NASA Technical Reports Server (NTRS) 19980017284: Rapid Temporal Changes of Midtropospheric Winds

12 Pages·0.59 MB·English

by  NASA Technical Reports Server (NTRS)

'_//_S_'_/7"-/'_ - c_ _ -_ 20"7069
RAPID TEMPORAL CHANGES OF MIDTROPOSPHERIC WINDS
./>5/3 /,,,J_4A
/
O_> 7 (cid:0)LIT
FRANCIS J. MERCERET
Printed in d_eUnited States ofAmerica
Reprinted from JOURNAOLFAPPLIEDMETEOROLOGY
Vol. 36, No. 11,November 1997
© 1997 American Meteorological Society

NOVEM1I_9E97R MERCERET 1567
Rapid Temporal Changes of Midtropospheric Winds
FRANC1SJ. MERCERET
Applied Meteorology Unit,NASA(cid:0)Kennedy Space Center, Horida
IManuscriptreceived 7October 1996,illlinal1-ornl17April 1997)
ABSTRACT
The statistical distribution of the magnitude of the vector v_'indchange over 0,25-. I-, 2-, and 4-h periods
basedondata from October 1995through March 1996overcentral Florida ispresented, Thewindchanges at
altitudesfrom6to 17kmweremeasured usingthe KennedySpaceCenter50-MHzDopplerradar windprofiler.
Qualitycontrolled profileswereproduced every 5rain for 112gates,each representing 150mmaltitude.Gates
28through 100_ere selected foranalysis becauseoftheir significance toascending spacelaunch vehicles.The
distribution waslk)undtobeIognormal.Theparameters of thelognormal distribution depends_stematically on
the time interval. This dependence is consistent with the behavior of structure functions inthe l'''' spectral
regime. There is a small difference between the 1995data and thc 1996data, which may reprcsent a weak
seasonaleffect.
I. Introduction Wind profiles were collected every 5 min over a pe-
riod of 117 days from 29 September 1995 through 26
This study was motivated by the need to quantify the
March 1996 except when the DRWP was down for
benefit of continuing to operate the Kennedy Space Cen-
maintenance or repair. The data were extensively quality
ter (KSC) 50-MHz Doppler radar wind profiler
controlled. Vector wind differences were computed at
(DRWP). In its operational support configuration, the
each level for the target time intervals (AT) for each
instrument measures wind speed and direction at 150-m
profile using only data that passed the quality control
intervals from 2011 to 18661 m. The data are currently
(QC) screening. More than 25000 profiles were ac-
used to evaluate wind persistence between the last pre-
cepted for analysis. The first four moments, the prob-
cision wind sounding balloon ("Jimsphere"), about l-h
ability density, and the cumulative probability distri-
before launch, and launch of a space shuttle. Other
bution of the vector magnitude were computed at each
launch vehicles may soon begin using the DRWP in a
level for each AT.
similar manner.
The benefit of the DRWP is avoidance of the risk This paper presents the details of how the study was
conducted and the results showing the statistical distri-
associated with undetected wind changes large enough
butions of the vector wind changes as a function of AT.
to threaten the vehicle or the mission. To assist in as-
Results include statistics for the distribution at any one
sessing this risk, the statistical distribution of the vector
gate in the target region and also for the maximum
magnitudes of 0.25-, 1-, 2-, and 4-h wind changes at
vector wind change magnitude in the entire region.
altitudes between 6 and 17 km was determined over a
6-month period during the winter season. The 6-17-km
region is the region in which large launch vehicles are 2. The KSC 50-MHz DRWP
most sensitive to wind changes. Winter is the season in
which the largest short-term temporal changes in winds The data used for this study were collected using the
at the target altitudes over Florida are expected. Wind 50-MHz DRWP located near the Space Shuttle Landing
changes rather than wind shear were examined because Facility (SLF). It is a hardware twin to the one at White
space vehicle load programs calculate loads based on Sands Missile Range described by Nastrom and Eaton
the wind profiles, and it is the deviation from the ex- (1995). The consensus software supplied with the orig-
pected wind profile that poses the hazard regardless of inal profiler was replaced by a "first-guess median fil-
whether large shears are involved. ter" algorithm developed at Marshall Spaceflight Center
(Wilfong et al. 1993) that also provides capability for
real-time interactive QC of the data. Details are provided
in Schumann et al. (1993).
Corre.wonding author address."Dr.FrancisJ. Merceret, Applied
Profiles are generated every 5 min. Each profile con-
Meteorology Unit--NASA, MailCode PH-B3,Kennedy SpaceCen-
ter,FL32899. tains data at each of 112 range gates that represent a
E-mail:[email protected] slice of the atmosphere 150 m thick. Gate 1 is at a
1568 JOURNAl. OF APPLIED METEOROI. OGY V(}it:Mtc36
nominal altitude of 2011 m while gate 112 is at 18 661 TAm.I_: 1. Quality control [lag thresholds.
m. The variables measured at each gate include the sig-
Threshold
nal power, noise power, Doppler return frequency, and What causes flag value
spectral width for each of the two oblique beams and
Excessive shear internally computed by
the vertical beam. These data are converted into esti-
DRWP O.I s
mates at each gate of the horizontal wind direction and Excessive vertical speed 3 m s '
speed, the vertical speed, and the signal-to-noise ratio Excessive spectral width 3 m s
in each beam. Exceeds first-guess propagation limit 6
Missing profile--data place holders 1999) N/A
Fails small median test See text
3. The data and its analysis Excessive adiacent gate directional shear 30 pet gate
Excessive adiacent gate speed shear 7 m s t l-mrgate
a. The data sample Insufficient signal-to noise ratio (SNR) 15dB
Fails manual QC See text
At each of the i 12range gates, the KSC DRWP pro-
duces 288 wind speed and direction values daily. Data
were collected for this study from 29 September 1995
through 26 March 1996. Excluding days on which the The vertical speed threshold is used to identify pre-
DRWP was inoperative, there were 61 days in 1995 and cipitation. Heavy rain contaminates the radar signal and
56 days in 1996 that yielded 117 days in the sample. produces bad winds. Typically, heavy rain produces a
Some of these days were incomplete due to profiler large vertical wind indication based on the radar track-
outages, and the QC process removed additional data. ing the falling raindrops.
Since the QC was done gate by gate, the sample size The shear value computed internally by the radar soft-
varies; but it is typically 25 000 to 28 000 per gate. ware provides a coarse warning that the measurements
When data are combined for gates 28-100 (6-17 km), are invalid and would usually not be triggered except
the resulting sample size approaches two million. by an aircraft in the beam or a major failure in the
The time available for this study was limited and a system. The more stringent speed and direction shear
full year of data could not be collected or processed. tests are designed to eliminate meteorologically un-
The upper air climatology of central Florida is such that realistic winds produced by sidelobe or interference sig-
the winter season brings stronger winds aloft, including nals and to flag those signals.
The small median test is the one described in detail
meandering jet cores with sharp spatial gradients. This
season was selected for the study as a "'worst case" in by Carr et al. (1995). It examines not only continuity
terms of the risk to space launch vehicles. Obviously, of the winds in the vertical but temporal continuity as
this is a worst case only in the sense of being worse well. The test requires threshold values be designated
than the summer season. It is unlikely to be the worst for three different heights. This work used 5.7 m s ' at
2 km, 10.2 m s nat 9 km, and 8.4 m s ' at 16 km based
case over a period of years.
To estimate the strength of seasonal variation, the on about 20 years of local windsonde data. These criteria
sample was nearly evenly divided, with the first half are more restrictive than those used by Carr et al. (1995).
from 1995 and the second from 1996. Typically, central On many of the days during the experimental period,
Florida winds aloft and their gradients are somewhat personnel of the Applied Meteorology Unit (AMU) ex-
stronger in the January through March period than in amined the spectra in real time and made notes of any
October through December. The mean layer wind speeds obvious sidelobe or interference signals that could ad-
for gates 28-100 followed the typical pattern. For 29 versely affect the data. When the Titan program con-
September through 31 December 1995 the dataset gave ducted real or simulated launch operations, the data were
amean speed of 23.19 m s _.For the data from 1Jan- not only monitored but were manually quality controlled
uary through 26 March 1996 the result was 35.46 m before transmission to the data system. In these cases,
s _. The respective standard deviations (m s ') were the real-time manual QC operator made extensive notes.
Both the AMU notes and the Titan notes were used to
13.66 and 14.31. The sample size for both sets was near
a million, and both of the differences are statistically assist the after-the-fact manual QC process described
significant at the 0.005 level. Comparison of the 1995 next.
and 1996 results with each other and with the combined After the automated QC was run, the file was dis-
1995-96 data was used to examine whether a seasonal played graphically. Wind speed and wind direction were
effect was significant during these months. separately but simultaneously presented as time-height
diagrams in which each range gate at each 5-rain in-
terval was plotted as a colored point. The resulting color
b. The qualio' control (QC) process
contour map permitted visual detection of wind speed
Each range gate in each profile constitutes one record or wind direction changes and the identification of nat-
in the daily data files. Each record includes a QC flag ural features like jet streaks and artifacts like sidelobes.
to track individual quality control tests. The active flag AMU and Titan QC logs were used for guidance.
bits are listed below in Table 1. During visual inspection of the data, the time and al-
No\q_Mm_r 1997 M E RC E R ET 1569
45 essary despite the loss of information about the variability
of the results with height. The height levels from gate
4 •
28 to gate 100 were selected for their relevance to as-
35i cending launch vehicles. In addition, the maximum value
w
for the magnitude of the vector wind change anywhere
-_ 25, in the target region was computed for each profile. The
2 I i same statistics were computed for this dataset.
® i A combination of gates 32-34 (a shuttle's transonic
"_ i5• region) was created to permit "quick-look" examination
of the extent to which vertical inhomogeneity might be
a concern. The mean wind speed in this layer averaged
05- 30% lower than in the layer from gate 28 to gate 100.
0 ....... --From MomentsI The wind difference data from the two regions differed
' • Obsolved by an amount that was statistically significant due to the
-05
! 10 10Q enormous sample sizes involved, but the differences ap-
Magnilude of Vector Velocity Change (m $-1)
peared insignificant to the author from the point of view
Ftc,,. 1.The cumulative probability (vertical axis) of the magnitude of risk assessment. After determining that the distribution
of the I-h v,'ind change at an),' level in the region fron_ gate 28 to
of these data had roughly the same shape and parameter
gate I00 inclusive exceeding the specitied (horizontal axisl value.
values as those in the thicker layer, detailed analysis of
Probabilities are expressed as norlnalized standard deviations : (e.g.,
a probability of 0.5 equates to c = O. 0.8413 equates to :: - I). The this dataset was deferred due to limited resources. Lim-
points are the measured cumulative distribution. The line is the lug- ited resources also prevented any attempt to stratify the
normal distribution derived from the moments as described in the
data temporally to look for diurnal effects.
text.
Examination of the statistical results from the first two
days of the September 1995 data suggested that the data
titude of any apparent sidelobe or interference signatures were distributed lognormally. The experimental design
were noted. Data that appeared to be unreliable were was enhanced to include generation of six estimates of
manually flagged by setting a QC flag bit. the parameters of a lognormal distribution from the four
Nearly all of the manual QC consisted of flagging the moments by combining them in pairs. Each pair of mo-
interior (in time-height space) of sidelobe and interfer- ments generates an estimate for the two lognormal pa-
ence signatures whose boundaries were flagged by the rameters as shown below. The mutual consistency of
automated process. Of the 117 days in the sample, 44 these estimates served as an indicator that the distribu-
required some manual flagging. Less than 1% of those tions really were lognormal. This was confirmed by using
data were manually flagged. the mean of the six estimates of each parameter to gen-
Once a data file had been accepted, the delta files erate a straight line on a log probability plot on which
were generated and examined visually. Delta files are the actual distribution was overlaid for comparison.
identical to the data files except that they contain the The formula for the nth central moment of alognormal
wind changes rather than the winds. Sometimes a side- distribution is given by Aitchison and Brown (1966) as
lobe or interference signal stood out better in the wind
M" = exp(n/x + n'_tr'-/2), (1)
changes than in the winds themselves. When this oc-
curred, the process described above was repeated. When
where /.L is the mean of the logarithms and tr is the
the data file had been reaccepted, the delta files were standard deviation of the logarithms of the lognormal
rerun and reexamined. The data were used for analysis
variable. For any two values of n for which M" are
only when the basic data and the delta files had been known, the result is a pair of simultaneous linear al-
examined and accepted. During analysis, any record for
gebraic equations for /x and tr _,which may be solved
which the QC flag is nonzero was ignored.
readily.
c. The analysis scheme
4. The results
The initial experimental design called for computing
the first four statistical moments (mean, variance, skew- This section presents the analysis showing that the
ness, and kurtosis) and the probability density and dis- magnitude of the vector wind change is lognormally
tribution of the magnitude of the vector wind change at distributed. The variation of the parameters of the log-
each level in the region of interest. It became quickly normal distribution with the length of the time interval
obvious that this was impractical. Greater sample sizes over which the change takes place is also presented. A
with all of their advantages for statistical analysis could brief discussion of possible seasonal variation in the
be obtained by combining data from multiple levels while distribution is included. Finally, the implications of
simultaneously reducing the information output to aman- these results for spaceflight operations and for the design
ageable size. Limited resources made this strategy nec- of profiler networks are discussed.
1570 JOURNAl.OFAPPLIEDMETEOROOI.GY Vot.trM36:*:
contained 1 882 502 data points. To the extent that the
i!iI i wind differences may be related to wind variances, these
3 _i _ _ri,_!¸! .,/' ! i, ,, results agree with those of Nastrom and Eaton (1995),
who found the variances over a 1-h period at similar
altitudes were lognormally distributed.
Figure 2 shows the same information as Fig. 1except
'eI i I i a,ji,./z ! i that the data are the maximum values of the magnitude
of the vector wind change in gates 28-100 for each
:ii
profile rather than the individual gate values. The sample
J°i size is 25 983. This figure shows that the maximum
N fi
values in the entire region are also lognormally distrib-
uted.
The data from 1995 and 1996 separately and the in-
--From Mornen_s
-2L ; • _ I L• =ot:_,-,_ _!i dividual monthly data behave in the same manner, as
1 1o lOO do the data for time periods of 0.25, 2, and 4 h. Figures
MagnitudeofVKtor Volo¢ltyCbarge[ms-I) 1 and 2 above are typical of any of these data subsets.
FiG.2.Thecumulative probability of the largest single value of
the l-hwindchange inthe regionfromgate 28togate 100inclusive
exceeding the specified value.The format isthesame as inFig. I. b. The parameters of the lognormal distribution
depend on AT
a. The magnitude of the vector wind change is
The two parameters of the lognormal distribution,/x
/ognormally distributed
and or,were computed for velocity magnitudes in meters
Figure 1 shows the cumulative distribution of the per second. Figure 3 shows the variation of # with the
combined gates 28-100 data for the entire sample of time over which the changes are measured (AT) for the
1-h wind changes. The individually plotted points are gates 28-100 combined data. Figure 4 shows the vari-
the actual measured distribution. The straight line rep- ation of _ for the layer maxima. Despite the significant
resents a pure lognormal distribution with parameters scatter, a clear, roughly linear relation between the value
equal to the average of those derived from the six pair- of/z and the logarithm of AT is apparent in both cases.
wise combinations of the first four moments of the mea- Linear regressions of /2,versus In(AT) yield r-_> 0.9
sured distribution. The horizontal axis is the magnitude (significant to <0.005) in all cases.
of the vector wind change over a 1-h period. This sample Figures 5 and 6 present the same information for o'.
25
2
X
= "-'_" _" i * sep-95]
1 ,,-- _ • Oct-95 /
I .ov.95i
X Dec-95 I
• X Jan-96 1
"__t,,,I ! •+ FMeabr--9966 I
0.5 ................. 1995 II
X_ _ " " - 1996
_AH
0
01 1 10
Delta T(hrl
FIG.3.Therelationship between the mean(#) of thenatural logarithmof the magnitude of
the vector windchange (m s _)change atany heightlevel inthe region from gate28to gate
100inclusive andthe timeinterval(AT) overwhichthechangetakesplace. Thelinesrepresent
combined data fromSeptember-December 1995,January-March 1996.andthe entire dataset.
The points represent individual months toshow the scatter.
NO'¢t_MBI-R 1997 M E R C E R E T 1571
2.5
1.5
Nov-95
;< Dec-95
X Jan-96
• Feb-96
+ Mar-96
-- --1995
0.5
- - - 1996
--All
0
0.1 10
DeltaT(hr)
Ft(;. 4. The relationship between the mean (/.t) of the natural logarithm of the magnitude ofthe
largest single vector wind change (ms _)change in the region from gate 28 to gate 100 inclusive
and the time interval (AT) over which the change takes place. The format is the same as ill Fig. 3.
Although the scatter is much larger, a similar relation AT between 0.25 and 4 h. They can also be used with
with opposite sign is suggested. Linear regressions of appropriate caution to extrapolate these results to larger
or versus In(AT) yield r: > 0.4 (also significant to and smaller time intervals. Extrapolation to larger time
<0.005) in all cases. intervals is less justified because semidiurnal, diurnal,
The relationships between the lognormal parameters and synoptic features should begin to dominate sto-
and AT can be used to interpolate these results to any chastic ones at these scales.
0.8
0.7
0,6
I •
I
0.fi - T
!
_0.4 r ....
_3 i
• SeP'gSil
• Oct-95 l
0.3
Nov-95 " I
×X DJaenc--9965'i I
0.2 • Feb-96 -_
! + Mar'961 I
----1996 ]
--- 1996 ] 1
01 !
--All I 1
0.1 1 10
DeltaT(hr)
F[G. 5. The relationship between the standard deviation (_r) of the natural logarithm of the
magnitude of the vector wind change (ms ') change at any height level in the region from gate
28 to gate 100 inclusive and the time interval (AT) over which the change takes place. The
fl)rmat is the same as in Fig. 3.
1572 JOURNAL OF APPLIED METEOROLOGY V_l.t,Ml.:36
0.8 J
07
O6
m 04
J • • Sep-95
I Ocl-9S
03 i Nov-95
Dec-95
:_ Jan-96
02 • Feb-96
i Mar-g6 i
m m1995
0 1 " - - 1996
L_Arl
0 i ........... _.......................... ,
01 1 10
Delta T(hr)
F[(_. 6. The relationship between the standard deviation [_r) of the natural logarithm of the
magnitude of the largest single vector v,ind change (ms _1change in the region frmn gate 28
to gate 100 inclusive and the time interval (AT) over which the change takes place. The format
is the same as in Fig. 3.
c. The implications of the distribution normally distributed with twice the mean and twice the
standard deviation of the distribution of the log of the
Examination of the previous figures suggests that part
differences. If the parameters of the distribution are
of the scatter may be due to secular changes in the
known as a function of temporal separation, then the
environment during the experimental period. The/x and
mean value of the squared differences can be calculated
o"values tend to change in opposite directions with AT.
as afunction of the temporal separation. Since the whole
The /z values for 1996 were slightly higher than for
vector magnitude was used in this study rather than any
1995, but the o-values were lower, thus maintaining the
single component, the result will be a "magnitude struc-
same negative correlation. Although the scatter is large,
ture function" equal to the sum of the three component
a t test shows the differences in /._are statistically sig-
structure functions.
nificant at the 0.001 level, and a chi-squared test shows
The small number of independent estimates of the
that the differences in o-are significant at the 0.005 level.
behavior of the distribution parameters with temporal
In any case. the secular variation is not large compared
with the scatter and it remains to be seen whether it is separation and the scatter in the data suggest that pre-
cision calculations are unwarranted. To compare these
meteorologically significant. These effects, whatever
results with those published in the literature, the vari-
their cause, also naturally appear in the probability of
ation of/z and o-with AT was approximated by astraight
exceedance curves presented in the next section.
line fit through the 0.25- and 4-h values for the com-
The [ognormal distribution of the magnitude of the
bined dataset. This resulted in linear equations for #
vector wind changes facilitates comparison of these re-
and o-as follows:
suits with atmospheric structure functions. Structure
functions for wind are usually defined for each velocity 2/.z = a, + b,, lnAt (4)
component in terms of spatial separation by or in terms
of temporal separation by 2o"= a,, + b,, lnAt, (5)
where the factor of 2 converts the fit for differences to
D,,(r) = ([u,(x) - uL_t + r)]:) (2)
one for squares of differences. Time differences in sec-
or in terms of temporal separation by onds were used for dimensional consistency.
D.('r) = ([u,(t) - u,(t + r)]-'). (3) The resulting mean value of the squared differences,
D(At), may be reduced to the form
where the brackets denote the average and the subscript
D(At) = C_At :At '_"'_", (6)
i denotes the ith component (Stull 1989).
If the logarithm of the wind differences is normally where C_ = exp(a_, + 0.5a_,), N = b, - a,,b,,, and k =
distributed, then the logarithm of their square is also 0.5b_,. When the data from Figs. 3 and 5 are used to
Novt.;_,II_I_R 1997 M E R C E R ET 1573
..........!....................
01
.
0.01
-8
o•
l,U
,_ 0.001
I
i35 - - - 15 Min (all) 1
I'-- --1 Hour (aft) f
_. 0.0001 _4 Hours (all) ]
X 15 Min (1995) /
• 1Hour (1995) !
- 4 Hours (1995)'
000001 ._..!. 115HouMrin(1(919969)6) i
4 HotJrs (1996)
i
i
i
0.000001 I
10 100
Delta Vmag (m/s)
FIe;. 7. The probability (vertical axis) of the magnitude of the l-h wind change at any height
level ill the region from gate 28 to gate 100 inclnsivc exceeding the specified (horizontal axis)
value for various values of AT. The lines are results from lbe enlire dalasel. The points separately
present the 1995 and 1996 data to shov,, the varialion from late fall to early spring.
generate the constants in the linear equations given the -5/3 power law is followed fairly well. The data
above, the result is C_ = 0.01, N = 0.8, and k = 0.001 here fall within those scales, and the slope of the struc-
to one significant figure. ture function is consistent with that spectral behavior.
Over the range from 0.25- to 4-h, the term containing
k varies from near unity to l.l 7. If we treat this term
d. Significance for spaceflight operations and the
as approximately constant, the result resembles that
design of upper-air networks
widely reported in the literature for structure functions
in the inertial subrange if the assumption is made that Using the lognormal model with parameter values
spatial and temporal variations may be treated similarly based on the data collected in this study, we have gen-
using a Taylor's hypothesis. The spatial structure func- erated the required curves for 0.25-, 1-, and 4-h inter-
tion under these conditions is given by vals. The data for the 2-h interval were also generated,
but are not plotted in the figures to enhance legibility.
D(r) = Ce "-i_r_.t_, (7)
Figure 7 presents the curves for probability of exceed-
where C is a constant of order unity (Ottersten 1969). ance at any preselected gate from 28 to 100. The vertical
The value of N obtained here is within 20% of the axis is the probability that the magnitude of the vector
inertial subrange value. Moreover, the value of C, ob- wind change over the specified time period will exceed
tained here implies that e is of the order of 0.001 m-_ the corresponding value on the horizontal axis. Figure
s ', which compares well with values obtained in the 8 presents the probability that the threshold will be ex-
free atmosphere from aircraft observations (Merceret ceeded anywhere in the entire layer.
1976: Vinnichenko and Dutton 1969), although up to For spaceflight operations, the benefit of rapid access
two orders of magnitude smaller values have been re- to wind profile data is the avoidance of unanticipated
ported in the upper troposphere and lower stratosphere wind changes, which pose the threat of destruction of
(Vinnichenko 1970). Significantly, the data obtained a launch vehicle (Mecbam 1995). For assessing the haz-
here were taken during the winter where both the wind ards to space launch operations, engineers must first
speed and wind shear are greatest, suggesting the largest determine what wind change between the last balloon
degree of production and dissipation of mechanically and launch poses an operational threat to the launch
generated turbulence. vehicle. The curves presented here can then be used to
No spectral analysis has yet been undertaken with quantify the likelihood of that threat. If the risk is un-
these data, but Vinnichenko (1970) concluded that in acceptably high, then availability of day of launch
the mesoscale (periods from 1 day to about l0 min), DRWP capability should be a mission requirement. A
1574
JOURNAl. OF APPI. IED METEOROLOGY VoI.I!MF. 36
1
0.1 T• 1 _I i " 1
4. "k
c
\
wl
"a OOOl
,,_ _ m 1Hour (all) i . i_a ]
"o_ --4 Hours (all) I
_- 0.0001 x 15 Min (1995) ] l '_• r ' ::
• 1Hour (1995)
4Hours (1995) ,t:_
15 Min (1996) , _ . [
000001 1Hour (1996) , %_.
4Hours (1996) : "_,t,_
0.000001 i ..... i '
10 100
Delta Vma9 (m/s)
Fro. 8. The probability (vertical axis) of the magnitude of the largest -h wind change in the
region from gate 28 to gate I00 inclusive exceeding the specified (horizontal axis) value for various
values of _T. The lk_rmat is the same as in Fig. 7.
larger dataset and additional analysis could be used to can also be used to assist m the determination of whether
stratify the risk by altitude and season. the cost of operating a profiler is warranted for a par-
For design of upper-air networks, the required tem- ticular application given a specified acceptable risk
poral sampling rate depends on the need to assume that level.
significant features are observed and the data are rep-
resentative (Douglas and Stensrud 1996). Sampling
Acknowledgments. The author thanks Susan DeRussy
rates for current profiler networks are typically one per
of Computer Sciences Raytheon for saving the data used
hour based on consensus averages of l0 or more profiles
(NOAA 1994: Spencer et al. 1996). These results allow in this study. Robin Schumann, Ann Yersavich, and
Winnie Lambert of ENSCO, Inc. (AMU), provided the
quantitative assessment of the wind changes that might
daily logs of profiler performance. Ms. Schumann also
be missed with sampling or reporting intervals greater
wrote the software that transferred the MIDDS data from
than a specified time. This information can be used to
magnetic tape to PC readable diskette and confirmed
optimize the trade-off between cost and complexity of
file integrity and format. Ms. Yersavich ran the software
data collection and the accuracy and completeness of
the data. to transfer and process the data to readable diskette.
Captain Scot Heckman of the 45th Weather Squadron
performed some of the real-time QC for Titan opera-
5. Summary and conclusions tions.
An examination of six months of winter season The author appreciates the contribution of Christine
50-MHz DRWP data over central Florida has shown Boykin at Johnson Space Center to the choice of the
that the magnitude of vector wind changes for periods height levels and time periods selected for study. Greg-
from 0.25 to 4 h is iognormally distributed. The log- ory Taylor of ENSCO, Inc. (AMU), directed my atten-
normal parameters vary systematically and simply with tion to the paper of Carr et al., which contained the
the time interval. small median test used herein. Wade Batts of Marshall
Probability of exceedance curves for wind changes Spaceflight Center supplied the climatological data used
at any single 150-m gate in the region between 6 and to select the small median parameters. Gregory Taylor
17 km and also for the maxima within the entire region and Timothy Wilfong provided thorough reviews of the
were also generated. These curves can be used to assess first draft of this paper. Their thoughtful comments made
the risk to space launch programs of short period wind this a better paper as did those of several anonymous
changes and to assist in determining the required tem- reviewers. The author appreciates the assistance of Ann
poral resolution for proposed profiler networks. They Yersavich, who prepared the manuscript for publication.