Comparison of tropospheric scintillation prediction models of the Indonesian climate
© Chen and Singh; licensee Springer. 2014
Received: 13 February 2014
Accepted: 12 June 2014
Published: 2 July 2014
Tropospheric scintillation is a phenomenon that will cause signal degradation in satellite communication with low fade margin. Few studies of scintillation have been conducted in tropical regions. To analyze tropospheric scintillation, we obtain data from a satellite link installed at Bandung, Indonesia, at an elevation angle of 64.7° and a frequency of 12.247 GHz from 1999 to 2000. The data are processed and compared with the predictions of several well-known scintillation prediction models. From the analysis, we found that the ITU-R model gives the lowest error rate when predicting the scintillation intensity for fade at 4.68%. However, the model should be further tested using data from higher-frequency bands, such as the K and Ka bands, to verify the accuracy of the model.
Tropospheric scintillation is receiving more attention because of the demand for higher bandwidth due to the congestion on the C and Ku bands. Higher-frequency bands tend to be affected by tropospheric scintillation, which is an event that causes the rapid fluctuation of the magnitude and phase of millimeter radio waves to occur in satellite communication systems. When the signal encounters turbulence in the atmosphere, rapid variations in the refractive index along the path will lead to fluctuations in the signal level received (Mandeep and Hassan 2004; Mandeep et al. 2006; Karasawa et al. 1988a). These fluctuations, called scintillations, are generally constant around the mean signal level. Tropospheric scintillation depends on the season and daily weather conditions. On satellite links, significant scintillation effects are mainly caused by strong turbulence in clouds and usually occur in the summer afternoon (Mandeep and Hassan 2004).
Tropospheric scintillation intensity has been proven to increase with high carrier frequency, low elevation angle, and small receiving antenna. In general, signal fade caused by rain attenuation on communication signals is more significant compared to signal fade caused by tropospheric scintillation. However, considerations regarding tropospheric scintillation when designing the link budget become vital for low-fade-margin systems that operate at high frequency and low elevation angle (Mandeep et al. 2006). A satellite communication system that operates at a high frequency (>10 GHz) and a low elevation angle (<10°) may experience more degradation from scintillation than from rain attenuation. Typically, satellite links above 10 GHz may suffer from tropospheric scintillation fluctuations of up to several decibels peak-to-peak with duration of the scintillation events, of a few seconds.
Scintillation occurs under clear-sky conditions and during rain. However, distinguishing actual scintillation from rapid variations of rain attenuation in the presence of rain is not straightforward to Marzano and d'Auria (1998). In addition, it is of less interest to investigate scintillation under rainy conditions for low-availability satellite system design purposes because rain attenuation is usually much more pronounce than scintillation fades (Marzano et al. 1999; Ortgies 1993; Otung 1996; Otung and Savvaris 2006). For those reasons, only clear-weather scintillation is accounted for in this project (impairments such as rain attenuation and noise are eliminated). Noise was eliminated by visual inspection on all date sequence. Thus, accurate estimates of signal degradation due to this effect must be included in the design of satellite communication systems (Singh and Hassan 2003; Geoffroy et al. 1997; Kamp et al. 1997; Kamp 1998; Van De Kamp et al. 1999; Yu et al. 2006)
Indonesia, like other Southeast Asian countries, is near the equator and has a hot and humid tropical climate and two monsoon seasons, one between October and February and the other from April to October; the first is characterized by thunderstorms. Temperatures and humidity are high throughout the year. This climate varies from those of European countries where the climate is cold and dry most of the year. The Karasawa model (Karasawa et al. 1988b), for example, was developed in Yamaguchi (Japan). Other models, such as the Otung model (Otung 1996) and the Van de Kamp model (Van De Kamp et al. 1999) were developed using a European database. Because a large variation exists between the climate in Europe and Indonesia, an analysis of the tropospheric scintillation prediction models against the measured scintillation intensity for Indonesia (tropical climate) is needed.
The scintillation data are taken from a 12.247-GHz JCSAT3 beacon at an elevation angle of 64.7° at a sampling rate of 1 s. The beacon is located at Bandung (6.9° S, 107.6° E). The data for this analysis were taken from January 1999 to December 2000. The beacon receiver has a sampling rate of 1 Hz which is sufficient to analyze tropospheric scintillation data. Temperature, humidity, wind direction, and speed were placed closed to the antenna used to measure the surface parameter. Rain rate was measured using a rain gauge.
Non-rain events were separated from the rain events for the experimental data. The rain events were determined with the use of a rain gauge. The clear-sky level was indicated by using a spectrum analyzer and a rain gauge whereby rain periods have been removed. The raw data were inspected visually to remove any spurious samples as much as possible resulting from loss of lock due to the satellite propellant saving option and satellite movement. The data were extracted by passing through a fifth-order high-pass Butterworth filter with a 0.04-Hz cutoff frequency based on performing a spectral analysis. After the filtering process, the resulting data consists of positive (enhancement) and negative (fade) scintillation amplitude fluctuations above the mean level. The scintillation intensity is calculated as the standard deviation of the amplitude fluctuations over 1 min (Garcia-del-Pino et al. 2012).
Antenna height above sea level
Antenna configuration/azimuth angle
Tropospheric scintillation prediction models
Specifications of the scintillation models
14- and 11-GHz satellite link, elevation angle 6.5, diameter of 7.6 m, Yamaguchi, Japan, and 1 year of data (1988)
19.8-GHz satellite link, elevation angle of 28.7, diameter 7.6 m, Sparsholt, UK, and 1 year of data (1996)
DPSP, STN2, STH2
18.7-, 39.6-, and 49.5-GHz satellite link, elevation angle of 30.6, diameter of 1.8 m, Milan, Italy, 1 year of data (1998)
12.5-, 20-, and 30-GHz satellite link; diameter of 0.6, 1.8, and 3.7 m; Darmstadt, Germany, and 1 year of data (1993)
Van de Kamp
19.8- and 29.7-GHz satellite link, elevation angle of 12.7, diameter of 1.8 m, Helsinki, Finland, 1 year of data (1998)
ITU-R tropospheric scintillation model
DPSP tropospheric scintillation model
Karasawa tropospheric scintillation model
Ortgies scintillation model
Otung scintillation model
Van De Kamp model
where Whc is the long-term average water content for heavy clouds.
Marzano's STH2 and STN2 models
Results and discussion
Comparison of monthly scintillation intensity
Comparison of the scintillation intensity percentage errors for each month of the average years
Overall means, standard deviations, and RMS values for the measurement site over a 2-year period
The DPSP model gives the highest RMS error rate at 194%, compared to the other models. The error rate and root-mean-square error are shown in Tables 3 and 4. Comparing Figures 2 and 3, we can clearly see that models that predict the variance of the signal log-amplitude are more accurate in predicting the scintillation intensity for the climate in Bandung. This is because scintillation in the climate in Bandung follows a gamma distribution as shown in Figure 1, and the prediction models of category 1 are developed assuming that scintillation follows a gamma distribution. The prediction models in category 2 assume that the scintillation follows a lognormal distribution. This assumption is only true for the climates in Europe and not for the global climate.
Comparison of mean scintillation amplitude
Predictions of the monthly scintillation amplitude are calculated for the average measurement years at 0.01% of the time. We introduced a correlation in the analysis in the form of a statistical parameter describing the degree of the relationship between two variables. We compared the correlation output for each of the prediction models with the measured tropospheric scintillation data. For positive values, there is a tendency for an interrelation to exist between either data or a positive relationship with the measured data and vice versa for negative values.
Mean RMS error and correlation for fade scintillation
From Figure 4, the Otung model tends to overestimate the measured data for a lower percentage of time due to high Nwet values. The poor comparison at a lower percentage of time which could be due to the effect of rain present at a small percentage of time has been excluded. Most of these models are based not only on clear-air signal fluctuation.
Mean RMS error and correlation for enhancement scintillation
Mandeep et al. (2011) conducted measurement at USM during the El Niño year with lengthened dry season. The measurement site is at 57 m above sea level compared to ITB which is at 700 m above sea level and during the La Niña year. The antenna elevation angles at USM and ITB are at 40.10° and 64.70° with antenna diameters of 2.4 and 1.8 m, respectively, which highlights that the scintillation amplitude decreases as the elevation angle increases and increases as the antenna diameter increases. For example, at 0.1% of the time, the fade scintillation amplitude is 1.3 dB at USM and 0.5 dB at ITB. The average temperatures at USM and ITB were around 27°C and 23°C, respectively, during the measured years. This indicates that the scintillation intensity reduces as the temperature reduces. Overall, the findings show that the elevation angle, altitude, and temperature do affect the intensity of tropospheric scintillation.
The error rates in the other scintillation prediction models tested in this experiment were very high and not suitable for predicting the scintillation intensity. From Figure 8, we can see that European countries have very low Nwet values throughout the year. Countries like the UK, Italy, and Spain have average Nwet values of less than 40 ppm. Models like DPSP, STH2, STN2, Ortgies-T, and Ortgies-N have difficulty predicting the scintillation for Bandung where the average Nwet is approximately 110 ppm. The regression of these models is conducted using limited data, and the meteorology parameter is not tested against the data from other climate regions. Therefore, these models cannot be applied globally, especially to tropical countries like Bandung.
A review and evaluation of currently existing tropospheric scintillation prediction models have been presented in this paper. The prediction models are compared with the data obtained from the 12.247-GHz JCSAT3 beacon slant path at a satellite earth station, which is located at Bandung. The measurement from the satellite earth station confirmed that the ITU-R model gives the best scintillation intensity predictions for countries that have tropical climates.
The authors would like to acknowledge the Universiti Kebangsaan Malaysia and the Association of Radio Industries and Businesses (ARIB) of Japan for providing the data in the form of a CD.
- Geoffroy P, Marzano FS, D'auria G, Riva C, Vanhoenacker-Janvier D: Evaluation of statistical models for clear air scintillation prediction using Olympus satellite measurements. Int J Satellite Comm 1997, 15: 73–88. 10.1002/(SICI)1099-1247(199703)15:2<73::AID-SAT570>3.0.CO;2-NView ArticleGoogle Scholar
- International Telecommunication Union: ITU-R P.618–9. Propagation data and prediction methods required for the design of Earth-space telecommunication systems. ITU, Geneva; 2007.Google Scholar
- Kamp MMJL: Asymmetric signal level distribution due to tropospheric scintillation. Elec Let 1998, 34(11):1145–1146. 10.1049/el:19980805View ArticleGoogle Scholar
- Kamp MMJL, Tervonen JK, Salonen ET: Tropospheric scintillation measurements and modelling in Finland. Paper presented at the tenth international conference on antennas and propagation. Edinburgh; 1997. 14–17 Apr 1997 14–17 Apr 1997Google Scholar
- Karasawa Y, Yamada M, Allnutt JE: A new prediction method for tropospheric scintillation on Earth-space paths. IEEE Trans Ant Prop 1988, 36(11):1608–1614. 10.1109/8.9712View ArticleGoogle Scholar
- Karasawa Y, Yasukawa K, Yamada M: Tropospheric scintillation in the 14/11-GHz bands on Earth-space paths with low elevation angles. IEEE Trans Ant Prop 1988, 36(4):563–569. 10.1109/8.1146View ArticleGoogle Scholar
- Mandeep JS: Extracting of tropospheric scintillation propagation data from Ku-band satellite beacon. Int J Phy Sci 2011, 6(11):2649–2653.Google Scholar
- Mandeep JS, Hassan SIS: Comparison of 1-minute rainfall rate distribution for tropical and equatorial climates. Space Commun 2004, 19: 193–198.Google Scholar
- Mandeep SJS, Syed ISH, Kiyoshi I, Kenji T, Mitsuyoshi I: Analysis of tropospheric scintillation intensity on earth to space in Malaysia. Amer J App Sci 2006, 3(9):2029–2032.View ArticleGoogle Scholar
- Mandeep JS, Hassan SIS, Ain MF, Igarashi K: Tropospheric scintillation measurement in Malaysia at Ku-band. J Electr Wav App 2008, 22(8–9):1063–1070.View ArticleGoogle Scholar
- Mandeep JS, Anthony CCY, Abdullah M, Tariqul M: Comparison and analysis of tropospheric scintillation models for Northern Malaysia. Acta Astro 2011, 69(1–2):2–5.View ArticleGoogle Scholar
- Marzano FS, d'Auria G: Model-based prediction of amplitude scintillation variance due to clear-air tropospheric turbulence on Earth-satellite microwave links. IEEE Trans Ant Prop 1998, 46(10):1506–1518. 10.1109/8.725283View ArticleGoogle Scholar
- Marzano FS, Carlo R, Alessio B, Fabio C: Assessment of model-based scintillation variance prediction on long-term basis using Italsat satellite measurements. Int J Satellite Comm 1999, 17(1):17–36. 10.1002/(SICI)1099-1247(199901/02)17:1<17::AID-SAT615>3.0.CO;2-9View ArticleGoogle Scholar
- Ortgies G: Frequency dependence of slant-path amplitude scintillations. Elec Let 1993, 29(25):2219–2220. 10.1049/el:19931490View ArticleGoogle Scholar
- Otung IE: Prediction of tropospheric amplitude scintillation on a satellite link. IEEE Trans Ant Prop 1996, 44(12):1600–1608. 10.1109/8.546246View ArticleGoogle Scholar
- Otung IE, Savvaris A: Estimating tropospheric scintillation intensity on earth-space propagation paths. Elec Let 2006, 42(7):381–382. 10.1049/el:20064281View ArticleGoogle Scholar
- Garcia-del-Pino P, Riera JM, Benarroch A: Tropospheric scintillation with concurrent rain attenuation at 50 GHz in Madrid. IEEE Trans Ant Prop 2012, 60(3):1578–1583.View ArticleGoogle Scholar
- Singh MSJ, Hassan SIS: Probability density functions of tropospheric amplitude scintillation on a satellite link. Paper presented at the 4th national conference on telecommunication technology. Shah Alam; 2003. 14–15 Jan 2003 14–15 Jan 2003Google Scholar
- Van De Kamp MMJL, Tervonen JK, Salonen ET, Poirares Baptista JPV: Improved models for long-term prediction of tropospheric scintillation on slant paths. IEEE Trans Ant Prop 1999, 47(2):249–260. 10.1109/8.761064View ArticleGoogle Scholar
- Yu P, Glover I, Watson PA, Davies OT, Ventouras S, Wrench C: Review and comparison of tropospheric scintillation prediction models for satellite communications. Int J Satellite Comm 2006, 24(4):283–302. 10.1002/sat.842View ArticleGoogle Scholar
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited.