Acta Geod. Geoph. Hung., Vol. 47(1), pp. 1–11 (2012) DOI: 10.1556/AGeod.47.2012.1.8 DEVELOPMENT OF A SITE-SPECIFIC ZHD MODEL USING RADIOSONDE DATA D Singh, J K Ghosh, D Kashyap Department of Civil Engineering, Indian Institute of Technology Roorkee, 247667, India, e-mail: [email protected] [Manuscript received June 8, 2011; accepted January 4, 2012] Estimation of precipitable water vapor (PWV) in the atmosphere using ground based GPS (Global Positioning System) data requires an appropriate model for computation of zenith hydrostatic delay (ZHD). Presented herein is a site-specific ZHD model (SSM) for a station at New Delhi, India. The model has been developed by regressing one-year atmospheric vertical profile data collected through radiosonde. The model based on surface atmospheric pressure at the station, has been validated invoking data of three more years. The ZHD values estimated through the model disagree at the 0.3 mm level with ZHD values obtained from raytracing of radiosonde data. Further, Saastamoinen ZHD model provides an error about 0.23 mm rms while about 0.19 mm by the developed model (SSM). Thus, developed SSM can be used for precise estimation of PWV. Keywords: GPS; PWV; site specific; zenith hydrostatic delay 1. Introduction Water vapor plays an important role in several atmospheric and geophysical processes such as transfer of energy, formation of clouds, weather system etc. Therefore, an accurate estimation of water vapor with high spatial and temporal resolution is required for operational weather forecasting and climate research. Various instruments like radiosonde, ground based remote sensors, radiometers are long been used for estimation of water vapor. These operational meteorological measurements do not have adequate resolutions. Thus, restricts accuracy in short-range precipitation forecast. However, ground based GPS receiver provides continuous and near real time estimates of precipitable water vapor (PWV), ensuring all weather coverage and cost effectiveness. This requires model based computations of zenith hydrostatic delay (ZHD). However, the ZHD model is a major source of error. Estimation of GPS PWV with a sub-millimeter accuracy thus requires proper modeling of ZHD (Bosser et al. 2007). At present, several global and regional ZHD models are available. Traditionally, ZHD is estimated by invoking Saastamoinen (1972) global ZHD model. But, most of the existing ZHD models including Saastamoinen have been derived using the available radiosonde data from European and North American continent (Satirapod and Chalermwattanachai 2005). However, ZHD model c 1217-8977/$ 20.00 2012 Akadémiai Kiadó, Budapest D SINGH et al. 2 Table I. ZHD models and their input parameters Models and year Saastamoinen (1972) Hopfield (1969) Black (1978) Baby (Baby et al. 1988) Davis (Davis et al. 1985) MOPS Hydrostatic delay model (MOPS 1998) Improved mean gravity model (Bosser et al. 2007) Unified model (Raju et al. 2007a) Site Specific model (SSM) (Raju et al. 2007a) Nature Parameters global global global global global global global regional site specific ps , Φ, h Ts , ps , β ps , Ts ps , Φ, h ps, Φ, h Φ, day of year Φ, hsfc , t ps ps ps – atmospheric surface pressure (mbar) Φ – latitude h – ellipsoidal height Ts – surface temperature (K) β – temperature lapse rate (◦ C/km) hsfc – surface height (orthometric height (m)) t – time (month) parameters are region specific due to strong spatial heterogeneity and temporal variability of the atmospheric constituents. Therefore, a model developed for one region may not be applicable to another region. In this paper, a site-specific ZHD model has been proposed based on radiosonde data of New Delhi. 2. Literature review Any error in ZHD affects computation of zenith wet delay (ZWD) and hence GPS based PWV. Thus, estimation of GPS based PWV with sub millimeter accuracy requires sub millimeter accuracy in ZHD models. Prominent ZHD models developed so far are depicted in Table I along with their input requirements. Most of the ZHD models are global in nature. However, Hopfield and Saastamoinen models are most widely used and provide reasonably accurate ZHD. These models treat air as an ideal gas, which is not too critical. However, the assumptions of a unique temperature lapse rate and height-independent gravity are not very realistic. Saastamoinen (1972) estimated the accuracy of the hydrostatic components with 2–3 mm RMS. Mendes and Langley (1998) found that ZHD could be estimated with sub millimeter accuracy from the Saastamoinen model if accurate measurement of surface pressure is available. The limitation of the Saastamoinen ZHD model is that it requires ellipsoidal height. Further, the temporal variability of surface pressure and temperature have not been considered in the Saastamoinen ZHD model (Bosser et al. 2007). Since, the variation of mean gravity is nearly sinusoidal, accuracy of Saastamoinen gravity consideration is 0.001 m/s2 which leads to an error of 0.2–0.4 mm (Bosser et al. 2007). Other sources of errors in ZHD models are error in refractive constant and dry gas constant (Bosser et al. 2007). Elgered et al. (1991) reported that the Acta Geod. Geoph. Hung. 47, 2012 SITE-SPECIFIC ZHD MODEL 3 (rms) error in the refractivity constant contributes to 2.4 mm error in ZHD computation. Davis et al. (1985) improved the Saastamoinen model by incorporating Thayer (1974) refractivity constant. Mendes (1999) concluded that ZHD can be predicted from surface measurement of pressure having an error below 0.5 mm and concluded that among Hopfield, Saastamoinen, Baby et al. (1988), Davis et al. (1985), models performance of the Saastamoinen model is far better than the other hydrostatic models. He further observed that the accuracy of the Saastamoinen ZHD model is at sub-millimeter level whereas the other models agreed at the millimeter level. In another study, Janes et al. (1991) have reported an accuracy at the 2–3 mm level for ZHD predictions from empirical models (Saastamoinen and Hopfield) when compared to ray tracing through standard atmosphere and Radio soundings. The precision of the ZHD model also depends on the quality of meteorological sensor. Hauser (1989) has estimated that gravity waves could induce 1.7 hpa (mbar) at ground i.e. 4 mm error in ZHD. Turbulence in the boundary layer is responsible for a smaller (∼0.1 mbar) deviation (Bock and Doerflinger 2001). It has has shown that a ZHD error of 1 mm leads to a PWV estimation accuracy of 0.15 mm, and 0.40 mbar pressure error causes ∼1 mm ZHD error (Bai and Feng 2003). Janes et al. (1991) has mentioned that sensitivity of ZHD to surface pressure is 2 mm/hpa. Moreover, the sensitivity of zenith delay to temperature and relative humidity is about 5–20 mm/◦ C and 1–3 mm/% respectively. Therefore, a mm level accuracy in ZHD thus achievable with precise meteorological sensor (Bock and Doerflinger 2001). A study conducted over the Indian subcontinent using the upper air data for three years (1995–1997) has shown that a unified ZHD model (regional model) is inferior to the site-specific models (SSMs). The mean of absolute difference (MAD) obtained from a unified ZHD model is about ∼0.96 cm, 0.90 cm from Hopfield model and 1.57 cm from Saastamoinen model. The site-specific ZHD models based on atmospheric surface pressure have shown a mean absolute difference from 0.17 cm to 1.80 cm over the Indian subcontinent. This performance is comparable with that of the Hopfield model (Raju et al. 2007a). The mean absolute difference of the site-specific ZHD model developed by Raju et al. (2007a) for New Delhi is 0.27 cm with a standard deviation of 0.2 cm when compared with ZHD from radiosonde data. The Radiosonde data comprised of vertical profiles of p, T up to 25 km and above (Raju et al. 2007b). Thus, site-specific ZHD model provides better