Finnish Meteorological Institute Vaisala
Wind profiler principles

WIND PROFILER

1 Working Principles of the Wind Profiler
1.1 A Remote Sensing Instrument
1.2 A Wind Profiling Radar
1.3 Range Gates and Resolution
1.4 Wind Profiler Signal Scattering
1.5 Doppler Principle
2 Antenna Pointing
3 Signal Processing
3.1 Coherent Averaging
3.2 Fourier Transform
3.3 Spectral Averaging
3.4 Moments Calculations
3.5 Consensus Data
4 Radio Acoustic Sounding System (RASS)
5 Wind Profiler Altitude Performance
5.1 Wavelength
5.2 Size of the Antenna
5.3 Transmitted Power
5.4 Recovery Rate of the Electronics
5.5 Pulse Duration
5.6 Inter-Pulse Period
5.7 Pulse Coding
6 Sampling Choices
6.1 Sampling Delay
6.2 Gate Spacing
6.3 Number of Range Gates
6.4 Interdependency of Parameter Choices
7 Height Coverage and Range Resolution for RASS
8 Atmospheric Effects on Profiler Performance
8.1 Humidity
8.2 Turbulence
8.3 Precipitation
8.4 High Winds
8.5 Temperature


Vaisala, 2005


1 Working Principles of the Wind Profiler

1.1 A Remote Sensing Instrument

The wind profiler is an atmospheric remote sensing instrument. An atmospheric remote sensing instrument gives information about a volume of the atmosphere at a distance without being physically located in the volume. Your eyes are a remote sensing device that can give you information about an object some distance away without actually touching the object. An in-situ instrument, such as an anemometer, gives information about a specific point in the atmosphere. Because a remote sensing instrument generally measures a volume and an in-situ instrument measures a single point, the data from these two different kinds of instruments cannot be easily compared.

A Remote Sensing and an In-Situ Instrument

1.2 A Wind Profiling Radar

The profiler is a pulse Doppler RAdio Detection And Ranging instrument, commonly called a radar. It transmits a pulse of electromagnetic energy at a chosen frequency and in a chosen direction. When the pulse encounters a "target", electromagnetic energy is scattered. A small portion of this scattered energy, called backscatter, will return to the radar, which can then compute the distance to the target from the time delay between the transmission and reception of the return signal.

1.3 Range Gates and Resolution

The generic name "profiler" comes from the radar's ability to show data for many heights of the atmosphere at the same time, thus giving a profile of the atmosphere. The profile from equally-spaced heights is created by sampling backscatter at equally-spaced time intervals. The sequential intervals during which backscatter is sampled are called range gates. The atmosphere above the profiler is divided into range gates, and all data from a chosen time span are placed into the same range gate.

Range Gates

Backscatter is arriving from a volume of the atmosphere, rather than a single point. The distance defined by the pulse duration defines the spatial resolution of the radar. The height assigned to the range gate is the center of the resolution volume or cell. The finer the range resolution of the profiler, the smaller the vertical size of the volume. A short pulse gives a fine range resolution, and a long pulse gives a coarse range resolution.

Representation of a Resolution Cell

1.4 Wind Profiler Signal Scattering

The signal scattering targets or the wind profiling radar are refractive irregularities in the atmosphere. A refractive irregularity is anything that can change the course of a wave through a medium. In this case, the medium is the atmosphere, and the wave is the radio frequency signal that the profiler transmits. Maximum backscattering power occurs when irregularities are about half the size of the radar wavelength (Bragg scatter).

The atmosphere is deceptively turbulent, powered by wind and the uneven heating of the Earth's surface. This motion in the atmosphere creates variations in temperature, humidity, and pressure over relatively short distances. The variations, called eddies, can start out quite large, but because they are unstable, will break up into smaller and smaller eddies. These eddies are the refractive irregularities detected by the profiler.

1.5 Doppler Principle

Tracking refractive irregularities, which are carried by the wind, reveals information about the wind itself. The profiler computes height by using the time interval between transmission of the pulse and reception of the return signal. However, wind speed and direction are determined by using the Doppler principle. A wave will shift in frequency because of the motion of the target relative to the observer. A frequency higher than the transmitted frequency indicates that the wind is moving towards the profiler. A frequency lower than the transmitted frequency indicates the wind is moving away from the profiler. The profiler detects these small shifts in the frequency of the backscatter and translates them into wind velocity data.

Page top

2 Antenna Pointing

The profiler makes measurements in at least three directions in order to compute wind measurements (speed and direction). The transmitted pulse is directed to an antenna that has a beam width of less than ten degrees. One beam is directed vertically. The four oblique beams are tilted from vertical (23.5 degrees for the 4-panel antenna at 915/924 MHz or 15.5 degrees for antennas at other frequencies or 9-panel antenna at 915/924 MHz) and directed in the four orthogonal directions. The phased transmission of the electromagnetic pulse to the radiating elements built into the antenna subsystem, tilts the orthogonal beams.

The Antenna Pointing Directions of the Radar Profiler

The beam sequence, including the number of beams and the order in which they are transmitted, is operator-controlled but normally should include the vertical beam and at least two tilted, orthogonal beams for the wind computations to work. A complete rotation through the beam sequence is called a sample. For each range gate, many samples are processed together to obtain an average for each range gate.

Page top

3 Signal Processing

The signal processing that the samples undergo is divided into two parts, the time-domain stage and the frequency-domain stage. In the time-domain stage, the samples are averaged or filtered. Time-domain averaging is also called coherent averaging (integration) because the samples come from targets that are more or less fixed in phase in relation to each other during the time the sample was taken and because the sample is taken using a coherent measurement system.

Signal Processing Flowchart

3.1 Coherent Averaging

Overall, a longer averaging time improves signal detectability. However, it is possible to choose an averaging time that is so long that it reduces the highest unambiguous speed of radial wind measurements, also called the full-scale velocity, detected by the wind profiler to an unacceptable value. This results in a radial wind measurement that may be out of the range required by the user. The number of samples taken is determined by the desired velocity range measured, the inter-pulse period, and the frequency of the radar. Operators must estimate the highest possible radial wind speed that might occur and choose a number that will allow the wind profiler to correctly detect that wind velocity if it does occur.

After coherent averaging, some of the DC voltage is removed by mathematical signal processing techniques. At this point, the time-domain stage of signal processing is complete.

3.2 Fourier Transform

In the second stage of signal processing, the sample is converted from the time domain to the frequency domain using an algorithm called a Fast Fourier Transform (FFT). The operator may choose the number of FFT points that are used in the FFT algorithm. The more FFT points used, the finer the velocity resolution is. However, the number of FFT points also helps determine the length of time necessary to acquire a spectral sample. A typical number of FFT points chosen is 64 to 256 for wind data, and 2048 for RASS data.

The next segment of signal processing uses a technique called windowing, which is then used to reduce some of the mathematical artifacts of the FFT and coherent averaging. After spectral averaging, signal processing techniques are used to remove ground clutter, which is backscatter from stationary targets that can be caused by things like buildings, power lines, and towers.

3.3 Spectral Averaging

The next processing step is spectral averaging. The operator may select the number of spectral averages that are used in the averaged spectrum, which is used in Moments selection. The inter-pulse period (IPP), the number of coherent averages, the number of FFT points, and the number of spectral averages determine the dwell time, also called the averaging time. Generally, operators should choose values that create a dwell time of 20 to 30 seconds. The radar profiler can save averaged spectral data in files with the suffix .SPC. The spectral data file also contains complete spectral moments data.

3.4 Moments Calculations

Four quantities are calculated for each set of spectrally averaged data during the frequency-domain processing:
- Doppler shift of the peak;
- Spectral width;
- Noise power; and
- Signal-to-noise ratio.
The result from these calculations is called spectral moments data or just moments, and the profiler can save them in files with the suffix .MOM.

Diagram of a Spectrum

3.5 Consensus Data

The moments and spectra are considered "raw" data. The profiler can also produce wind and/or (with optional RASS) virtual temperature data. The data in this file have been processed with a consensus averaging algorithm. The algorithm uses two evaluation values for each radial direction to determine whether data are valid. One value is a velocity range in which the samples must fit. The second value is a percentage number of samples taken during the consensus period that must fit within that range for the mean consensus value is accepted as valid. Both values are chosen by the operator.

4 Radio Acoustic Sounding System (RASS)

The Radio Acoustic Sounding System (RASS) provides profiles of virtual temperature data. Virtual temperature is a temperature measurement that is uncompensated for humidity or pressure. The RASS system, usually composed of four acoustic sources (one on each side of the profiler antenna) transmits an acoustic wave directed vertically. The profiler uses the acoustic wave as a target, receiving and processing the resulting backscatter and effectively measuring the speed of sound propagation. The profiler can compute virtual temperature profiles because the speed of sound is easily related to air temperature.

Raw temperature data are stored in the moment and spectral data files, but separated from wind data into consensus files with a file name that uses the prefix letter T for temperature (W is for winds).

Page top

5 Wind Profiler Altitude Performance

The minimum and maximum height coverage of the profiler depend on a combination of several factors. Some of these factors are built into the profiler and, therefore, beyond the control of the operator. These factors include:
- The wavelength of the radar
- The size of the antenna
- The amount of transmitted power
- The recovery rate of the electronics

5.1 Wavelength

The wavelength a radar uses depends on the kind of measurements you want to make with the profiler. The radar profiler was designed to provide atmospheric data from the lower atmosphere. A radar profiler that uses a 915-MHz frequency has a 32.8-cm wavelength and the 1290-MHz frequency has a 23.2-cm wavelength. The relatively short wavelength allows a relatively small antenna size.

5.2 Size of the Antenna

The aperture of the antenna, which is controlled largely by practical considerations, determines two other characteristics of the antenna, the beam width and the antenna gain. The beam width, which is inversely proportional to the aperture, is less than 10 degrees in the four-panel Vaisala Windprofiler LAP-3000 configuration. The antenna gain is directly proportional to the aperture. The small aperture of the Vaisala Windprofiler LAP-3000 4-panel wind profiler antenna produces an array gain of about 24 dBi.

5.3 Transmitted Power

The amount of power in the transmit pulse affects the height coverage of the profiler directly. The higher the peak transmit power, the greater the height coverage. The amplifier used in the Vaisala Windprofiler LAP-3000 has a peak output power of about 600 W. The profiler has a duty-cycle capability of 0 to 10%, determined by the parameters chosen by the operators.

So, the first three factors, the wavelength the radar uses, the size of the antenna, and the transmitted power, are part of the system configuration and combine to determine the maximum height coverage of the profiler.

5.4 Recovery Rate of the Electronics

The minimum height coverage is limited by the recovery rate of the profiler electronics and nearby ground clutter. Some of the backscatter energy returns to the profiler antenna during and immediately after the transmit pulse from hard clutter targets. Unfortunately, the early backscatter is often quite strong and overwhelms the receiver's ability to ignore it and find the desirable data. Together, the limits of the receiver's recovery rate, which is fixed by the system configuration, and the nearby ground clutter determine the minimum height coverage of the profiler.

5.5 Pulse Duration

There are, however, two factors that the operator can control that affect the height coverage of the profiler. The first factor that affects the height coverage is the pulse duration, also called the pulse length or the pulse width. Lengthening the pulse raises both the minimum and maximum height coverage. Operators may choose the longest pulse possible to gather data from the highest range gates.

Often, though, we are striving for the highest maximum height coverage, the lowest minimum height coverage, and the smallest range resolution possible. Unfortunately, the pulse duration that the operator selects to achieve the best performance level for maximum height coverage is the worst choice for minimum height coverage and fine range resolution. A short pulse duration is good for the range resolution and minimum height, but is a poor choice for the maximum height.

5.6 Inter-Pulse Period

The other operator-controlled factor affecting range/height coverage, is the inter-pulse period, called the IPP for short. The IPP is the time interval between any two pulses. First glance at the IPP suggests making the IPP as long as possible to give the backscatter from the higher reaches of the atmosphere time to return to the profiler. But in reality, lengthening the IPP decreases the profiler's duty cycle, that is, the amount of time the profiler is transmitting compared to the time the profiler is receiving (listening). Decreasing the duty cycle decreases the average power the profiler is producing, and decreasing the average power reduces the available height range of the profiler.

The Inter-pulse Period and the Pulse Duration

Decreasing the IPP, that is, shortening the time interval between the pulses, raises the profiler's height coverage. However, operators must select an IPP that is not less than the height range they wish to sample. In other words, if the IPP is too short, the height coverage is decreased. Stretching the time between the pulses allows the backscatter from the higher elevations the time necessary to return to the wind profiler.

Choosing an IPP that is too short also has a detrimental effect when the backscatter from a desired height is returning concurrently with the backscatter from a second pulse (or after the profiler has begun sampling data from a second pulse). The backscatter from the second pulse is coming from a lower height than the backscatter from the first pulse. Now the same range gate contains data from two different heights, so data are compromised. This is called range aliasing.

5.7 Pulse Coding

A solution that resolves some of these limitations is pulse coding. Instead of transmitting one simple pulse, the profiler transmits complementary pairs of pulses composed of phase-coded pulse cells connected to each other to form one long pulse. The pulse cells are transmitted in a known sequence and with a known phase relationship to each other. A higher maximum height range is obtained using the longer pulse duration composed of multiple pulse cells rather than an uncoded pulse. A finer range resolution is obtained from the individual phase-coded pulse cells instead of the entire pulse. The operator can choose a pulse duration that has 2, 4, 8, 10, 16, or 32 pulse cells, increasing the pulse duration transmitted by factors of 2, 4, 8, 10, 16, or 32. However, the total pulse duration must not exceed 12 microseconds. The longer pulse duration allows the user to increase the wind profiler's duty cycle, which boosts the maximum height coverage.

Uncoded and Coded Pulse

Because the codes are complementary (in pairs), the operator is required to select values for coherent averaging that are multiples of 2. If the lowest possible height coverage of the profiler is a high priority, then the operator should choose to use an uncoded pulse instead of coded pulse.

Page top

6 Sampling Choices

After operators have chosen the IPP, the number of pulse code cells, and the pulse duration, then the next set of decisions the operators make will affect how the altitude samples are taken.

6.1 Sampling Delay

The sampling delay determines the first height sampled by the profiler. The sampling delay is the time interval between the end of the transmit pulse and the first range gate. This value is corrected in the calculation of height for the instrumental delay inherent in the profiler. The value chosen should not normally result in a first sample height of less than 120 meters.

The Sampling Delay and the Gate Spacing

6.2 Gate Spacing

Gate spacing is the altitude difference between the range gates. Gate spacing is typically the same interval as the transmit pulse. If pulse coding is used, then gate spacing MUST be the same interval as the transmit pulse. Sometimes operators, using an advanced or text editor, may choose a gate spacing that is less than the transmit pulse. This results in a backscatter-weighted average over the volume corresponding to the pulse duration.

6.3 Number of Range Gates

The number of range gates represents the number of range samples collected for each pulse. This value determines the maximum altitude that measurements are collected. Operators should choose a reasonable number that reflects achievable expectations for the atmospheric conditions at the site.

6.4 Interdependency of Parameter Choices

The choice of sampling parameters has many effects on profiler operation. Pulse duration determines range resolution. The IPP and number of coherent averages determines the maximum radial wind velocity that can be unambiguously resolved.

Some parameter relationships are more intricate. For instance, the dwell time is determined by four factors: the IPP, the number of coherent averages, the number of FFT points, and the number of spectral averages. Changing one or a combination of these factors changes the dwell time.

Some parameters are interdependent. The maximum radial velocity is divided by half the length of the FFT to calculate the velocity resolution, which is the value of the velocity range encompassed by each spectral point. Changing any factor that determines the maximum radial velocity also changes the velocity resolution.

In summary, the pulse duration, the recovery rate of the receiver and antenna system, and the effect of ground clutter combine to give the profiler a minimum height coverage of about 100 to 200 meters. The range resolution is, for example, 60 meters for a 400 ns pulse (45 m for a 300 ns pulse), and 400 for a 2800 ns pulse. Several factors, such as the wavelength used by the radar, the size of the antenna, and the average power, set real limits on the maximum height coverage possible. Other factors controlled by the operator, such as the pulse duration and pulse coding, combine to achieve a maximum height coverage of two to five kilometers. This very broad range of maximum obtainable height coverage is largely determined by atmospheric scattering conditions, however.

Page top

7 Height Coverage and Range Resolution for RASS

In normal operation, the profiler gathers wind velocity data for the largest portion of an averaging period and virtual temperature data for the remaining portion. Operating for fifty minutes in wind mode and 10 minutes in the operational RASS mode is a common arrangement. However, the averaging periods may be shorter.

The RASS has the same range resolution and minimum height coverage as the profiler, but the RASS's maximum height coverage is typically 1 to 2 kilometers. Strong winds often transport the acoustic signal away from vertical alignment with the radar antenna beam. Also, the atmosphere absorbs the acoustic signal (largely determined by temperature, humidity and pressure).

All the radar parameters, pulse duration, IPP, coherent average, spectral averaging, sampling delay, and number of range gate, have the same function for RASS as for wind measurement. The exception is pulse coding, which cannot be used for RASS.

The maximum radial velocity must be greater than 380 m/s in order to measure the propagation velocity of the acoustic signal. The number of points used in the Fourier transform is increased to maintain velocity resolution with the increased maximum radial velocity. Typical values for maximum radial velocity are ~ 400 m/s and 2048-point FFT when taking RASS data.

The operator may also choose the separate parts of the spectrum to select acoustic propagation peak and the vertical velocity peaks. The operator must also select the range of audio frequencies that are used. If the operator chooses too large a range, the audio energy is spread over many useless frequencies, reducing the range coverage of RASS. If the operator chooses too small a range, at some altitudes no acoustic propagation peak may be found.

Normal audio frequency values are chosen to accommodate diurnal temperature variation. The range for the acoustic propagation window in the spectrum is typically 2 to 3 Celsius larger than the range of audio sweep.

Page top

8 Atmospheric Effects on Profiler Performance

The range performance of all profilers depends on atmospheric conditions, which can change dramatically and rapidly. An inexperienced operator, when confronted with a reduced height range or large "holes" in the data, may assume that the profiler is not working properly, when in fact the profiler is working correctly, but the atmospheric conditions have changed. The conditions that most affect profiler performance are: humidity, turbulence, precipitation, high winds, and temperature.

8.1 Humidity

The amount of moisture in the atmosphere affects the height range performance of the profiler. Generally, the more moisture there is, the better the profiler works for winds because the air has larger refractive index variations. Often, wind data will show a band of range gates for which the profiler will give no values, even though the profiler is showing data in range gates above and below. This "hole" in the data is often caused by a layer of dry air, which can be verified by examination of corresponding radiosonde data. Because of the dryness of the air, data from profilers located in extremely cold places such as the high latitude regions often have such holes. Marine environments make good profiler sites because of the moisture usually prevalent in those regions.

The RASS benefits from high humidity levels also. When the atmosphere contains more moisture, there is less attenuation (decrease) of the acoustic signal with range.

8.2 Turbulence

The amount of turbulence in the atmosphere also affects the range performance of the profiler. The more turbulence there is in the atmosphere, particularly turbulence with a scale of one-half the profiler wavelength, the better the profiler works. The profiler data has a low signal-to-noise ratio when the air is stable with laminar airflows, a condition commonly found at night. This is because there is little thermally-created turbulence available to reflect the profiler's signals. On the other hand, regions around hills and mountains make good sites because the topography creates turbulence. Convective conditions can produce strong turbulence and correspondingly good profiler data. However, highly localized convective conditions can produce erroneous data. Turbulence is also beneficial to RASS operation. Turbulence helps distribute the acoustic wavefront, helping increase the range in the presence of winds. The wavelength of the acoustic signal must be half that of the radar signal in order to measure the velocity of propagation of the acoustic signal.

8.3 Precipitation

Most types of precipitation such as rain, snow, and hail can affect the performance of the profiler. When precipitation moves in a direction that's different from the air around it, the vertical beam measures the movement of the hydrometeors rather than the vertical component of the wind. This is because precipitation returns stronger signals than clear air. The precipitation occurring during a thunderstorm can overwhelm the data collected from the regions above the storm, creating a "shadow" in the data. However, if the precipitation is carried with the wind, then the horizontal winds might still be measured because the particle velocity in the off-zenith beam can be corrected with the vertical beam measurement, assuming spatial homogeneity.

For RASS, virtual temperature measurements are usually poor quality during precipitation. During precipitation, if the hydrometeor fall velocity is measured and it differs from the vertical wind velocity, the resulting virtual temperature measurements will be incorrect.

8.4 High Winds

Ground clutter most often affects the quality of data in the lower range gates. High winds can cause clutter signals from objects such as trees and power lines to exhibit sufficient Doppler velocity width that the profiler's ability to screen out this clutter is overwhelmed. Choosing sites with minimal ground clutter will improve the range and data quality of the profiler.

High winds can adversely effect the RASS virtual temperature measurement in two ways. Increased ground clutter can create incorrect vertical velocity values used for temperature correction. High winds may also reduce the range of measurement of RASS by displacing the acoustic signal away from the radar beam.

8.5 Temperature

Temperature has more of an effect on RASS than on the profiler's wind measurement. Acoustic attenuation varies as a function of temperature, humidity, and pressure. Cold dry air exhibits highest attenuation, which can exceed -40 dB per kilometer. Very moist or warm air propagates acoustic signals better, resulting in improved range for virtual temperature measurement.

Page top