[SG] Atmosphere

General influence of atmosphere refraction

• Two velocity in atmosphere: phase and group velocity
  
  • Background: a simple harmonic wave propagating in air looks like $u(x, t) = A_0 * sin(\omega t - kx + \phi_0)$, where $\omega$ is angular frequency, represented in radians per second, k is angular wavenumber, represented in radians per meter.
  • group velocity is the propagating speed of an evenlope wave, which is modulated(overlap several waves) to carry information. On the contrary, a simple harmonic wave(showed before) can not carry any information.
  • Phase velocity: $$v_{ph} = \omega / k = \lambda / T$$
  • Relationship between group velocity and phase velocity: $$v_{gr} = v_{ph} - \omega * \frac{d v_{ph}}{d\omega}$$
  • This means when $v_{ph}$ is independent with the wavelength(no dispersion), then $v_{gr} = v_{ph}$.
• Structure of atmosphere: troposphere and ionosphere
  
  • troposphere: ranging from 0km to 10 km above the ground
    
    • Note that for microwave the tropospheric refraction is non-dispersive. This means the signal delay caused in troposphere in this frequency range is irrelevant with the frequency.
    • While for optical range the tropospheric refraction is frequency-dependent and non-dispersive.
  • ionosphere: range from 50km to 1000km above the ground
    
    • the free electrons and ions in ionosphere is related to the solar radiation.
• The signal delay $\delta \rho_{atm} = \int_L(n(s) - 1) ds + (S-G)$
  
  • Here, S is the curved path of propagating light, while G is the geometric distance.
  • The first term can achieved highly at 200m, while the latter is negilible (excepet for low zenith angle) , 1 to 2 centimeter.
  • Generally speaking the refractive index depends on frequency. By definition $n_{ph} = c / v_{ph}$ and $n_{gr} = c / v_{gr}$.

Ionospheric refraction correction

• Ionospheric Refraction: phase advance, group delay
  
  • For frequency bigger than 100MHz, the mathematical model of Ionospheric refraction: $n_{ph} = 1 - a * \frac{N_e}{f^2}$ and $n_{gr} = 1 + a * \frac{N_e}{f^2}$ Dispersive!!: depends on the frequency
  • a is constant: $a = a = 40.3 m^3 s^{-2}$; $N_e$ refers to the free electrons density along signal path.
  • Correction for ionospheric refraction: $$\delta \rho_{ph} = -a * \frac{E}{f^2}$$
    $$\delta \rho_{gr} = a * \frac{E}{f^2}$$
    Here E denotes the Total Electron Content: $E = \int N_e(s) ds$
  • ionospheric refraction correction is negative for phase measurement (phase advance), while positive for code/pseudorange measurement(group delay).
  • Mostly only the VTEC (Vertical Total Electrons Content) $E_V$ is provided, to get E in a specific direction we need a map function.(See page 101)
  • Final formula: $$\delta \rho_{ph} = -a * m_{ion}(z)\frac{E_V}{f^2} = -a * \frac{1}{cosz^*}\frac{E_V}{f^2}$$
  • Using the above formula, we assume all electron concentrated in an infinitesimally thin layer above the earth surface.
  • The smaller the frequency, the bigger the inonspheric correction. So for GNSS (L-Band) is 1 - 15m. For VLBI using X-Band is much smaller: 4cm - 60cm. For Altimetry (X or Ku-Band) is 1 cm - 15 cm.
  • For SLR/LLR which employ optical wave (frequency: $5 * 10^{14}$) the delay casued by ionospheric refraction is negilible.
  • Ionospheric-free linear combination to eliminate the ionospheric delay.

Tropospheric delay for radio wave: Dry part or Wet part

• For radio waves range from 100MHz to 15GHz the tropospheric delay is non-disperive. While for optical signal is dispersive.
• Tropospheric delay can still be modelled as $\delta \rho_{tro} = \int (n(s) - 1) ds$. Here we introduce refractivity $N = 10^{6} * (n - 1)$. So $\delta \rho_{tro} = 10^{-6} \int N(s) ds$
• Model of refractivity: $$N = N_d + N_w = k_1 * \frac{P_d}{T} * Z_d^{(-1)} + k_2 * \frac{e}{T}*Z_w^{(-1)} + k_3 * \frac{e}{T^2} * Z_w^{(-1)}$$
  
  • Refractivity not only depends on pressure and compressibility of air, but also depends on electrical characterstics of molecules, where $N_2$ or $O_2$ are different with $H_2O$.
  • e : partial pressure of water vapour, a function of relative humidity
  • Total delay amounts to 2.3m in zenith direction, and about 25 at an elevation of 5 degree.
  • Dry part contributes to 90% of tropospheric delay, and can be modelled quite well based on the pressure measurment on the ground, and changes rather slowly.
  • Wet part amounts to maximum about 40 cm, but is highly variable in time and space.
• Tropospheric delay can be writen as the product of zenith delay and map function $m_{trp}(z)$. Normally the mapping function for dry part and wet part is different. So the following formular is given: $$\delta \rho_{tro} = m_{trp, d}(z) * \delta \rho_{tro, d}^0 + m_{trp, w}(z) * \delta \rho_{tro, w}^0$$
• Tropospheric correction can be modelled based on meterological data including air pressure P, temperature T and relative humidity H(Using water vapor radiometry). You can also based on standard meterologival model, but is not accurate enough.

Tropospheric delay for optical part of spectrum (SLR/LLR)

• As for tropospheric delay for optic wave, the contribution of dry part (partial pressure of dir air) is one order higher than the wet part(partial pressure of water vapour)
• The distance correction at normal condition in zenith direction amounts to 2.45m and 7m at the elevation angle of 20 degree.
• The magnitude of tropospheric correction of dry part for optical and radio wave are similar(2.3 -2.4m). While for wet part, optical is much smaller(up to 5mm).
• a change of meterological condition result in a change of tropospheric correction at several mm level, due to the low coefficient of wet air pressure. Tropospheric correction for optical wave is less variable.
• Thus troposhperic delay can be modelled sufficiently accurately based on the meterological measurement at station.
• By contrast, for VLBI or GPS, the change of meterological condition result in significant change of tropospheric correction(up to 40cm).

Estimation of troposphere Parameter: zenith troposphere delay

• Normally ZTD is estimated for every station involved and is valid for 30mins to 120mins.
  
  • By contrast, for the influence of dry part of tro towards radio waves the pressure measurment is valid 12hours, with a resulting error less than 2cm.
• The main problem of tro parameter estimation is the correlation between the parameters, because they all change with the change of zenith angle
  
  • station height
  • receiver clock correction
  • troposphere delay
  • phase ambiguities
  • offset and variation of antenna phase centers
• Ways to eliminate correlation:
  
  • at least 1-2 hours observation
  • in static mode
  • open sky(good satellite geometry)
  • Typically estimate one ZTD per hour