precision than unified ZHD model over the Indian subcontinent. However, radiosondes can generally ascend up to about 20 km and atmosphere above this height may provide a significant contribution to ZHD (Vedel et al. 2001). Such delay arising above the known atmospheric profile is termed as upper tropospheric correction (Mekik 1997). Thus, the objective of this research work is to develop a site-specific ZHD model incorporating the upper air correction and treating the gravity as a function of height. Acta Geod. Geoph. Hung. 47, 2012 D SINGH et al. 4 3. Background theory of zenith hydrostatic delay As an electromagnetic wave propagates in the atmosphere, it gets continuously refracted due to the varying index of refraction between different layers of the air starting from the top of the atmosphere up to the ground. Since, the troposphere is non-dispersive to GPS signals, its refractive index (n) is the ratio of the speed of light in vacuum to the phase velocity of the signal. As n is just slightly larger than one, a more convenient quantity (namely refractivity N ) is defined as N = 106 (n−1 ). The delay experienced by a signal in traveling through the neutral (non-ionized) part of the atmosphere is generally referred to as tropospheric delay. This is about 80% of the total delay. The tropospheric path delay, attributed mainly due to retardation, is described by the following equation (Bock and Doerflinger 2001) −6 N (s)ds , (1) ∆L = [n(s) − 1]ds = 10 where, ds is the incremental path length of the signal. Refractivity of air is usually described by empirical equations related to thermo dynamical state variables of the air viz. temperature, pressure, and water vapor pressure. Smith and Weintraub (1953) suggested the following relationship e p , (2) + 3.73 · 105 N = 77.6 T T2 where p is the total atmospheric pressure (in mbar), T is the atmospheric temperature (in degree Kelvin), and e is the partial pressure of water vapor (in mbar). This expression is assumed to hold up to about 0.5% under normal atmospheric conditions. In most contexts, the first term in Eq. (2) is considerably larger than the second. Thayer (1974) provides a more accurate formula for refractivity as follows p e e d −1 N = K1 Zw Zd−1 + K2 + K3 , (3) T T T2 where Ki are constants that are empirically determined in laboratory, pd is the partial pressure due to dry gases (mbar), e is the partial water vapor pressure −1 are the compressibility (mbar), T is the temperature of air (K), Zd−1 and Zw factor for dry air and water vapor respectively. The compressibility factors, which are corrections for non-ideal gas behavior, have nearly constant values that differ from unity by a few parts per thousand (Mendes et al. 2000). The uncertainties in the constants of Eq. (3) limit the accuracy with which the refractivity can be computed, to about 0.02% (Davis et al. 1985). Equation (3) can alternatively be written as follows e e −1 Zw + K3 , (4) N = K1 Rd + K2 T T2 where K2 = K2 − K1 Acta Geod. Geoph. Hung. 47, 2012 Rd Rw . (5) SITE-SPECIFIC ZHD MODEL 5 Rd is the specific gas constant for dry air, Rw is the specific gas constant for water vapor, and is the density of moist air. Rueger (2002) has also given a very accurate formula for computing the refractivity. Several authors have determined refractivity constants empirically and theoretically. However, the estimates are not always in good agreement for the water vapor content (Davis et al. 1985). The first term in Eq. (4) is no longer a pure dry component as the total mass density contains the contribution of water vapor. Hence, this term is referred to as hydrostatic component of refractivity. Integration of the hydrostatic component of refractivity along zenith direction constitutes the zenith hydrostatic delay, ZHD, which can be expressed as (Bosser et al. 2007) follows ZHD = 10 −6 ra K1 Rd dz , (6) rs where rs is the geocentric radius of the receiver antenna (m), ra is the geocentric radius of the top of the neutral atmosphere (m), and z is the integration variable. The empirical ZHD models (i.e. Saastamoinen, Hopfield etc.) have evolved from this equation under varying assumptions. 4. Study area and source of radiosonde data Radiosonde measures temperature, pressure, and humidity at various ascending points. The radiosondes are expensive, and the cost of these devices restricts the number of launches to twice daily (0000 and 1200 UTC) at a limited number of stations (Bevis et al. 1992). Balloons are used to lift radiosonde devices into air at the required ground station. These balloons usually can ascend a height up to 20 km. The radiosonde sends down the measured data, which gets stored in files. The instruments measure temperature and relative humidity with accuracies of ∼0.2◦ C and ∼3.5%, respectively, with diminishing performance in cold, dry regions (Elliot and Gaffen 1991). The pressure obtained from standard radiosonde has an accuracy of 0.5 hPa (Guharay et al. 2010). These data contains vertical profiles of Temperature, Pressure and relative humidity but poor in spatial and temporal resolution. From these profiles, the altitude profiles of refractivity N can be obtained (Eq. (4)) and further, from profiles of N , the tropospheric delay gets estimated. In order to develop a site-specific ZHD model, a radiosonde launching station located at New Delhi at 28.58 N, 77.20 E having elevation 216 m above MSL is selected. Four years of radiosonde data for New Delhi station has been downloaded from Wyoming university website (www.weather.uwyo.edu/upperair). Two files of radiosonde data of each day (0000 UTC and 1200 UTC) are available. The records that do not contain atmospheric pressure has been rejected at pre-processing stage along with vertical profiles which are not available up to or beyond 20 km. Further, there appears some horizontal drifting of radiosondes and termination of observation after certain altitudes. Acta Geod. Geoph. Hung. 47, 2012 D SINGH et al. 6 5. Estimation of zenith hydrostatic delay from radiosonde profiles Under the assumption that dry air is in hydrostatic equilibrium and the equation of state, Eq. (6) can alternatively be written (Mekik 1997, Vedel et al. 2001) as: ZHD = 10 −6 ps K1 Rd 0 dp , g (7) where, ps is the surface pressure in mbar. As pressure and gravity (g) change with height, both have been considered as variable in Eq. (7). Since the pressure decreases with the ascent of balloons, ps is the upper limit of the integral. The lower limit of zero pressure has been assumed to occur above 100 km from the earth surface. The hydrostatic part of the integrals however need to be accounted for after the last ascending point of the radiosonde profile, since the atmosphere will still add to the integral in Eq. (7) (Mekik 1997). This can be obtained from Eq. (7) denoted as upper tropospheric correction and is given as follows pt , (8) ∆ZHD = 10−6 K1 Rd gr where, pt and gt are the pressure and gravity at the top of the radiosonde vertical profile. Equation (8) can be obtained from Eq. (7) by substituting the lower pressure as zero and the top pressure as pt . Equations (7) and (8) can be combined to yield the total zenith hydrostatic delay as follows ZHDRS = 10−6 K1 Rd t (pi − pi+1 ) i=1 gi + 10−6 K1 Rd pt , gt (9) where i and i + 1 indicate the sequence of layers of radiosonde data profiles from the surface towards vertical direction and t is the top vertical profile of the data. Pressures pi and pi+1 denote the atmospheric pressure at geo-potential height hi and hi+1 . Here g1 denotes the absolute gravity at the site (ground). The zenith hydrostatic delay estimated by radiosonde data is represented by ZHDrs . In the present work Eq. (9) is used to estimate total zenith hydrostatic delay invoking four years radiosonde data from New Delhi. The refractivity constant K1 has been considered to be 77.604 K/mbar, the dry gas constant Rd as 287.05 J/kg.K, the dry molecular weight Md as 28.9644 kg/kmol and the universal gas constant R as 8314.34 J/Kmol.K (Katsougiannopoulos et al. 2006). Another important parameter used in Eq. (9) is the gravity. But due to non-availability of absolute gravity at the radiosonde launching station at New Delhi, the available data of absolute gravity measured at NPL New Delhi, in 1971 by WRGRN (World Relative Gravity Reference Network, USA, Air Force) has been used. The baseline distance between the two stations is about 7 km and the elevation difference is 12 m. The spatial variation of absolute gravity is one mGal per mile (1.6 km) at mid latitude (Griffiths and King 1965). Thus, the absolute gravity of the location (28◦ 34◦ 9’N, 77◦ 7E, 228 m) has been found to be 979 137.8 mGal (1 mGal = 0.00001 m/s2 ) invoking a gravity gradient of 0.3086 mGal/m (Li and Gotze 2001). Acta Geod. Geoph. Hung. 47, 2012 SITE-SPECIFIC ZHD MODEL 7 Fig. 1. A scatter plot showing the variation of estimated zenith hydrostatic delay (2007) Fig. 2. A scatter plot showing the variation of surface pressure annually (2007) 6. 6.1 Results Results of estimated ZHD at New Delhi Approximately 700 samples have been considered for the estimation of ZHD. Applying the Eq. (9) and using vertical atmospheric pressure profiles from the radiosonde data of the year 2007, ZHD has been estimated for the year at 0000 UTC and 1200 UTC. The annual mean of ZHD has been found to be 2.224 m having a standard deviation of 0.0156 m. It has been observed that the minimum value of estimated ZHD is during June–July and the maximum is during December– January. The estimated ZHD varies from 2.184 m to 2.255 m corresponding to a surface pressure of 958 mbar to 989 mbar. Therefore, the range of the estimated ZHD is 0.071 m. Figure 1 shows the annual variation of the estimated ZHD for station New Delhi for the year 2007 along with mean and standard deviation. This difference in the variation of the estimated ZHD is mainly caused by large annual variability in atmospheric dry constituents at the station New Delhi. The annual mean, minimum and maximum of surface pressure (ps ) at the New Delhi station for the year 2007 are 975 mbar, 958 mbar and 989 mbar respectively and its variation is as shown in Fig. 2. Acta Geod. Geoph. Hung. 47, 2012 D SINGH et al. 8 Fig. 3. A scatter plot of estimated ZHD versus surface pressure along with regression line (R2 = 0.9998) Figures 1 and 2 reveal that the trends of both ZHD and surface pressure have the same pattern and thus, can be inferred that variation in ZHD is related to the variation in the surface atmospheric pressure and thus, can be considered as the guiding parameter towards development of the proposed model. 6.2 Development of site specific ZHD model (SSM) for New Delhi In order to develop a site-specific ZHD model, ZHD is estimated through numerical integration of respective atmospheric vertical profiles obtained from radiosonde data using Eq. (9). A regression line represented by Eq. (10) gets best fitted to the scatter plot of the estimated ZHD against surface atmospheric pressure (Fig. 3) ZHDRS (m) = 0.0022677 · SurfacePressure (mbar) + 0.0121318 . (10) The model (Eq. (10)) provides the integrated hydrostatic delay all the way up to surface. It has been found that the model provides highly correlated (0.9998) value with that obtained from radiosonde observation with rms error as 0.19 mm and it is accurate up to 0.13 mm. 6.3 Validation of site-specific ZHD model In order to validate the developed model, site specific ZHD values have been computed using the model and that from radiosonde data sets of the years 2006, 2008, 2009. Considering the ZHD values estimated from radiosonde data (as true value), some significant statistical parameters have been computed as shown in Table II. It can be found that rms error is always less than or equal to about 0.3 mm and mean absolute difference is less than 0.2 mm. Acta Geod. Geoph. Hung. 47, 2012 SITE-SPECIFIC ZHD MODEL 9 Table II. The statistical parameters of the developed site-specific ZHD model for New Delhi Year R2 RMS m Bias m MAD m 2007 2006 2008 2009 0.9998 0.9996 0.9995 0.9996 0.00019 0.00028 0.00033 0.00027 0.000003 0.000097 0.000140 0.000100 0.00013 0.00016 0.00019 0.00016 2007 0.9997 Saastamoinen 0.00023 0.000110 0.00018 Fig. 4. A scatter plot comparison of ZHD using different models for 2007 at New Delhi The ZHD values using the Saastamoinen model and that estimated from radiosonde profile data at Delhi for the year 2007 is given in Table II. It can be observed that the rms is 0.00023 m and MAD as 0.00018 m for Saastamoinen model while for site specific ZHD model rms is 0.00019 m and MAD is 0.00013 m for 2007. Hence, the site-specific ZHD model provides a slightly better result than the Saastamoinen ZHD model. Hence, the developed model can be used for a precise estimation of the precipitable water vapor (PWV) for New Delhi. 6.4 Comparison with other ZHD models A comparison of the ZHD values obtained from the developed site-specific model (SSM developed), the SSM model developed by Raju et al. (2007a), the Saastamoinen ZHD model (Saastamoinen) and the ray-traced radiosonde data (Est ZHD (RS)) have been done, as shown in Fig. 4. It can be observed that ZHD values from Saastamoinen and from the site-specific model are close to each other and also coincide with the ray-traced ZHD, while there is a large systematic difference of about 19 mm to the SSM model by Raju et al. (2007a). The large deviation might be due to a difference of the methods of estimation of ZHD. In case of Raju’s SSM, ZHD is estimated by numerical integration of pressure and temperature vertical profiles of radiosonde data available up to 25 km and Acta Geod. Geoph. Hung. 47, 2012 D SINGH et al. 10 above. While, in this study, ZHD is estimated by numerical integration of pressure and gravity profiles available up to 20 km and for atmosphere above 20 km, an upper tropospheric correction has been considered. 7. Discussions In this study, an account of the ZHD of GPS signals has been explained briefly. This delay plays a crucial role in GPS PWV estimation. The study focuses on the development of a site-specific zenith ZHD model based on radiosonde profiles. The model permits estimation of the delay due to dry air present in the atmosphere. The hydrostatic delay introduced by troposphere depends on the geographic location and atmospheric condition. Therefore, in this study a site-specific ZHD model has been developed that is applicable to a particular place viz. the radiosonde launching station at New Delhi. The model treats the ZHD as a function of surface pressure and thus, permits continuous estimation of ZHD. The model has been validated by comparing its values with the estimated ZHD obtained from radiosonde data collected for the years 2006, 2008 and 2010 respectively at the New Delhi station. The precision of the model is about 0.3 mm (RMS) and 0.2 mm in respect of the mean absolute difference. The model invoking the surface pressure, performs slightly better than the Saastamoinen ZHD model. However, the study reveals that the ZHD estimate is quite sensitive to the surface pressure. An error of 1 hPa in the surface pressure measurement introduces an error of about 2.3 mm error in ZHD estimate. Thus, it is important to measure the pressure accurately for ZHD estimation. 8. Conclusions A site-specific ZHD model has been developed as a function of surface atmospheric pressure model. The model considers the geometrical height same as the geo-potential height i.e., neglects their difference (Wallace and Hobs 1977). Absolute gravity of the station has been extrapolated from that of a station that is about 7 km away and has 12 m difference elevation. In addition, for different parameters such as the vertical gravity gradient, refractivity constant, dry molecular weight, universal gas constant standard values have been considered. In spite of all these assumptions/approximations, the site-specific model performs slightly better than Saastamoinen (1972) ZHD model. Acknowledgements Authors are thankful to the Department of Science and Technology for funding of the project. Further, authors acknowledge with thanks the Wyoming University, for making available Radiosonde data online. References Baby H B, Gole P, Lavergnat J 1988: Radio Sci., 23, 1023–1038. Bai Z, Feng Y 2003: J. Global. Position. Systems., 2, 83–89. Bevis M, Businger S, Herring T A, Rocken C, Anthes R A, Ware R H 1992: J. Geophys. Res., 97, 15787–15801. Black H D 1978: J. Geophys. Res., 83, 1825–1828. Acta Geod. Geoph. Hung. 47, 2012 SITE-SPECIFIC ZHD MODEL 11 Bock O, Doerflinger E 2001: Phys. Chem. Earth (A), 26, 373–383. Bosser P, Bock O, Pelon J, Thom C 2007: IEEE Geosci. Remote Sens. Lett., 4, 3–7. Davis J L, Herring T A, Shapiro I I, Rogers A E, Elgered G 1985: Radio Sci., 20, 1593–1607. Elgered G, Davis J L, Herring T A, Shapiro I I 1991: J. Geophys. Res., 96, 6541–6555. Elliot W P, Gaffen D J 1991: Bull. Amer. Meteorol. Soc., 72, 1507–1520. Griffiths D H, King R F 1965: Applied Geophysics for Engineers and Geologist, University of Bermingham. Elsveir, United Kingdom Guharay A, Venkat Ratnam M, Nath D, Dumka U C 2010: Radio Sci., 45, 1–14. Hauser J P 1989: J. Geophys. Res., 94, 10182–10186. Hopfield H S 1969: J. Geophys. Res., 74, 4487–4499. How to compute Geoidal separation NGA EGM 96 (online) available at: http://earth-info.nga.mil/nga-bin/gandg-bin/intpt.cgi (Accessed 20 May 2010) Janes H W, Langley R B, Newby S P 1991: Bull. Geod., 65, 151–161. Katsougiannopoulos S, Pikridas C, Rossikopoulos D, Ifadis I M, Fotiou A 2006: In: Shaping the change. 23th FIG Congress Munich, Germany Li X, Gotze H-J 2001: Geophysics, 66, 1660–1668. Mekik C 1997: Tropospheric delay models in GPS. International Symp. On GIS/GPS, Istanbul, Turkey Mendes V 1999: Modeling the neutral atmosphere propagation delay in radiometric space techniquees. PhD. Dissertation. Technical report, No. 159, Department of Geomatics Engg, University of New Brunswick Mendes V B, Langley R B 1998: In: Proceedings of the institute of Navigation 54th Annual meeting, Denver, CO, USA, 337–347. Mendes V B, Prates G, Santoss L, Langley R B 2000: In: Proceedings of the 2000 National Technical Meeting of the institute of Navigation, Anaheim, CA, 433–438. MOPS 1998: Minimum operational performance standards for Global position system/wide area augmentation system airborne equipment. Technical report, Document No. RTCA/DO-229A, prepared by SC 159. Raju C Suresh, Saha K, Parameswaram K 2007a: J. Atmos. Sol-Terr. Phy., 69, 875–905. Raju C Suresh, Saha K, Thampi B V, Parameswaram K 2007b: Ann. Geophysics., 25, 1935–1948. Rueger J M 2002: Refractive index formula for radio waves. FIG 22th International Congress Washington, D.C., USA Saastamoinen J 1972: In: Papers presented at the Third Int. Symp. on The use of artificial satellite for Geodesy. Geophysical Monograph Series, S W Henriksen, A Mancini, B H Chovitz eds, AGU, Washington DC, Vol. 15, 247–251. Satirapod C, Chalermwattanachai P 2005: J. Global Position. Systems, 4, 36–40. Smith E K, Weintraub S 1953: In: Proceedings of IRE, 41, 1035–1037. Thayer D 1974: Radio Sci., 9, 803–807. Vedel H, Mogensen K S, Huang X Y 2001: Phys. Chem. Earth., 26, 497–502. Wallace J M, Hobbs P V 1977: Atmospheric Science: An introductory survey. Academic Press, Elsveir, USA World Relative Gravity Reference Network (online) available at: http://geomaps.wr.usgs.gov/gump/gravity− base− stations/wrgrn/WRGRN-orig1971Section4− Asia.new.pdf) (Accessed 20 may 2010) Wyoming University Upper air data (online) available at: www.weather.uwyo.edu/upperair (Accessed 20 May 2010) Acta Geod. Geoph. Hung. 47, 2012
© Copyright 2026 Paperzz