Observational Techniques: Sampling the Mountain Atmosphere

Abstract

In this chapter several types of instruments are surveyed that have contributed to increased understanding of the structure of atmospheric flows and processes that drive these flows in complex terrain. After a brief example of the use of in-situ measurements, those that are in contact with the air they are measuring, a review of basic remote sensing principles serves as a background for a discussion of several types of remote sensors. The review focuses on active remote sensing systems, those that transmit and receive their own signal, including sodar, radar, radar wind profilers, and lidar. Examples of the kind of data available from each type of system allow an appreciation for how these systems contribute to advanced understanding of atmospheric processes by providing data above the surface and revealing the evolving vertical and horizontal structure of complex terrain flows. Advantages of deploying combinations of complementary instruments are described. The roles of measurements and mesoscale numerical models are discussed in a section where modeling studies have used high-resolution remote sensing and other advanced measurement systems to verify the models. Finally a brief section describes turbulence measurement techniques by remote sensing that may be able to provide profiles of turbulence quantities in mountain studies.

Appendix

To begin to understand Doppler processing it is necessary to probe deeper into sine-wave behavior. Recall that the three basic specifications for a sine function (Fig. 8.57, top panel) are its frequency f, its amplitude A, and its phase φ:

$$ {\text{s}}\left( {\text{t}} \right){ } = {\text{ A \, sin }}\left( {{2}\pi f \cdot {\text{t }} + { }\varphi } \right) $$

and that the frequency and wavelength are interchangeable, related by the speed of propagation of the wave according to the relationship: c = λ∙·f or λ = c/f. Radar and lidar have fundamentally different approaches to the Doppler determination.

Fig. 8.57

Sine-wave signal as a function of time t (top). Outgoing signal at an instant of time as a function of distance x, reflecting off an object at distance D ₁ . Return signal indicated by dashed curve (middle). Same as middle panel, except reflecting object is at distance D ₂ (bottom)

The radar case is illustrated in Fig. 8.57. First consider for convenience a continuous-wave (CW) signal, and for now consider this signal at an instant of time as a function of the distance x from a source (Fig. 8.57, middle panel), instead of as a function of t. Also for convenience, consider the instant when the CW signal has initial phase φ₀ = 0 at its source, i.e., where x = 0—the source would probably be the instrument antenna. Then, let the signal intercept a target at distance D₁ and return to the source location. Note that where the signal intercepts D₁, it is most often at a different phase from the phase at the source φ₀, because in general D₁ is not at an integral multiple of λ from the source. The signal “reflected” back to x = 0 has also returned at a φ different from φ₀. If the target is not moving at D₁ along x, we can change perspective and look at the “reflected” signal received back at x = 0 as a function of time. This transmitted source signal is a temporally varying sine wave of frequency f, as in the top panel of Fig. 8.57. The received signal back at the source position x = 0 as a function of time will also be the sine wave synchronized (at frequency f) with the transmitted signal, but with a constant phase offset Δφ₁. (The mean particle/droplet motion produces a relatively tiny change in the frequency of the backscattered signal, but at radar wavelengths this frequency shift is not used to calculate Doppler velocities.)

If the target is located at a different distance D₂, then the phase intercept at D₂ will be changed (Fig. 8.57, bottom panel) and the phase offset received back at the source position will also have shifted, i.e., Δφ₂ will be different from Δφ₁. Where this argument becomes related to Doppler processing for radar, is if the target moves from D₁ to D₂ in a time of Δt, then the speed of movement is related to the rate of change of the phase offset, expressed as [(Δφ₁ − Δφ₂)/Δt]. This leads to a derivative expression d(Δφ)/dt, which is proportional to the speed of movement of the target (see Doviak and Zrnić 1993, Eq. 3.30; Battan 1973, Eq. 8.2; Rinehart 2010, p. 99). At Doppler radar λ’s, this phase-shift approach works for atmospheric returns, even though the scatterers are not a solid target as portrayed here, because (1) the scatterers in each pulse volume as a whole have not moved more than a distance equivalent to λ in the time interval between pulses, and (2) the distributed scatterers are not significantly rearranged in the time interval between shots in such a way that the phase information within each pulse volume is distorted. The latter could be stated as, for scatterers seen at radar λ’s, the atmospheric decorrelation time due to finescale turbulence motions is larger than the pulse repetition time of the measurements. Doppler radars therefore actually measure the phase shift from pulse to pulse, which has units of frequency, but is not really a measurement of the frequency of the signal. Therefore this method has sometimes been referred to as “pseudo-Doppler” processing.

For the much smaller λ’s of lidar these conditions are not met. The distributed aerosol-particle population does move a distance of more than the lidar λ in the time between successive pulses, and turbulent motions do rearrange the particles with respect to each other in such a way that the phase information in each range gate is too scrambled to be useful for velocity calculations. Lidar systems perform a more traditional Doppler approach, by directly measuring the frequency of the return signal and comparing it to the outgoing frequency, calculating the Doppler shift as in (8.5).