as ITU P.1812 does not allow time variability over 50%. F(50,90) field strength can be calculated using the formula the FCC provides for calculating F(50,90) given F(50,10) and F(50,50) field strengths.
I hope to have more detailed results to report in a future column and, ideally, maps comparing coverage from TVStudy’s Longley-Rice model and ITU P.1812.
If you are interested in this project, email me.
