The foliage effect on the height time series from permanent GPS stations
© The Society of Geomagnetism and Earth, Planetary and Space Sciences, The Seismological Society of Japan; The Volcanological Society of Japan; The Geodetic Society of Japan; The Japanese Society for Planetary Sciences; TERRAPUB. 2010
Received: 20 July 2008
Accepted: 25 October 2010
Published: 26 January 2011
The effect of deciduous trees growing above antenna height on data collected by permanent Global Positioning System (GPS) stations was investigated. Signal blockage due to foliage and branches was found to have the same effect as an increased elevation cutoff angle, i.e., there was a change in the computed position. Height estimates were affected the most, showing a decrease with tree growth. Empirical Orthogonal Function (EOF) analysis on the height-time series from five test sites and two stations surrounded by trees showed a similar EOF mode of signal. Signal availability, computed as the ratio of the complete to possible set of observations, decreased with increasing tree growth and showed seasonal variation, with the observation ratios being higher during the winter months when the leaves had fallen. A similar seasonal variation was observed in multipath error and signal attenuation due to foliage. The multipath error index MP2 was computed using the TEQC program and found to increase at a significant rate at sites with growing trees. Signal attenuation was analyzed using 1-σ uncertainties from the estimation process of daily GPS data processing. While 1-σ uncertainties did not show any seasonal variations at sites without trees, they were highly dependent on conditions related to the seasonal change of foliage when deciduous trees were near the antenna.
Key wordsGP Smultipath foliage effect signal attenuation
About 90 permanent Global Positioning System (GPS) stations were operational in South Korea as of July, 2008. These sites are used for various GPS applications, such as real-time navigation, precise surveying, geodesy, and Earth science studies. Since the average operation period of those stations is about 8 years, the site velocities derived from the continuous GPS measurements are an indispensible geodetic tool for crustal deformation studies of the Korean peninsula and the surrounding area (Hamdy et al., 2005; Jin and Park, 2006). For example, Jin and Park (2006) processed 4 years of data and computed the horizontal velocities of 45 GPS sites in South Korea and found that the GPS-derived seismic moment accumulation rate is significant and consistent with recent historic earthquakes and fault zones in South Korea. However, truly three-dimensional (3-D) tectonic studies of the Korean peninsula deformation based on GPS measurements have not yet been conducted.
As a preliminary step for the vertical displacement analysis of the Korean peninsula, we selected 53 GPS stations in Korea and processed 7 years of continuous data. Based on this analysis, we determined that the average uplift rate of the 53 sites is 2 mm/year (Park, 2007). However, two of the sites, INCH and PAJU, located in Incheon and Paju, respectively, in the northwestern part of South Korea show a relatively low vertical velocity, with a rate of almost zero, which is 2–3 mm/year lower than that of nearby sites. As the inter-site distances are at the most several tens of kilometers, the abnormal rates of vertical velocity at INCH and PAJU must be due to local effects. A visit to the two sites to check the site-specific environments revealed the presence of trees growing nearby that were 2–6 m higher than the antennas. This led to the suspicion that the trees growing above the GPS antenna may be the reason for the apparently low uplift rates.
The availability and attenuation of GPS signals due to foliage depends on various parameters, such as tree height, type of tree, thickness of leaves, density of trees, distance from the antenna to the tree, canopy shape, and tree base (Lachapelle et al., 1994; Spilker, 1996). Lachapelle et al. (1994) investigated the seasonal effect of deciduous tree foliage on GPS signal availability and accuracy for vehicular navigation and found that the range availability was superior on the test date prior to foliage growth. Spilker (1996) characterized foliage attenuation as attenuation in dB/m of foliage penetration for stationary and mobile users. Koh and Sarabandi (2002) investigated the attenuation, depolarization, and fluctuation of a microwave signal going through a tree canopy, developing a Monte Carlo coherent scattering model. This model is considered to be particularly effective for analyzing the performance of GPS receivers under tree canopies.
The locations and GPS equipment of the five GPS stations analyzed in Fig. 1.
2. The GPS Height-Time Series Analysis
Among the 53 permanent GPS sites analyzed by Park (2007), two sites, namely, ICNW and PAJU, both near the Seoul metropolitan area, showed abnormally low rates of crustal uplift. This was especially evident for the INCH station, which showed the lowest vertical velocity of all the 53 sites. The initial suspicion was that INCH may actually be subsiding because part of the Incheon area is reclaimed land. Therefore, in our study, we included a second site (ICNW) in the Incheon area as a control to check if subsidence is actually occurring in this region. The ICNW station, which is within 10 km of INCH, is the first permanent GPS site in Korea to be installed right at the tide gauge site so as to be able to monitor the crustal deformation of the tide gauge station. Two additional sites, SUWN and SEOS, were also included in this analysis. The site locations of the GPS stations and the types of GPS receiver and antenna at each station are listed in Table 1.
Of the five sites included in our regional analysis, four stations (excepting ICNW) have been in operation since at least May 2000; ICWN was installed in September 2005. Thus, for an objective comparison, we only processed data from September 2005 through to January 2008. GIPSY-OASIS software (GPS Inferred Positioning System-Orbit Analysis and Simulation Software), developed by the Jet Propulsion Laboratory (JPL), was used to process the GPS data. In particular, the precise point positioning strategy (Zumberge et al., 1997) was used with earth orientation parameters and GPS satellite orbit and clock solutions provided by JPL. These JPL products are non-fiducial ones (Heflin et al., 1992), and the orbit and clock information are sampled at 300-s intervals. After data processing, we used transformation files (referred to as x-files) to align the site position to the IGS05 frame.
Antenna phase center variations, ocean tidal loading displacements, and atmospheric loading effects were corrected as standard procedures of high-precision GIPSY-OASIS data processing. Relative calibration tables published by the National Geodetic Survey of the USA were used in order to consider antenna phase center variations in terms of elevation angles (Mader, 1999). The ocean tidal loading displacements were also corrected for on an epoch-by-epoch basis by utilizing 11 major constituents from NAO.99b model (Matsumoto et al., 2000). The atmospheric loading effects at 6-h intervals were computed using the APLO program (Petrov and Boy, 2004), and the daily average displacement was derived from these atmospheric loading effects. The daily GPS height-time series was then corrected using the daily atmospheric loading values.
For a more quantitative analysis, we performed an Empirical Orthogonal Function (EOF) analysis on the times series of Fig. 1. EOF analysis is a powerful tool that is used to assess the degree of correlation among times series of the same physical phenomenon recorded at different places (Elosegui et al., 1997). It has been successfully used in the areas of GPS geodesy, such as GPS coordinate time series and precipitable water vapor studies (Davis and Elgered, 1998; Aoki and Scholz, 2003). Because EOF analysis does not allow for data gaps, the missing points in Fig. 1 were interpolated by cubic splines (Aoki and Scholz, 2003).
The next step involved visiting the five sites and checking the site environment for any possible problem in monument stability. The GPS antennas at INCH, PAJU, and SEOS are installed at the top of 4-m-high stainless pillars, but both the monument and mount of these stations were found to be stable. SUWN and ICNW also have sturdy mounts and monument structures. The sky visibility at ICNW and SEOS is excellent, with signal reception above the 5° elevation angle secured in all directions. SUWN, an International GNSS Service (IGS) site, is situated just east (50 m) of a small forest, but the treetops are not much higher than the antenna. Therefore, we concluded that these three sites (SUWN, ICNW, and SEOS) meet IGS Site Guidelines (Dow et al., 2005); in contrast, a number of problems were evident at sites INCH and PAJU.
3. Signal Availability due to Foliage
The effect of deciduous tree foliage on GPS can be classified in two categories: signal availability and signal attenuation. Signal availability is a function of the GPS signal not penetrating thick foliage and, consequently, the GPS antenna being unable to receive the signal. Signal attenuation occurs when (1) the GPS signal goes through thin foliage and reaches the antenna; (2) the signal reflects or bounces off the foliage. The latter case of signal attenuation can be considered as multipath error. In this section, we introduce signal availability analysis. Signal attenuation, due to foliage, is investigated in the next section.
The second-degree polynomial function, as a function of elevation cutoff angle, was fitted in weighted least-squares to the coordinate differences; this is given as a solid line in Fig. 6. In the case of the vertical coordinate, the fitted function was In this equation, ΔU is the height difference in centimeters, and εco is the elevation cutoff angle in degrees. This function was used to determine the height changes at 5° intervals. The calculated changes were −0.8 cm from 5° to 10°, −1.0 cm from 10° to 15°, −1.5 cm from 15° to 20°, −2.0 cm from 20° to 25°, and −2.5 cm from 25° to 30°. This result led us to conclude that the decreasing rate in the estimated heights grows rather rapidly with increasing elevation cutoff angle. If we assume that the foliage is thick enough to block the GPS signals, its effect will be similar to that of increased elevation cutoff angles, namely, the estimated height will be lower.
The observation ratio was found to have a clear seasonal variation, with a value of approximately 90% in the winter season, becoming smaller during the summer. In addition to the seasonal fluctuations, the observation ratio during the summer season continuously decreases over time, from approximately 86% in 2006 to approximately 84% in 2007, suggesting a relationship between decreases in the observation ratio and annual tree growth and the corresponding increase in signal blockage. Figure 7 also clearly shows that the minimum ratio occurs during the months of July and August, which coincide with the period when the leaf area index in Korea is the highest (Kim et al., 2005). The observation ratio also shows a decreasing trend over time during the winter season when all the leaves of deciduous trees fall, with monthly averages for January of 89.2% in 2006, 88.4% in 2007, and 87.8% in 2008. This continuous, gradual reduction is believed to be caused by the growing stems and branches.
4. Signal Attenuation due to Foliage
This section deals with multipath error and signal attenuation due to foliage and its seasonal variation. To analyze multipath error, we used multipath indices computed by TEQC. The TEQC multipath indices MP1 and MP2 reflect the multipath error on pseudorange observables on L1 and L2, respectively (Estey and Meertens, 1999). Among the currently available GPS carriers, the signal strength of L2 is weaker than L1 and, therefore, L2 signals are more susceptible to signal attenuation and to multipath. This makes MP2 a better indicator of the foliage effect; as such, we utilized MP2 instead of MP1 in this study. Earlier studies indicated that MP1 and MP2 values are dependent not only on the site environment, but also on the GPS receiver type and firmware version (Park et al., 1998). Thus, for an objective evaluation of the site-specific multipath effect, MP2 values from the same kind of receiver with the same firmware version have to be compared. Of the five stations analyzed in this study, SEOS, INCH, and PAJU have the same kind of receiver (Trimble 4000SSi) with exactly the same firmware version. ICNW and SUWN were excluded from this analysis because both stations are equipped with a unique receiver: an Ashtech iCGRS receiver at ICNW and a Trimble NetRS at SUWN. The MP2 values can be calculated at selected measurement intervals, but to simplify the analysis we took daily averages from MP2 values sampled at 30-s intervals. The daily average MP2 values at each site were computed for more than 4 years to detect seasonal trends, if any.
INCH and PAJU exhibit very different characteristics from SEOS due to growing trees (Fig. 8). Like SEOS, the two sites also have clear annual signals, but the magnitudes are higher: 8.2 cm for INCH and 5.0 cm for PAJU. Another feature that separates the two sites from SOES is that the MP2 values show significant increasing rates from 2004 onwards. This increase in MP2 values can only be caused by the growing trees: as the trees grow, not only the height of the treetop grows, but the number of leaves and the foliage thickness increase. The MP2 growth ratio for INCH and PAJU are 9.5 and 3.4 cm/year, respectively. The growth rate is higher for INCH than PAJU because INCH is almost completely surrounded by trees, while clear sky is visible in some directions at PAJU. The distances to the tree trunks are shorter at INCH than PAJU.
The height uncertainties are consistent at DOND, staying in the range of 5–5.5 mm regardless of season. However, at INCH and PAJU, there are clear signs of an annual signal, with higher uncertainties in the summer season and lower ones in the winter. Notably, the uncertainties at INCH and PAJU are at the same level during the winter months, even though they are slightly lower at PAJU, while during the summer months, 1-σ values of INCH are definitely higher. The larger difference between INCH and PAJU during the summer can be explained by the presence of thicker foliage at INCH. The slightly higher uncertainties at INCH during the winter season are believed to be caused by more stems and branches at the site.
IGS Site Guidelines state that the site location should have a clear horizon with minimal obstruction at elevations >5° (Dow et al., 2005). Thus, buildings or structures near the GPS antenna that can block the signal should be avoided. Growing trees are even worse sources of error, not only because of their growth, but also due to seasonal changes of foliage. In our study, we have demonstrated the blockage of GPS signals due to foliage by analyzing seasonal changes in the ratio of complete to possible observations for permanent GPS stations. As expected, the observation ratio was higher in the winter season, when all of the leaves of deciduous trees have fallen. With the growth of trees grow above the antenna at one of the test sites, in 2007 the observation ratio of the summer season was found to drop from approximately 86% to in 2006 to approximately 84%. This signal blockage due to trees has the same effect as an increased elevation cutoff angle, decreasing the height estimate. The multipath error index MP2 was computed for more than 4 years for three permanent GPS sites with different foliage conditions. Seasonal signals in MP2 were observed at all sites, but those sites with thick foliage or higher tree density were found to have higher amplitudes and an increasing trend in annual variation. Signal attenuation due to foliage and the resulting degradation of accuracy were investigated with the coordinate uncertainties in the parameter estimation process. The 1-σ values in the height estimates also showed annual variation; their amplitudes were higher for the sites with thick foliage.
This work was supported by INHA UNIVERSITY Research Grant (INHA-36093). We thank Ministry of Land, Transportation, and Maritime Affairs for generously making their GPS data available to the authors.
- Aoki, Y. and C. H. Scholz, Vertical deformation of the Japanese islands, 1996–1999, J. Geophys. Res., 108 (B5), 2257, doi:https://doi.org/10.1029/2002JB002129, 2003.View ArticleGoogle Scholar
- Davis, J. L. and G.. Elgered, The spatio-temporal structure of GPS water-vapor determinations, Phys. Chem. Earth, 23, 91–96, 1998.View ArticleGoogle Scholar
- Dong, D., P. Fang, Y. Bock, F. Webb, L. Prawirodirdjo, S. Kedar, and P. Jamason, Spatiotemporal filtering using principal component analysis and Karhunen-Loeve expansion approaches for regional GPS network analysis, J. Geophys. Res., 111, B03405, doi:https://doi.org/10.1029/2005JB003806, 2006.Google Scholar
- Dow, J. M., R. E. Neilan, and G. Gendt, The International GPS Service (IGS): Celebrating the 10th anniversary and looking to the next decade, Adv. Space Res., 36 (3), 320–326, 2005.View ArticleGoogle Scholar
- Elosegui, P., A. Rius, J. L. Davis, G. Ruffini, and S. Keihm, A regional GPS experiment for estimating the spatial and temporal variations of water vapor, in Proceedings of the ION GPS-97 10th International Technical Meeting of the Satellite Division of The Institute of Navigation, 241–248, Kansas City, Missouri, 1997.Google Scholar
- Estey, L. H. and C. M. Meertens, TEQC: The multi-purpose toolkit for GPS/GLONASS data, GPS Sol., 3 (1), 42–49, 1999.View ArticleGoogle Scholar
- Hamdy, A. M., P.-H. Park, and H. C. Lim, Horizontal deformation in South Korea from permanent GPS network data, 2000–2003, Earth Planets Space, 57, 77–82, 2005.View ArticleGoogle Scholar
- Heflin, M. B., W. I. Bertiger, G. Blewitt, A. P. Freedman, K. J. Hurst, S. M. Lichten, U. J. Lindqwister, Y. Vigue, F. H. Webb, T. P. Yunck, and J. F. Zumberge, Global geodesy using GPS without fiducial sites, Geophys. Res. Lett., 19 (2), 131–134, 1992.View ArticleGoogle Scholar
- Jin, S. and P.-H. Park, Strain accumulation in South Korea inferred from GPS measurements, Earth Planets Space, 58, 529–534, 2006.View ArticleGoogle Scholar
- Kim, S. H., J. H. Park, C. S. Woo, and K. S. Lee, Analysis of temporal variability of MODIS Leaf Area Index (LAI) product over temperature forest in Korea, IGARSS, 4343–4346, 2005.Google Scholar
- Koh, I.-S. and K. Sarabandi, Polarimetric channel characterization of foliage for performance assessment of GPS receivers under tree canopies, IEEE Trans. Antennas and Propagation, 50 (5), 713–726, 2002.View ArticleGoogle Scholar
- Lachapelle, G., J. Henriksen, and T. Melgard, Seasonal effect of tree foliage on GPS signal availability and accuracy for vehicular navigation, in Proceedings of the ION GPS-94 7th International Technical Meeting of the Satellite Division of The Institute of Navigation, Salt Lake City, 1994.Google Scholar
- Larson, K. M. and J. Levine, Carrier-phase time transfer, IEEE Trans. Ultrasonics, Ferroelectrics, and Frequency Control, 46 (4), 1001–1012, 1999.View ArticleGoogle Scholar
- Mader, G. L., GPS antenna calibration at the National Geodetic Survey, GPS Sol., 3 (1), 50–58, 1999.View ArticleGoogle Scholar
- Matsumoto, K., T. Sato, T. Takanezawa, and M. Ooe, Ocean tide models developed by assimilating TOPEX/POSEIDON altimeter data into hydrodynamical model: A global model and a regional model around Japan, J. Oceanogr., 56, 567–581, 2000.View ArticleGoogle Scholar
- Park, K.-D., Vertical deformation rates of the Korean peninsula derived from seven years of data at permanent GPS sites, in Proceedings of the 14th GNSS Workshop, Jeju, Korea, 2007 (in Korean).Google Scholar
- Park, K.-D., R. S. Nerem, and M. S. Schnewerk, Site-specific multipath characteristics of IGS and CORS GPS stations, Eos Trans., Suppl., 79 (40), F183, 1998.Google Scholar
- Petrov, L. and J.-P. Boy, Study of the atmospheric pressure loading signal in very long baseline Interferometry observations, J. Geophys. Res., 109, B03405, doi:https://doi.org/10.1029/2003JB002500, 2004.Google Scholar
- Spilker, J. J. Jr., Foliage attenuation for land mobile users, in Global Positioning System: Theory and Applications, Vol. 1, edited by Parkinson, B. W. and Spilker, J. J. Jr., 569–583, American Institute of Aeronautics and Astronautics, Inc., Washington, D.C., 1996.View ArticleGoogle Scholar
- Wdowinski, S., Y. Bock, J. Zhang, P. Fang, and J. Genrich, Southern California permanent GPS geodetic array: Spatial filtering of daily positions for estimating coseismic and postseismic displacements induced by the 1992 Landers earthquake, J. Geophys. Res., 102 (B8), 18057–18070, 1997.View ArticleGoogle Scholar
- Zumberge, J. F., M. B. Heflin, D. C. Jefferson, M. M. Watkins, and F. H. Webb, Precise point positioning for the efficient and robust analysis of GPS data from large networks, J. Geophys. Res., 102 (B3), 5005–5017, 1997.View ArticleGoogle Scholar