Auusr 1965
WILLIAM M. GRAY
Calculations of Cumulus Vertical Draft Velocities in Hurricanes from
Aircraft Observations
Waa M. Gray
Colorado State University, Fort Collins
(Manuscript received 24 December 196, in revised form 10 June 1965)
ABSTRACT.
In cumulus convective atmospheres the most significant length scale of vertical motion is of cumulus draft
‘ize—of horizontal lengths two to four orders of magnitude less than the synoptic scale and of magnitude
fone to three orders greater. This draft motion on seales between the gust and synoptic motion has been
least investigated. The purpose ofthis paper isto propose a new method of calculating draft velocities from
aircraft and present results of calculations with this method in hurricanes.
"An aircraft Doppler radio navigation instrument which s capable of meseuring horizontal wind variations
to-a space resolution ofa fraction of a nautical mile ean be employed with other instruments to measure an
aleraft's piteh angle changes. From this determination, along with other standard aircraft measurements
suchas radar and pressure altitude, power setting, etc, itis possible to make determinations of average ver-
tical air motion to space reslutions of 0.4 to 0.7 nautial mile,
Calculations of vertical motion are performed along a number of radial leg aight tracks flowa by the
National Hurricane Research Project B-80 aireraft during the 1058 season with the above method. Typical
draft velocity was 5-15 knots andl draft widths 1-3 nautical miles. Maximum derived gust velocities within
‘the drafts were typically inthe range of §-15 knots, Comparison of results with those of the Thunderstorm
‘Project is made. Vertical accelerometer data are also presented
463
1, Background and purpose
Most_ meteorologists have realized that in cumulus
‘convective atmospheres the most significant length scale
of vertical motion is of cumulus draft size. Due to the
lack of proper instruments for accurate measurements,
this scale of vertical motion has been least investigated.
But from the meteorological point of view this scale of
vertical motion may be most responsible for the im-
Portant vertical mass, energy, and momentum trans-
Ports occurring in the convective atmosphere (Riehl
and Malkus, 1958; Rich], 1961; Malkus, 1961).
‘The Thunderstorm Project’ (Byers and Braham,
1949) has been the only systematic investigation of
draft scale vertical motion. Vertical draft velocities
‘were approximated from aircraft altitude changes under
specific fight conditions, Few direct measurements or
calculations of vertical motion have been made over
‘oceans or in storm systems such as the hurricane; yet
the importance of cumulus convection in the hurricane
far surpasses that of any other synopticscale storm
system,
Beginning in 1935, the U. S. Weather Bureau began
instrumentation of two Air Force B-50 and one B47 air-
craft to investigate the meteorological parameters of
the hurricane, Recent instrument developments such
as the Doppler radio navigation system make this the
first systematic investigation of the meso- and micro-
scales of motion in which accurate detailed wind ob-
servations are obtainable. The hurricane radial leg
penetrations which were flown (Fig. 1 is typical) by
the National Hurricane Research Project during the
1957-38 seasons were mostly executed at constant
LEO 18 AUS
560 mb
kc .
‘Fc. 1. Typical hurricane fight track.464 JOURNAL OF APPL
power setting and heading, Recordings of rpm and mani-
fold pressure of each B-50 engine are available from
photopanel data recorded every two to five seconds
(usually every two seconds near the storm center).
‘These recordings give the power changes. Aircraft
heading and indicated airspeed (LAS) were also re-
corded at the same_time intervals on the photopanel
Engincer’s Hight logs of fuel burnoft are available every”
hour,
Wind velocities were computed and recorded by an
reforence line _
1ED METEOROLOGY Voir &
ASN.6 analogue computer and a General Precision
‘Laboratory AN/APN-82 radio navigation system which
‘employed the Doppler shift principle. Radar Altitude
(RA) and Pressure Altitude (PA) were also recorded,
As observations are taken over water, accurate RA
‘measurements aire possible—unhindered by irregular
‘topogtaphic variations experienced over land, National
Aeronautics and Space Administration's Velocity,
Gravity, Height (VGH) continuous recordings (Rich-
ardson, 1951) of IAS, PA, and aircraft vertical accelera-
30} _Indicated Air Speed
0: =
: z Nea acres
30,
a Tine i
92 Pressure Aifiude 93
Ftc, 2. Typical VGH recording.
aa Tae —
clean) — :
1 tone Fj
eens)
ewe ach
Teceee
uno CBE ew ee)
‘uo wares ou yor = |
vO TT a
Fic, 3. Data available
‘on most radial fight legs.Avcrse 1968 WILLTAM
tion were also taken on most of these flights (Fig. 2).
‘The University of Chicago Cloud Physics Group
corded continuous values of liquid water by paper tape
tnd hot wire over a large portion of the middle tropo-
spheric radial flight legs (Ackerman, 1962). Fig. 3 is a
plotted portrayal of the typical data available on a
majority of the middle tropospheric fight iegs.
Never before have so many meteorological observa
tions been so systematically collected on a selective
class of storm systems, ‘These observations are unique
in that wind variations and accurate RA values are for
the first time simultaneously available in large numbers
‘on the meso~ and microscules of motion. These measure
ments offer a means of calculating average vertical air
motion to resolutions (0.4-0.7 n mi) not. previously
attainable. The purpose of this study is to describe this
‘method and present results of vertical motion computa~
tions within hurricane cumulus and cumulonimbus
dlouds.
Statement of problem. To measure vertical air motion
from an aircraft, two determinations must be made:
(1) the vertical motion of the aireraft relative to the
round, and
2) the vertical motion of the air relative to the
aireraft,
‘The sum of these two measurements gives the vertical
air motion relative to the ground,
2, Computational method
The vertical motion of the aircraft relative to the
ground is obtained from time differentiation of the
pressure altimeter, modified by the absolute (or radar)
altimeter, when the horizontal pressure gradient was
significantly large. To describe the method of calculat-
{ng the vertical motion of the air relative to the aircraft,
1 short discussion of the aerodynamic theory must be
given.
Aerodynamic theory. From the aerodynamic theory of
aircraft fight (Duncan, 1952), the lift (L) and drag (D)
acceleration of an aircraft, when the aireraft is con-
AIRFOIL aT EQUILIBRIUM
NO VERTICAL NOTION
M. GRAY 465
sidered from the particle dynamic point of view is
/2pVESCryary a
D=1/2pV ?SCopy @
where
L=lift acceleration,
D=drag acceleration,
p=air density,
Smeffective wing area of aircraft,
C1= coefficient of lift,
Co=coefficient of drag,
‘M-=mass of aircraft at time equations are applied,
V = true airspeed of aircraft.
Equilibrium values of lift (L.) equal to gravity (@)
(= 1/2pV #5Czjae= g] and drag (Da) equal to thrust
(1) (D=1/2pV ¢SCoju=T] acceleration can be de-
fined at the places where the B-50 flew with constant
values of power setting, Vi, heading, and altitude,
‘The observed values of V at these places are similarly
defined as equilibrium values of true airspeed (V,).
‘The pitch angle (9) of the aircraft is defined as the
inclination of its longitudinal axis to the horizontal
the angle of attack (a) as the inclination of the aircraft's
longitudinal axis to the relative wind blowing into the
aircraft. Over short periods of 8-10 minutes of flight
‘at constant power setting and plane heading, equilibrium
conditions of @ and a (Fig. 4) may be considered locally
due to the slow percentage fuel burnoff rates.
ibrium conditions will be denoted 8%, and
a Equilibrium conditions are not present when the
plane enters areas where vertical motion is occurring.
Here the actual pitch (@,) and actual angle of attack
(@.) differ from the equilibrium values by deviational
amounts 0, and az (Fig, 3). Deviational values of 8, a,
and V; are then defined as
4, 8 @)
ag=0e—Oe, @
Vua= Veo Vw 6)
AIRFOIL IN AN UPDRAFT
‘Fie, 4, Airfoil at equlibriamno vertial motion,
Fo 5. Aifel in an upaialt466
At places of equilibrium, 84, a, and Va=0; and 8,=6.,
a4=a1, Via Vee Over short periods of time, changes of
actual values equal changes of deviational values—due
to the constancy of the equilibrium values. No vertical
air motion is occurring when equilibrium con
exist, By defining deviations from thei
‘values, phugoidal effects are eliminated.
Incremental deviations of lift (AL) and drag (AD)
acceleration from the equilibrium values of L and D
can then be expressed (Duncan, 1952) as
Sy dcr
a= VE (vi ant? Vea), 6)
rare
pS, dCo
apa v( vi act VC) ”
Mw da
where
aC1/da=change of coefficient of Lift with change of
angle of attack (=5.1),
dCy/da= change of coeficient of Drag with change of
angle of attack (~0.4),
and other symbols as previously defined,
Vertical motion equation. If the deviations of aircraft
pitch and angle of attack from their equilibrium values
an be obtained, then by geometry (Fig. 5) the vertical
‘motion of the air relative to the aircraft is given by the
expression Vs(as=6a). The vertical ait motion relative
to the sea’s surface for small angles is thus
w(air)= Vilas 0.4 W's, @)
where W,=the vertical motion of the aircraft. Meas
‘urement of ag and 6, were not made. The problem then
becomes one of calculating these values,
Determination of ex, and 04. When the aircraft flies at.
constant power setting and heading, and equilibrium
values have been determined, then’ the vertical and
horizontal equations of motion for the aircraft are
sgiven by
4
ave ale HL, O)
MP 5p ene-wo, (10)
a at _
where
\N'= vertical acceleration of aircraft measured at
its center of gravity,
dg/dt=rate of change of aircraft's deviational pitch
angle,
vertical acceleration from rising or falling
yuid water striking aircraft. Positive for
rising liquid water,
WD=drag acceleration from liquid water striking
aircraft,
wi
JOURNAL OF APPLIED METEOROLOGY
Vou 4
4V ufdt=rate of change of deviational true airspeed!
dU {dt rate of change of wind component on plane's
nose. Positive backward,
incremental Lift and Drag as defined in (6)
and (7),
AL, AD='
Ackerman’s (1962) paper tape liquid water measure-
ments along the same radial legs here studied give
‘maximum one second values of 8-9 gm m-? and maxi-
mum 5-10 second average values of 4-5 gm m-¥, The
‘accelerations produced on the over 10%1b aircraft by
these liquid concentrations are of insignificant magn
tude compared with the other terms of (9) and (10).
