Sa. Cohn et al., Radial velocity and wind measurement with NIMA-NWCA: Comparisons with human estimation and aircraft measurements, J APPL MET, 40(4), 2001, pp. 704-719
The National Center for Atmospheric Research (NCAR) Improved Moments Algori
thm (NIMA) calculates the first and second moments (radial velocity and spe
ctral width) of wind-profiler Doppler spectra and provides an evaluation of
confidence in these calculations. The first moments and their confidences
are used by the NCAR Winds And Confidence Algorithm (NWCA), to estimate the
horizontal wind. NIMA-NWCA has been used for several years in a real-time
application for three wind profilers in Juneau, Alaska. This paper presents
results of an effort to evaluate the first moments produced by NIMA and ho
rizontal winds produced by NIMA-NWCA through comparison with estimates from
"human experts'' and also presents a comparison of NIMA-NWCA winds with in
situ aircraft measurements. NIMA uses fuzzy logic to separate the atmosphe
ric component of Doppler spectra from ground clutter and other sources of i
nterference. The fuzzy logic rules are based on similar features humans con
sider when identifying atmospheric and contamination signals in Doppler spe
ctra. Furthermore, NIMA attempts to mimic the human experts' assignment of
confidence to the moments. A Human Moment Analysis (HMA) tool was developed
to assist the human experts in quantifying moments. This tool is described
and a methodology of tuning NIMA rules based on human truth specification
is presented. NIMA performed well on a dataset specifically chosen to be di
fficult. The average absolute error between the HMA estimate and NIMA-deriv
ed radial wind estimate was slightly more than 0.3 m s(-1) when data with l
ow NIMA confidence were excluded, which is comparable to the Doppler spectr
um resolution. The correlation between winds derived from NIMA-NWCA and fro
m HMA first-moment estimates exceeded 0.96 when the data with low NWCA conf
idence were excluded. The correlation coefficient between NIMA winds and in
situ measurements by aircraft was 0.93 when aircraft winds that were belie
ved to be accurate were used.