[Next] [Previous] [Top]

The Effect of Antenna Covers On GPS Baseline Solutions

Results

Figure 5 shows results using the Trimble conical cover with their Dorne Margolin choke ring antenna. Results using 10, 15, and 20 degrees as the data processing cutoff angle are shown in the top, middle, and bottom panels. The green "+" marks indicate L3 height solutions. Red "*" marks indicate L3 height solutions with additional troposphere parameters estimated at one of the sites. For days 170, 177, and 178 neither the raydome cover or the metal mounting plate are attached to the antenna. Days 171, 172, 175, and 176 have the metal plate mounted underneath the antenna without the conical dome attached. Days 173 and 174 have the dome and the conical cover mounted on one of the antennas.

A few conclusions can be made from results presented in the figure. The first is that the conical dome causes a vertical height bias that is amplified by the estimation of a troposphere estimate. Days 173 and 174 have the conical dome, and the metal mounting plate, on one of the antennas. On these days, the height component of the solutions disagree with ground truth by up to 20 mm. A second conclusion is that the magnitude of the disagreement varies as a function the elevation cutoff angle used for processing, particularly when solutions with tropospheric delay parameters are estimated. Again, for days 173 and 174 when the conical dome is mounted on one of the antennas, the height solution varies by almost 30 mm when comparing solutions using a 10 and 20 degree elevation cut off. The third conclusion is that the metal ground plane which is mounted underneath the antenna also causes a bias in the height component of the solutions. The metal plate was mounted underneath one of the antennas without the conical dome on days 172, 173, 175, and 176. Biases ranging from 2 mm for 10 degree elevation cutoff to 5 mm for 20 degree elevation cutoff when compared to ground truth are evident for solutions without troposphere estimates. For solutions with troposphere estimates, variations of up to 10 mm (for troposphere solutions using a 20 degree cutoff angle) from ground truth on these days can be attributed to this extra piece of metal.

Figure 5: Baseline height solutions using the Trimble Dorne Margolin choke ring antenna and the Trimble conical cover dome. The top figure uses an elevation cutoff angle of 10 degrees, the second figure uses an elevation cutoff angle of 15 degrees, and the bottom figure uses an elevation angle of 20 degrees. In all figures, the green "+" symbols represent L3 height solutions. The red "*" symbols represent L3 height solutions with additional site specific troposphere parameters estimated. On days 170, 177, and 178 nothing was surrounding the antennas. On days 171, 172, 175, and 176 the metal base plate was installed underneath one of the antennas. On days 173 and 174 the Trimble conical shaped dome was attached to the base plate under one of the antennas.

The results in figure 5 use a Trimble Dorne Margolin choke ring antenna and their conical cover with metal mounting plate. To confirm that the problems seen with the choke ring antenna would occur with other antenna types, five days of data were collected with the Trimble L1/L2 GEOD GP antenna. The results of this test are plotted in figure 6. The three panels in this figure represent three different elevation angles used in processing the data. The top panel uses a 10 degree elevation mask, the middle panel uses a 15 degree elevation mask, and the bottom panel uses a 20 degree elevation mask. The green "+" symbols are the L3 height solutions, and the red "*" symbols represent L3 height solutions with additional troposphere parameters estimated at one site. Days 179 and 180 are control surveys. For these two days, only the antennas were mounted on the tripod. Solutions without troposphere parameters estimated agree to within 1 mm of ground truth while solutions with troposphere parameters estimated agree to within 5mm. Days 181, 182, and 183 have the conical dome and metal ground plate attached to one of the antennas.

Both types of height solutions in figure 6 can be seen to disagree with ground truth surveys by at least 20 mm. In particular, the L3 solutions without troposphere estimates (the green "+" symbols) seem to have a 20 mm height bias throughout all elevation cutoff angles. The 20 mm offset occurs in both the L3 solutions (with and without troposphere parameters estimated) and is consistent through all the elevation cutoff angles. This is similar to figure 5. For the days where the mounting plate only is attached to the antenna, the height errors are consistently in the 5 mm range. This implies that the errors are introduced by the metal mounting plate. Comparing figures 5 and 6, the metal mounting plate has a larger effect on solutions using the Trimble L1/L2 GEOD GP antenna than the Dorne Margolin choke ring antenna. In addition to the 20 mm offset caused by the mounting plate, the rest of the errors in figure 6 can be attributed to the cover. The L3 solutions with troposphere estimates vary as a function of elevation cutoff angle with the largest differences (44 mm) when compared to ground truth occurring when a 20 degree elevation cutoff angle is used. The results with troposphere estimates show that the solutions vary as the elevation cutoff mask is changed. In particular, the solutions with a 20 degree elevation cutoff have a height error of 44 mm.

Figure 6: Baseline height solutions using the Trimble GEOD L1/L2 GP antenna and the Trimble conical antenna cover. The top figure uses an elevation cutoff angle of 10 degrees, the second figure uses an elevation cutoff angle of 15 degrees, and the bottom figure uses an elevation angle of 20 degrees. In all figures, the green "+" symbols represent L3 height solutions. The red "*" symbols represent L3 height solutions with additional site specific troposphere parameters estimated. Days 179 and 180 are solutions without the cover on either antenna. Days 181, 182, and 183 are solutions with the Trimble conical antenna cover on one of the antennas.

Ashtech Dorne Margolin choke ring antennas come equipped with a conical dome mounted directly over the antenna. Effects of the dome on the antenna can be seen in figure 7. The top panel shows solutions using an elevation cutoff of 15 degrees, while the second panel uses an elevation cutoff of 20 degrees. The red "+" symbols represent L3 height solutions, while the green "*" symbols represent L3 height solutions with troposphere estimates. Days 041, 042, and 048 through 051 do not have a dome on either antenna. Days 044 and 045 have a dome on one of the two antennas. For days 046 and 047 the dome was removed from the first antenna and moved to the second antenna. Like the Trimble conical antenna cover, the Ashtech cover clearly causes an L3 height solution that differs from the ground truth surveys. This error is as much as 20 mm when estimating troposphere parameters, using an elevation cutoff of 20 degrees. Looking at the two plots, there appears to be a bias between the solutions with troposphere estimates and the solutions without troposphere estimates. This is probably because multipath at the two stations are not identical at the two antennas, and the difference is mismodeled as tropospheric delay. If one looks at only the troposphere solutions, they seem to be centered around a value of 2 or 3 mm for the 15 degree solutions and 6 or 7 mm for the 20 degree solutions.

Figure 7: Baseline height solutions using the Ashtech Dorne Margolin choke ring antenna and the Ashtech conical antenna cover. The top figure uses an elevation cutoff angle of 15 degrees, the second figure uses an elevation cutoff angle of 20 degrees. In both figures, the red "+" symbols represent L3 solutions. The green "*" symbols represent L3 solutions with site specific troposphere parameters estimated. The solutions on days 041, 042, 048, 049, 050 and 051 do not have a cover on either antenna. Solutions on days 044 and 045 have the conical dome on one of the two mark. The dome was switched to the other mark for days 046 and 047.

The one-eighth inch acrylic spherical dome has a much smaller effect than either of the conical covers reported above. The results for this dome are shown in figure 8. In this plot only L3 height solutions without troposphere parameters estimated are plotted. The L3 troposphere solutions are not plotted because the noise introduced by adding additional troposphere parameters masks the results shown in the L3 solutions without troposphere estimates. The red "+" symbols represent solutions using a 10 degree elevation cutoff, the blue "*" symbols use a 15 degree elevation cutoff, the green "o" symbols represent solutions with 20 degree elevation cutoff. Days 170, 171, and 172 were control days where the antennas were mounted without any covers. Days 173 and 174 are solutions where a one-eighth inch spherical acrylic dome was mounted in a non-concentric manner over one of the antennas. Days 175 and 176 have the acrylic dome mounted over the same antenna, but this time so that the nominal phase center is located as near as possible to the center of the dome. Days 177 and 178 the dome was removed from the antenna to check how well the control survey could be repeated. It appears as if the solution on day 173 is an outlier and that the rest of the solutions show that the one-eighth inch acrylic dome causes at most a 2 mm offset in the baseline height solution. In addition, there was no dependence of troposphere estimation on these solutions.

Day 173 appears to be an outlier because the solutions between days 173 and 174, where the dome was mounted non concentrically over one of the antennas, differ by approximately 3mm. This difference in these two solutions is more than three times the difference of all the other solution differences of similar antenna/dome configurations. The reason 173 appears to be the problem, instead of 174, is that the solution on day 174 has the same sign change as days 175 and 176. The differences between days 173 and 174 versus 175 and 176 is that the cover was moved so that it was positioned concentrically over the antenna. It is doubtful that the difference between the concentric and non-concentric setup would be severe enough to cause a complete sign change in the effect on baseline height. The results for day 173 have been checked for a blunder multiple times, and none has been found. However, the consistency in the rest of the solutions seems to suggest that something is different for this day.

Figure 8: Baseline height solutions for the one-eighth inch spherical acrylic cover. The red "+" symbols represent solutions using a 10 degree elevation cutoff, blue "*" symbols use a 15 degree cutoff, green "o" symbols use a 20 degree cutoff. All solutions are without troposphere estimation. The noise in the solutions with additional troposphere parameters estimated was too large to see the small changes associated with this figure. Solutions on days 170, 171 and 172 have no antenna covers on either antenna. Solutions on days 173 and 174 have an acrylic dome mounted in a non-concentric way over one of the antennas. Solutions on days 175 and 176 have the dome mounted so that the dome is concentric with respect to the mean phase center of the antenna. The solutions on days 177 and 178 do not have a cover over either of the antennas.

Tests of the one-quarter inch dome show that it causes a larger height error than the one-eighth inch dome, but not as large as the conical covers. These results are shown in figure 8. The L3 height solutions are plotted with red "+" symbols. The L3 height solutions with troposphere estimates are plotted with green "*" symbols. For the first solution, neither antenna had a dome on the antenna. For the last two days, one of the antennas had a one-quarter inch spherical dome mounted so that the mean phase center was centered with the cover. From this figure it appears as if the thicker one-quarter inch acrylic dome has a larger effect on the height solutions than the one-eighth inch dome. The L3 solutions with troposphere estimates are off by as much as 10 mm while the solutions without troposphere estimates are off by less than 5 mm.

Figure 9: Baseline height solutions for the one-quarter inch spherical acrylic cover. The solutions plotted here all use a 15 degree elevation cutoff for processing. The red "+" marks represent L3 height solutions without additional troposphere estimates. The green "*" marks represent the L3 height solutions with hourly troposphere parameters estimated. For the first solution, neither antenna was covered by a dome. For the second and third solutions, the one-quarter inch dome was covering one of the antennas. This data was taken on the UNAVCO facility roof and not at the TMGO. The ground truth for these marks is not as well known as the marks at TMGO.

The effect of the various domes, and how the effect changes with the elevation cutoff angle used for data processing is summarized in Table 1. From this table, it can be seen that the conical covers have a larger effect on baseline height than the spherical covers. They also have a larger dependence on what elevation cutoff angle is being used for data processing.

[Next] [Previous] [Top]

Last modified Thursday, 17-Nov-2005 03:58:00 UTC

Home | About Us | Contact Us | Support | Search | Facility | PBO | Education & Outreach

Comments: webmasterunavco.org
© 2009 UNAVCO, Inc.