Distegarding WL and WD and expanding AL and AD
in the form of (6) and (7) the two above equations
become
ay. i (verctart 2¥-0¥ 0) ve an
we (ve act VCiVu) Ve,
\—g0a. (12)
dp
(vee taut VCs)
WN a
All quantities of (11) and (12) are measurable from the
aircraft's recorded observations or known independently
from its flight tested characteristics, except aw and @y.
‘These equations can each be solved for aa, and a new
differential equation set up in the one unknown 8. Thus,
8, dV dU
Sten 6 Raa KV), (as)
a a dt
where
dC1/da dofda
VidCo/do dCifda
dCi /da pSV (CudCo/de cn)
VaCr/da ( ailda
or
(a4)
where
Due to the AN/APN-82 instrument response time,
it is necessary to evaluate dU/dt over selected time
intervals of at least 5-10 seconds (see Appendix). Thi
limits the time resolution to which (13) can be applied.
Over 5-10 second intervals the four é-coetficients in
(13) can be considered constant within a few per cent.
‘These coefficients were evaluated from Boeing aircraft
recorded B-50 instrumental flight test characteristics.
‘The aieais ovontal momentum inhibits from in;
stantancously adjusting is airspeed to large and rapid horizontal
‘Wind speed changes.Avoust 1965 WILLIAM
‘The gust scale turbulent nature of the atmosphere
(450-300 m wavelength) in convective areas indicates
hat values of AW, (20/2, (@Va/A), Vag often vary
y an order of magnitude and change their sign over
S10 second intervle The magnitude of A thus shows
large fluctuations over these intervals and is of the
general character shown in Fig. 6.
time ( see.)
Fo, stating typical variations of A parameter ove the ine
Intervals and mean values of (denoted),
‘These gust-scale wind gradient values could not be
measured. The response of the airspeed meter and
AN/APN-82 is such as to give only average values of
(aU dt) and (@V fat) over 5-10 second intervals, AX
values tend to cancel themselves over this interval.
‘The KaVia term does not significantly affect A. The
primary contributors to A are the measured average
values of (4U/di) and (dV.u/di). When these average
values of A over 5-10 second intervals have been de-
termined from the instrument recordings, (14) may
bbe solved for the mean value of Oy over this interval
Genoted &,), thus
[a+ (e0-
ime of interval under consideration in seconds,
1 value of Os at the beginning of the time interval
when =0,
Fig. 7 portrays the typical variation relationships be-
tween 8, Ba. 8s, and A over time intervals of 5, 8 and. 10
seconds. For most airspeeds flown, Ky was nearly 1. Vale
*The term A is itself # function of time and cannot explicitly
‘be constant in evaluation of 8,, However, if the value of 4 over
the $-10 second interval under consideration possesses the char-
acteristic Huctuations shown in Fig. 6, then the above evaluation
[is & satisfactory approximation of @y to within a few per ceat.
M. GRAY 467
ues of (1—e*1)/1 att values of $ and 10 seconds are ap-
proximately 0.2 and 0.1. The [0is— (4/Ks) ](—-**/1)
term is significant in the computations only when 82e
and A are of opposite sign or of quite different magni-
tudes. It usually contributed Jess than 10-15 per cent.
to the average value of @s. 01 approaches A with in-
creasing time interval (Fig. 7)
‘Substitution for draft scale vertical motion. Once mean
values of @y over the 5-10 second intervals have been
determined, they can be substituted in (12) and mean
values of ag determined over the similar interval, These
values may in turn be substituted into (8) and together
with the earlier determination of Wy, the mean vertical
air velocity over flight time interval of 5-10 seconds
(0.40.7 n mi) obtained. The aircraft traveled at ap-
proximately 250 knots, Values of & were always within
+£0.22, while values of dz ranged up to -£3-4°, To a
close «pprosimation then,
w(air)=— Vt Wp,
(16)
where the — (bar) signifies mean value over the 5-10
second interval. Since draft velocities possess charac
teristic widths of 1-3» mi, a resolution to 0.4-0.7 n mi
should describe a major portion of the vertical draft.
Acceleration measurements, Vertical gust velocities
with characteristic widths of a hundred meters may be
approximated with respect to the environmental draft
motion with the derived gust velocity formula (Pratt
and Walker, 1954) developed by NACA Gust Loads
Section (now NASA Structural Dynamics Branch).
‘This formula defines the derived gust as
2aNM
poSdCr/daViK,
‘This has been astamed inthe evaluation of
value of 4 over the 3-10 second time inter
acter of delta function, would the evaluation of By by (15) be
significantly in error, Nutnerieal averaging of 4. over 5-10 second
intervals under assumed, widely vanable variations of Pa, U
[AN have verted the above assumption of near constancy of 4.
Mae
pete eat tated
0,7, Portrayal o ype vals of a Ou Aya os
ieee selected ébmputation timc inSrvais468 JOURNAL OF APPL
where Ky is the gust factor (=0.72), Vr=Indicated
airspeed ‘and the other symbols are as in previous
equations.
“The vertical component of the measured gust is
primarily responsible for the vertical acceleration of the
aircraft and the value of ger. The vertical acceleration
‘may also be effected by the gust’s horizontal component,
but this can only account for about a quarter of the
observed vertical acceleration, Pilot maneuver has a
Gifferent characteristic frequency and can usually be
eliminated from the calculations.
Once these gusts have been approximated by the
derived gust formula, they might be superimposed on
‘the draft scale vertical motion already computed. This
will give an approximate comparison of the two scales
of motion, even though the derived gust may not be closely
equivalent to the atmospheric gust. Fig. 8 portrays typical
calculated values of vertical draft motion before and
after derived gusts have been superimposed.
Determination of time interval over which axerage
IED METEOROLOGY Vous 4
rertical motion is obtained. The rapidity of response of
the AN/APN-82 system to wind changes was the prin-
cipal factor in determining the 5 to 10 second time
interval over which the individual average vertical
motion computation was made. The AN/APN-82 sys-
tem responds at a rate of approximately 1 to 2 knots/
sec and one degree/sec for wind fluctuations of ap-
proximately 10 knots. For larger fluctuations. the
response time is more rapid.
From a study of the AN/APN-82 system and large
samples of its gathered winds, it was determined that
‘measured wind changes on the nose—i.e., dU/di—over
periods between 3 and 10 seconds are representative
‘of the actual wind changes—if the boundaries of these
time intervals are taken at places where the derivative of
the wind component on the nose changes sign, changes to
or from zero, or does not change at all. Such derivative
‘changes usually occurred once very 5 to 10 seconds and.
account for the variable time interval used. Fig. 9
illustrates a typical profile of the observed variation of
VERTICAL, VELOCITY
Fors.
REET SoALE wr
SUPERIMPOSED GUSTS
Fig, 8. Mustrating the portrayal of vertical motion across the eye wall of a radial Right leg. ‘Top diagram—rectangular
‘curve showing average vertical velocity aver varying time intervals of 5-12 seconds. Smooth curve with dots representing
‘overlapping smoothing interval of 5 seconds, Bottom diagram? —smeoth curve ol Lop diagram with superimposed derived gustAcoust 1965
10,9, Mustrating selection of time intervals over which aver-
age vertical motion computations are made. Boundary of time
Intervals chosen at places where AN/APN-S? wind shows abrupt
changes.
the wind component on the plane’s nose. The veritical
lines bracket the time intervals over which the equa-
tions are applied. Between these places the mean deriva-
tive of U approaches the derivative of actual wind. Tf
the derivative of the AN/APN-82 wind changes sign
at two successive places, then the actual wind must
have also changed sign at these places, The AN/APN-82
responds instantaneously to changes in the actual wind,
but lags only in time adjustment to the actual wind,
Fig. 10 illustrates this idea for hypothetical actual and
measured wind variations. The U measured by the
AN/APN-82 is lagging in time response to the actual
ind. When the actual wind component rises, the
WILLIAM M.
GRAY
Fic. 10. Tustrating how hypothesized actual sind and hy-
pothesizd AN/ADN.82 measured wind could be diffrent due to
apid fluctuation of actual wind and lay of AN/APN.82 Ume
sponse. Als illustrates. how hypothesized actual and AN/
‘APN-82 wind changes between vertical lines would be equal.
AN/APN-82 measured U also rises and continues rising
until it approaches the actual wind, or until the actual
wind decreases to become less than the measured wind.
At this precise point, the actual and measured winds
ate identical, and the derivatives of the measured and
actual winds approach equality. The average time in-
terval and distance equivalent was 8.7 seconds or
=0.6 n mi, In no case was the time interval less than
5 seconds (~0.4 n mi) or greater than 13 seconds
(~0.9n mi).
Fo. 11, Example of data recordings and computations along a radial fight leg toward the south from the center of Hurricane Cleo.
‘Top smoothed and rectangular curves those of computed average vertical motion with superimposed gust velocities (kn)—sinilar
to Fig. 8,
io. 12, Example of data ceordings and computations along a fight log at 570 mb towacd the southeast into the
‘center of Hurricane Helene, Data portrayal the same as Fig. 1470
MAXIMUM VELOCITY
of DRAFTS (Kn)
80 —
6o-
No.of DRAFTS
é
1
SEL oe wolno OE
£222 2/22 2s8
2220 mmo oO §
DOWNORAFTS UPDRAFTS
Fig, 13, Histogram of maximum vertical draft velocities.
COMBINED UPORAFT AND
DOWNDRAFT WIDTH
80 -
60-
40-
No. of DRAFTS
20-
Fio, 14, Histogram of draft widths,
3. Results?
Calculations of vertical draft velocity with super-
imposed derived gust velocity have been made on four
middle tropospheric flight levels in three hurricanes in
which National Hurricane Research Project aircraft
flew radial leg missions during the 1958 season. Plight
information is given in Table 1. Typical vertical draft
velocities with superimposed derived gust velocity along.
‘ovo radial legs are illustrated in Figs. 11 and 12, Sum-
rmarizing all calculations, it was found that downdraft
2 For & more complete description and discussion of the com-
putational method and results, the reader is referred to the Na-
Uonal Hurricane Research Project eprint manuscript, On the
‘Scales of Motion and Tnternal Stress Characteristics of the HUrr-
cane, Distribution from Miami, Fa,
JOURNAL OF APPLIED
METEOROLOGY Vousme 4
MAXIMUM GUST
160} VELOCITY ON UP
AND DOWNDRAFTS
1204
® 80-4 |
% \
5 I
2 1
404 | |
Io
pi
Peet H®@ aR
2222338
-¢% Qn 2 DS
Kn
ic. 15, Histogram of individual derived gust velocities which were
‘superimposed on individual drafts at midale tropospheric levels
‘Tape Information on four National Hurricane Research
‘Project fights used in velocity calculations.
Mini-
press Max. Flight legs of nmi of
Sure, wind, level, compu eorapit-
Storm Date ‘mb’ “kn mb’ ‘tation tations
TAug 970 &
Aug 9065
27 Aug 910120
25Sept. 950110
and updrafts were approximately equal in number and
magnitude, Draft widths averaged approximately 1.4
n mi and average maximum draft velocities were 7-8
‘knots. Histogram distributions of the draft magnitudes
and widths are portrayed in Figs. 13 and 14, The maxi-
mum derived gust velocities averaged approximately
half the maximum draft velocities.
Derived gust velocities. The VGH accelerometer traces
showed distinct areas of predominant turbulent air
along the radial fight legs. The characteristic period of
the derived gusts varied between 1.5 to 2.5 seconds
(200 to 300 meters) or 0.75 to 1.25 seconds (~100 to
150 meters) between maximum and minimum values.
Fig. 2 illustrates a typical turbulent area.
Along the 22 radial legs of accelerometer dat (all
Detween radit 10 to 60 n mi) studied—approximately
1100 n mi of flight—there were 84 separate areas
(totaling approximately 280 n mi—or one-quarter of
“hie sompstson should nt be vewelna tisk quantitative
sens for infinany instances undoubtedly only the edge of U
Sa see ehooumerel. vonyAvcust 1965 WILLIAM
the flight distance) of concentrated turbulent motion.
‘These turbulent areas averaged 3.5 n mi in width and
‘were always within or closely adjacent to the computed
draft areas, Outside the hurricane—when the aircraft
were going to or from the storms—similar turbulent
areas were seldom encountered.
‘The derived gust velocities in the 84 turbulent areas
averaged approximately 4 knots (40.15) in
magnitude and had average half wavelengths (or
widths) of 130 meters. Thus, in the average turbulent
area the aircraft was sustaining vertical acceleration
variations of -£03 g within periods of approximately
‘one second or in 130 meters fight distance. The average
maximum derived gust per turbulent area was 28.5,
knots (~:i0.35 g). There were 38 recorded derived
gusts greater than 12 knots (=E0.5 g) or 38 places
where the vertical acceleration of the aircraft would
have changed by 1.0 g or more in one second, A gust
‘occurred on an average of one every 1.8 sec
meters) in the turbulent areas.
‘The turbulent areas were usually, but not always,
concentrated within the drafts and had a tendency to
be stronger when the drafts were more intense. How-
ever, much variation to this general pattern was,
evident. Fig. 15 is a histogram of the individual maxi-
mum derived gusts which were superimposed on each
up- and downdraft. In general, accelerations were less
‘than those often experienced and expected within
Great Plains severe weather, but in accord with those
observed by the Thunderstorm Project.
Association of liguid water with drafts ond gusts.
Paper-tape liquid-water measurements (Ackerman,
1962) were usually well correlated with the calculated
drafts, The highest values from the paper tape usually
corresponded to the higher magnitude drafts (Figs. 11
and 12).
Temperature departures within drafts. A comparison
of vortex-thermometer observations (Hilleary and
M. GRAY any
Christensen, 1957; Hawkins ef al, 1962), outside and
within the computed drafts were also made.* Only on &
statistical basis were bouyancy requirements being met,
In some updrafts temperature was slightly lower than
that of environment; the reverse was true ina number of
the downdrafts. Some drafts had no temperature differ-
ences. Fig. 16 portrays the scatter of maximum up-
and downdraft temperature departures from sur-
roundings (surrounding temperatures defined as the
‘mean temperature over a 5 n mi distance straddling
the draft) vs. draft velocity. Draft velocity must have
been 6 knots or more to have been considered. The peak
value of the temperature within the draft was taken,
Sometimes it occurred on the inner-sides of the draft;
at other times near the middle. ‘The same scatter of
cloud temperatures was also shown by the tempera-
ture recordings of the Thunderstorm Project (Figs
22-28, p. 31 of that report).
‘No attempt will be made to explain the discrepancy
of some of these observations from bouyancy concepts.
‘The vortex-thermometer measurement is probably the
most suspect, but other considerations need also to be
examined. Such considerations are beyond the intended
scope of this study
4. Comparison of results with thunderstorm project
‘The ‘Thunderstorm Project is the only similar study
with which comparisons of present computations can be
made. A statistical comparison of the Thunderstorm
Project fights at 15 and 16 thousand feet in Florida and
"The virtual temperature corection was not made. ‘The
viconmnt relative humidity surrounding th draft asso high
That the mato enretion erence always les than 0.2
vortecthemometer tenpersturs wece adja to 4 con:
ase presae ght with ented laps tate of the mean topical
Simosphers. Height changes within tho drafig weve pot enough
{ovemine mice than O2C aiderence between the mein ane na
oud poe ne
‘TEMPERATURE DEVIATION
vs.
DRAFT VELOGITY
TEVPERATURE*C
DOWNDRAFT (kn)
ue. 16, Seater diagram of the maximum temperature deviation within y
pared with that of the surrounding temperature (surtounding temperature del
perature over a 5 n mi distance straddling the draft) versus up- and downdra
ka or greater were considered.
UPORAFT (Ka)
and downdraits com-
ned. a8 the mean tem
‘velocity. Only drafts472
Ohio with present hurricane computations at 13, 15
and 16 thousand feet (620 and 560 mb) is made.
‘Thunderstorm Project tables 7-13 (pp. 40-45 of that
report) tabulate draft velocities, draft widths, and
effective gust velocities between selective velocity and
width intervals (ie, 0-59, 60-118 knots, etc., and
0-0.17, 0.18-34 n mi, etc.). By assuming that all values
in each range interval would average out at the middle
(.e,, all values in range interval 6,011.8 would average
8.9) a statistical comparison of the overall Thunder-
storm Project results with those of the hurricane may
be made, Because the statistical results of the Florida
and Ohio data were nearly identical, they were averaged.
“Together they will be referred to as the Thunderstorm
data.
Draft comparison. Table 2 portrays average values of
Graft velocity and draft width for each study. The
distribution range of draft velocity and widths is also
similar if comparison is made of Figs. 13-14 of this text
with Thunderstorm Project tables 7-8 and 10-11 (pp.
40-42 of that text), There is close comparison between
statistical averages.
Gust comparison. ‘The maximum derived gusts in the
turbulent areas of the hurricane were similar to those
‘of the thunderstorm. The average maximum derived
gust velocity per 10 seconds of flight in the Chio
thunderstorms at 15 thousand feet was 8.0 knots. The
average maximum derived gust velocity per 0.5m mi
traverse through the Florida thunderstorms was 8.3
nots. These- values may be compared with the 8.5
nots average maximum gust per turbulent area
(avg, 3.5 n mi) encountered in the hurricane. Average
derived gust width in the Thunderstorm Project was
approximately 100 meters, and in the hurricane, ap-
proximately 130 meters.
Discussion. The Thunderstorm Project attempted to
fiy its aircraft at constant power setting and airspeed.
‘This was thought to eliminate any vertical air motion
relative to the aircraft. Draft velocities would then be
directly given from changes of aircraft altitude. No
*The Thunderstorm Project computed elective gust velocities
from vertical acceleration records. Folefon (1936) has converted
Thunderstorm fete gusts to deed gusts, “Ax devel gist
velocities were éompated in the hurricane stud, ‘Tlefson's gust
‘elocities are seed for comparison
JOURNAL OF APPLIED METEOROLOGY
Vou 4
evaluation of vertical motion was made when large
airspeed changes were observed in the cumulonimbus
because pitch angle changes and consequent vertical
motion relative to the aircraft were thought to be
present. However, constancy of airspeed at constant
Power setting does not strictly imply pitch angle
constancy if horizontal wind changes are accompanying
the draft. An increase of the horizontal wind component
on the nose would be reflected by an increase of airspeed
‘at constant pitch angle. This and other recent studies
hhave indicated large horizontal wind changes in con-
veetive clouds (Gentry, 1963; Fujita, 1962; Newton,
1963; Steiner and Rhyne, 1962; and McLean, 1961).
The problem is vastly more complex if one attempts
to measure directly vertical motion to gust-scale resolu-
tion. One must measure exactly the gust-scale pitch,
angle of attack, and vertical displacement of the air-
craft, Instrument requirements become quite critical,
and voluminous data reduction is necessary. Holling
and Malkus (in Jones, 1955), Bunker (1955), and ‘Tel-
ford and Warner (1962) discuss the severe instrumental
requirements needed for such measurements.
Not attempt has been made with the method here
presented to obtain similar direct vertical-motion meas-
‘urement to gust-scale resolution, Instrumental capa
ties would not allow this. The derived gust velocities
are presented only to show the magnitude and scale
‘of the accelerations experienced and the approximate
association of the gusts with the drafts,
‘This computational method is unique in that the
horizontal wind changes, as measured by the AN/APN-
82 Doppler wind instrument, are directly used to
measure pitch angle variation Pilot pitch-angle
maneuver is allowed for, and computational require-
ments are not excessive.
‘The average width of the up- and downdrafts were
approximately 14 n mi. There can be no doubt that
the most significant scale of vertical motion in the
hurricane is the cumulus draft scale. This is not to say
that an overall mass circulation through the storm
system is not taking place and is not highly significant.
‘This need average only a small fraction of @ knot over
the large inflow areas, This broader mass circulation
through the storm system is most likely to manifest
"Tate 2. Comparison of hurricane and thunderstorm project cumulonimbi
Hurricane fights
‘Thunderstorm fights
ae nea
a ee ee
a ee a a er
cSSess he ee
Se B FB ns
These computations should only be vis
viously mentioned, only
ince a ver
ved in a qualitative sense as the draft resolutions were only (0 0.4 100.7 m misandaspre-
lhe edge of many drafts may have been encountered.
‘velocity of but three knots vas neoded for consideration as a draft, a number of the draft values undoubtedly
came from non-cumulonimbus clouds, his mus be considered in comparison of data with Thunderstorm Flights.Aucust 1965
itself in a slightly greater number and/or strength of
upirafis to downdrafts. The present computations are
incapable of resolving this smaller magnitude and larger
space scale circulation.
Acknowledgments. The auithor expresses his gratitude
to Professor Herbert Riehl, Head of the Department of
Atmospheric Science, Colorado State University, for
his advice and encouragement in this investigation.
‘This research has been financially sponsored by the
National Hurricane Research Project of the U. S.
Weather Bureau,
Appreciation is extended to the National Hurricane
Research Project personnel for their friendlly assistance
and cooperation in obtaining and interpreting the data
here used. The author would like to acknowledge helpitt
discussions of the data used or subject matter with
Professor Horace Byers and Miss Bernice Ackerman
‘of the University of Chicago, Dr. Robert Simpson of
the U, S. Weather Bureau, Dr. R. Cecil Gentry, Direc-
tor of the National Hurricane Research Project; Dr.
Pat Squires of the National Center for Atmospheric
Research, Mr. U. Oscar Lappe of the New York Uni-
versity, Dr. Michio Yanai of the Japan Meteorolo
Research Institute, and Mr. Andrew Bunker of the
Woods Hole Oceanographic Institute.
“The author appreciates loan of the Velocity-Gravity-
Height (VGH) records from the National Aeronautics
and Space Administration’s Structural Dynamics
Branch and of the liquid-water oscillograph recordings
from the Cloud Physics Group of the University of
Chicago.
APPENDIX
Representativeness and Accuracy of Data Sample
‘The above calculations are felt to be closely repre-
sentative of the vertical motion occurring at middle
tropospheric levels and at radii between 10 and 60 1 mi
in the moderate hurricane. There is, however, a bias
to the data sample in eliminating a number of the
‘most intense vertical drafts. This is due to two effects:
1) In ten places where it was obvious from the
changes of radar altitude, indicated airspeed, and
liquid water concentrations that high vertical
velocity was occurring, the AN/APN-82 failed
to function properly. Along one radial flight leg,
the AN/APN-82 system did not function properly.
for a major portion of the fight leg. In such places
no vertical motion computation was. possible,
‘There was a tendency for the AN/APN-82 to
‘malfunction selectively in the strongest convee-
tive areas,
Although pilots were instructed to hold their
‘course at all times—commensurate with safety,
in some cases there was circumnvigation of radar
echoes.
Q
WILLIAM M. GRAY
413
Hilleary and Christensen (1957) and Hawkins et al,
(1962) have discussed the accuracy of the instrumenta~
tion used in this study. ‘The measitrement whose accuracy
might significantly alter the results is the wind speed
‘measurement by the AN/APN-82, No thorough testing
as to the precise accuracy and response characteristics
of the AN/APN-82 under all conditions has been con-
ducted, This would be necessary before definite confi-
dence limits could be placed’ on results. Previous
experience has shown that the AN/APN-82 winds are
‘most likely to be in error when
1) the winds are very light;
2) when the aircraft is in turns or is changing
altitude;
3) when the ocean surface is nearly flat due to calm
surface wind conditions;
4) when there is a sustained ocean current under
the aircraft [this would cause the AN/APN-82
radar energy to be reflected from a moving rather
than stationary surface. ‘The winds would thus be
‘uniformly in error by the amount of the water
motion (Grocott, 1963)];
5) in occasional heavy rain or thick cloud areas when
the AN/APN-82 may lock onto and reflect from
these sources rather than the ocean surface.
Conditions one through three were not encountered
in this study. Wind-speeds were always greater than
30 knots. Calculations were never made while the air-
‘raft was in turns or changing altitude, Sea conditions
were always unstable. Condition four should not
noticeably affect these results as the computations
required only the mensurement of the change of wind
and not the wind’s absolute value,
errors due to condition five were occasionally present.
There were ten obvious places (usually of 10 to 15
seconds duration) where the AN/APN-82 did not
function properly and wind speeds appeared unrepre-
sentatively light. In most of these cases very large
and almost discontinuous decreases occurred. In other
instances the AN/APN-82 was unable to record any
wind at all and reverted to a memory mode, None of
these areas of unrepresentative wind were used in the
computations.
Cloud and rain particles travel around the storm at
velocities close to that of the winds. These speeds are
‘much larger than the ocean speed. If reflection is coming
from the rain or cloud sources, the computed winds
would take on a very different speed character. Wind
changes of but 5 to 15 knots could not be due to inter-
ception of the more rapidly moving cloud particles.
‘NHRP instrument engineers believe that nearly total
interception of radar energy by rain or coud was
responsible for these few unrepresentative measure-
ments (i.e., condition five). The AN/APN-82 will ealcu-
late the wind from the strongest return signal. It does
not integrate the total energy return, Unless the returnama
from the ocean surface was less than that from another
source, the wind measurement is always based on the
return from ocean source alone, Partial or weak inter-
ception of rain, cloud, or spray by the AN/APN-82
would not significantly affect the wind determination
if the computation is based on the strongest return only.
‘Thus in most cases the AN/APN-82 correctly measures
the wind despite the presence of rain and cloudines
‘There is evidence which points to the basic reliability
of the AN/APN-82, In all but the few cases cited above,
the computed winds appear to be very reasonable. The
‘magnitude and characteristics of the wind fluctuations
tre much like those of the airspeed changes. Navigation
corrections after many hours of fight with AN/APN-82
svinds were usually within a few n mi. The wind changes
experienced in the clouds were not observed outside the
convective areas. The calculated vertical velocities of
this study appear to be reasonable in magnitude and
width with observations from radar, the Thunderstorm
Project, and previous theoretical speculation,
‘Any doubt about the accuracy of the present AN/
APN-82 wind measurement does not invalidate the
‘method presented. Future vertical motion calculations
‘with a more completely tested and accepted fine resolu-
tion wind-measuring device can be accomplished.”
"This discussion of the accuracy of the AN/APN-82 is meant
to refer only fo the winds measured from the NURP B-50 virraft
Guring the 1058 hurricane fights. The reliability. of the AN/
‘APN82 winds which were measured from the Research Plight
Faciity (RFP) DC-6 during and after 1960 and for the Na-
tional Severe Storms Project has not been investigated by the
futhor, The above discussion ie intended to apply only to data
Of the fight levels here studied.
REFERENCES
‘Ackerman, B,, 1962: Hurricane cloud physics research. Final Ree
‘port, Contract Cwb 9720, 68 pp. (Available from Dept. of
Geophysical Sei, Univ. of Chicago.)
Boring Airplane Company’: Various technical papers on B-50
aircraft flight ‘characteristics, (Available from Witchita,
Kansas, Office)
Braham, R. R., 1960: Hurricane cloud physics research, Final
‘Report, Contract Cwb 9175 and Cwh 9484, 152 pp. (Avail.
able from Dept. of Geophysical Sc., Univ. of Chicago.)
Bunker, A. F, 1955: Turbulence and shearing stresses measured
cover the North Adantic Ocean by an sirplane-aceleration
technique. J. Meteor, 12, 445-455,
Byers, H. and RR. Braham, 1949: The Thunderstorm. U. S
Govt. Printing Office, 287 pp
, 1961: On the structure of Hurticane Daisy (1958).
"Report No, 48, 102 pp. (Availabe from
U.S, Weather Bureau, Mian, Fla.)
Donely, P., 1949: Summary of information relating to gust loads
‘on aisplanes. NACA Technical Note 1976, 145 pp. (Available
from NASA, Washington, D.C.)
Duncan, W. J, 1952: The Principles ofthe Coniral and Stabitty of
“Aircraft. London, Cambridge Univ. Press, 384 pp.
Durand, W. F,, 1985! Aerodynamic Theory. Val. I and V., Ber-
lin, J. Springer, 867 and 347 pp.
Fujita, 7, 1962: A review of research on analytical mesometeor-
logy” 114 pp. (Available from Dept. of Geophysical Sc,
Univ. of Chicago.)
General Pression Laboratory (GPL), 1955: Doppler Ait-Naviga-
tion Technical Series. Section Tand 1, Volumes 1, 2,3, 6,7,
10, 150 pp. (Available from GPL, Pleasantville,
JOURNAL OF APPLIED METEOROLOGY
Voure 4
Gentry, R. Cy 1963: A study of hurricane rainbands, Nat, Hur,
Res, Proj. Report No, 69, 75 pp. (Available from U.
Weather Bureau, Miami, Fla.)
Gray, W. M,, 1962: On the balance of forces and radial accelera~
tons in hurricanes. Quart. J. R. Meter. Soe, 88, 30-458,
1965: On the scales of mation and internal stress character-
Istics of the hurricane. Nat. Hur. Res. Proj. Report, (Avail-
able ftom U. 8. Weather Bureas, Miami, Fa.)
Grocott, D. F.H., 1963: Doppler correction for surface movement
J. inst, Nasigation, 16, 51-63.
Hawkins, H. F, a ol, 1962: Inventory, use and availability of
‘Nat. Hur, Res. Proj. data gathered by aircraft. Nat. Hur
Res, Proj, Report No. 52, 240 pp. (Available from U. S.
‘Weather Bureau, Miami, Fla.)
Hilleary, D. T., and F. F. Christensen, 1957: Instrumentation of
Nat. Hur. Res, Proj. aircraft. Nat. Hur. Res, Proj. Report
No. 11, 71 pp. (Available from U. S, