Campbell CSAT3B, Инструкция по эксплуатации, Страница: 57 / 88

CSAT3B Three-Dimensional Sonic Anemometer
analyzed, the user should choose a filter bandwidth that is at the Nyquist
frequency, which is half the data output rate from the CSAT3B to the
datalogger or PC. This will eliminate aliasing or the mismatch of signals to
frequencies. The bandwidths that are available for selection are 5, 10, 20, and
25 Hz. If no filter is used as in Mode 3, the effective bandwidth will be 50 Hz,
since the bandwidth is limited by the Nyquist frequency of the fundamental
measurement rate of 100 Hz. Each bandwidth, including the effective 50 Hz
bandwidth from unfiltered data, results in a different time delay between
measurement and output. TABLE 8-2 summarizes the time delays for each
filter. These time delays will need to be accounted for when aligning CSAT3B
data with data from other fast-response sensors. Data alignment can be done
automatically with a datalogger or during post-processing on a PC. TABLE
8-2 also shows the closest integer sample delays for various output rates.
If the primary interest of a study is fluxes rather than spectral analysis, there is
no need to select a filter. With unfiltered data, the anemometer’s high
frequency response is only limited by the geometry of the anemometer head,
and in the case of self-triggered measurements, a 50 Hz effective bandwidth.
This minimizes the possibility of underestimating high-frequency signal
variations and covariations. This approach, however, aliases high-frequency
information to lower frequencies. This aliasing is apparent in spectra from an
upwards tail, when compared to the
−
5/3 power relationship (Kaimal and
Finnigan), at frequencies approaching the Nyquist frequency of one-half the
sample rate. This aliasing does not compromise the variances and covariances
(and therefore, fluxes) computed from aliased data. The variance and
covariance calculations are not frequency dependent, they simply measure a
signal’s total variation or total covariation from two signals, respectively.
Given this, under circumstances where measuring fluxes are the primary
interest, it is recommended to operate without a filter and, if possible, have the
datalogger provide the trigger.
TABLE 8-2. Time Delays by Mode and Filter
Mode
Delay between Measurement
and Output
1 1 Datalogger Scan Interval
3 10 ms
2 and 4
Ba
nd
w
idt
h
5 Hz 795 ms
10 Hz 395 ms
20 Hz 195 ms
25 Hz 155 ms
8.2.3 Data Output
After a measurement is triggered and optionally run through a filter, it is stored
in the CSAT3B data buffer until it is output to either a datalogger or a PC.
Mode 1
In the case where a datalogger provides the measurement trigger (Mode 1), the
trigger is issued each time the datalogger goes through a scan of the program
and executes the
CSAT3B() CRBasic
instruction. See Section 8.4.1, CRBasic
47