Analysis of Gravity Waves from Radio Occultation Measurements Martin Lange and Christoph Jacobi Institute for Meteorology, Stephanstr. 3, 04103 Leipzig mlange@uni-leipzig.de, jacobi@uni-leipzig.de Summary. In the height range 10-30 km atmospheric gravity waves lead to periodic perturbations of the background temperature field in the order of 2-3 K, which can be resolved in temperature profiles derived from radio occultation measurements. Due to the spherical symmetry assumption in the retrieval algorithm and the low horizontal resolution of the measurement weakening in the amplitude and phase shift of the waves occurs. The influence of the geometric wave parameters and the measurement geometry on a homogeneous spectrum of plane gravity waves in the range 100-1000 km horizontal and 1-10 km vertical wavelength is investigated with a 2D-model ranging ±1000 km around the tangent point and 10-50 km in height. The investigation shows that with radio occultation measurements more than 90% of the simulated wave spectrum can be resolved with relative amplitudes above the 1/e level. Considering the total variance, about 80% is retrieved in the worst case when the GPS receiver scans perpendicular to the wave crests and about 88% as mean value when the view angle of the receiver through the gravity waves is arbitrarily oriented. More realistic wave spectra lead to slightly smaller values. Key words: Gravity waves, gravity wave activity, radio occultations, satellite measurements, model simulations 1 Introduction Gravity waves (GW s) play a main role in the circulation of the middle atmosphere. GPS radio occultation (RO) measurements provide a new tool to estimate GW activity on a global scale (Tsuda et al., 2000). Unfortunately, due to the low horizontal resolution of about 200-400 km (Kursinski et al., 1997) and the spherical symmetry assumption in the retrieval algorithm for atmospheric parameters, weaker amplitudes and a phase shift is derived in the temperature profiles depending on the gravity wave parameters. Short waves with horizontal wavelengths smaller than 50-100 km are filtered out due to the averaging over positive and negative phase variations of the GPS signal, whereas waves with larger horizontal wavelengths than 1000 km match the spherical symmetry assumption (Belloul and Hauchecorne, 1997). Therefore the vertical profiles of small gravity waves cannot be derived directly, and the potential energy derived from the temperature variance used to describe climatologies of gravity wave activity is estimated too low.
480 Martin Lange and Christoph Jacobi The task of this study is to investigate to what extent plane gravity waves with horizontal wavelengths of 100-1000 km and vertical wavelengths of 1-10 km can be derived by the RO technique. The GW signal, that is seen by the GPS receiver of the low earth orbiter (LEO) satellite is investigated by applying the inversion technique on excess path delays introduced by simulated gravity wave fields in an isothermal atmosphere in dependence on the horizontal and vertical wavelengths and the view angle of the satellite through the wave. A 2D model is used, that ranges ±1000 km around the tangent point and from 10 to 50 km in height to derive the temperature profiles of the simulated waves. Ionospheric effects are not considered here and the influence of the upper part of the atmosphere is neglected which is justified because of the low density there. This is in correspondence with usual temperature retrievals, that use upper boundary temperatures from external analysis at about 60 km (Hocke, 1997). 2 Model setup 2.1 Retrieval of temperature variations from gravity waves To simulate the retrieval of given temperature perturbations from gravity waves with constant amplitude in the range λ x =100-1000 km and λ z =1-10 km (see Table 1) excess path delay of the GPS signal is calculated from the corresponding variations in refractivity for the height range 10-50 km. Temperature profiles are then retrieved from the inversion equations of Fjeldbo and Eshleman (1965). Dividing the atmospheric delay Φ 1 into a mean part and a wave part Φ 1 the corresponding perturbation of the refractive index n may be written: n(r) = λ/π r/2r d( Φ 1 )/dr +λ/π r+ r ( Φ 1 (r) Φ 1 (ξ))(ξ 2 r 2 ) 3/2 ξdξ. (1) Here λ denotes the wavelength of the GPS signal, r the distance of the tangent point from Earth s center and ξ the integration variable. Assuming small perturbations and a dry atmosphere, through the ideal gas law the density ρ depends only on the refractivity N and the perturbation equations for the atmospheric parameters are expressed as follows: N = 10 6 n = k 1 R ρ (2) ρ = p 0 /(RT 0 ) T/T 0 + p/(rt 0 ). (3) R is the gas constant, k 1 =77.6 K/hPa is a proportionality factor, T 0 and p 0 represent the temperature and pressure of the undisturbed atmosphere. Assuming adiabatic conditions the pressure variation may be expressed by
Analysis of Gravity Waves from Radio Occultation Measurements 481 z K y 50 km 10 km 2000 km x Fig. 1. Measurement geometry of the occultation. The model domain covers a horizontal range of 2000 km and extends from 10 to 50 km in the vertical. For a better view the vertical scale is exaggerated. the temperature disturbance and the temperature signal becomes directly related to the refractivity: T = 1/k 1 2/5 T 2 0 /p 0 N. (4) 2.2 Measurement geometry The measurement geometry is shown in Fig. 1. It shows a typical wave pattern propagating through the atmosphere. The lines of constant phase are tilted against the Earth s sphere. During the occultation of the GPS satellite the ray path intersects the constant phase lines of the gravity wave and the receiver scans the vertical profile of the periodic temperature disturbance top down. The model domain, where the temperature disturbance is calculated from the path delay of the GPS signal introduced by the simulated gravity waves, is indicated as the shaded rectangle. We assume a dry atmosphere which is justified in the upper troposphere and above. Therefore the lower boundary of the model is set to 10 km. This takes also into account that GW s are often generated in the height range between 5 km and 10 km. The model parameter are listed in Table 1. 2D Model domain 2000 km 40 km Range: x = ±1000 km around the tangent point, z=10-50 km Grid: 1000 500 grid points Grid resolution: 2000 m 80 m Wave Perturbation: T = T 0 cos(kx + m(z-z 0 )); z 0 =10 km, T 0 =5K GW spectrum: λ x =100-1000 km, λ z =1-10 km Table 1. Model and wave parameter used for the simulation of plane gravity wave temperature perturbations.
482 Martin Lange and Christoph Jacobi 3 Results 3.1 Visibility of plane gravity waves Fig. 2 shows a general picture of the visibility and the vertical phase shift of gravity waves in dependence on the horizontal and vertical wavelength. Following an approach similar to that of Preusse et al. (2001), who derived gravity waves from thin layer radiation of simulated GW temperature perturbations, visibility is defined here as the ratio between the derived temperature amplitude and the original amplitude. The dashed lines in the left of Fig. 2a) indicate visibility of 1/e and 1/e 2. The 90% isoline apears above the diagonal of the plot. Thus more than 50% of the waves are retrieved with an amplitude above that level. Calculating the total temperature variance T 2 for all waves (not explicitly shown), about 80% of the original variance is derived. Nevertheless the phase shift with respect to height, that occurs from the retrieval algorithm (panel b), is quite remarkable for a major part of the waves. Although it decreases with increasing visibility, about 40% of the waves undergo a phase shift of more than 45. At combinations of short horizontal and long vertical wavelengths, see the left side of panel b), where the phase shift exceeds 60, the contour lines become irregular. For example at a horizontal wavelength of 200 km the phase shift maximizes at a vertical wavelength of 6 km with values above 160 and decreases again at larger wavelengths. All the results hitherto described are obtained under the assumption, that the LEO satellite views perpendicular to the wave crests. The same temperature amplitudes may be derived for even shorter horizontal wavelengths, if the view angle between the line of sight and the horizontal wave vector is larger. Therefore the worst case of geometry has been considered for plane gravity waves. To estimate this effect, the visibility in dependence on the Fig. 2. Visibility of plane gravity waves (a) defined as amplitude ratio between the retrieved and the original temperature perturbation and phase delay φ of the derived wave perturbation (b).
Analysis of Gravity Waves from Radio Occultation Measurements 483 Fig. 3. Visibility of plane gravity waves (a) and phase difference φ (b) between the derived and the original wave dependend on the horizontal view angle between the line of sight and the horizontal k-vector of the wave. The respective horizontal wavelength is 200 km. vertical wavelength and the horizontal view angle is considered. Figure 3 shows the visibility (a) and the phase difference (b) between the derived and the original gravity wave perturbation dependent on the view angle α between the line of sight and the horizontal wave vector for a horizontal wavelength of 200 km. Since the horizontal wavelength seen by the GPS receiver increases with sec(α), the visibility approaches unity while turning the occultation ray in the direction parallel to the wave crests. Also the phase delay decreases with increasing view angle as expected from Fig. 2 for longer horizontal wavelength. At vertical wavelengths above 7 km the phase shift cycles through a maximum with increasing view angle reflecting the retrieval error when the wave vector is tilted against the vertical and spherical symmetry is disturbed. Considering a homogeneous distribution of the view angles through the gravity waves for the spectrum considered in Fig. 2 the total derived mean variance increases from 80% for α = 0 to 88%. Assuming a more realistic vertical wavelength spectrum with a slope of k 5/3 in power spectral density (not shown) the mean observed variance reduces additionally by 2-3% depending on the wave parameters. The larger contribution from long vertical wavelength accounts for that. 4 Conclusions The signal of temperature perturbations of plane gravity waves, derived from radio occultation measurements, depends on the geometrical wave parameters and the horizontal view angle between the occultation ray and the wave. Incomplete resolution of the amplitudes and vertical phase shift are introduced during the retrieval increasing the effect with shorter horizontal and
484 Martin Lange and Christoph Jacobi larger vertical wavelength. For the simulated gravity wave spectrum with constant amplitude in the range 100-1000 km horizontal and 1-10 km vertical wavelength about 80% of the total variance of the spectrum is derived in the worst case, when the view angle of the GPS receiver is perpendicular through the wave crests. For isotropic distributed view angles, as can be assumed for large numbers of measurements used to derive climatologies of gravity wave activity, a mean variance of 88% is derived. Considering a more realistic vertical wavenumber spectrum with a slope of k 5/3 in power spectral density the mean observed variance reduces additionally by 2-3% due to the larger contribution of the long vertical wavelengths. Phase delay and amplitude damping increase simultaneously, therefore the part of the spectrum, that is difficult to identify is small compared to the rest. Additional information on gravity wave parameters will be provided by a more dense network of radio occultation measurements in future that allows for multiple measurements near the same tangent point at time scales shorter than one wave period. At horizontal wavelength below 200-300 km the resolution of gravity waves is enhanced when fresnel diffraction theory is taken into account in the retrieval algorithm. Acknowledgements This study was supported by the Deutsche Forschungsgesellschaft under grant JA 836-4/1 and by INTAS under grant 991-1186. The project is imbedded in the CHAMP mission lead by the GFZ Potsdam. References Belloul MB, and Hauchecorne A, (1997) Effect of periodic horizontal gradients on the retrieval of atmospheric profiles from occultation measurements. Radio Sci, 32, 469 478. Fjeldbo G and Eshleman VR (1965) The bistatic radar occultation method for the study of planetary atmospheres. J. Geophys Res, 13, 3217 3225. Hocke K (1997) Inversion of GPS meteorology data. Ann Geophys, 15, 443-450. Kursinski ER, Hajj GA, Schofield JT, Linfield RP, and Hardy KR (1997) Observing earth s atmosphere with radio occultation measurements using the global positioning system. J Geophys Res 102, 23,429 23,465. Preusse P, Dörnbrack A, Eckermann SD, Riese M, Schaeler B, Bacmeister JT, Broutman D, and Grossmann KU (2001 Space based measurements of stratospheric mountain waves by CRISTA. 1. Sensitivity, analysis method and a case study. J Geophys Res, accepted. Tsuda T, Nishida M, Rocken C and Ware RH (2000) A global morphology of gravity wave activity in the stratosphere revealed by the GPS occultation data (GPS/MET). J Geophys Res, 105, 7257 7273.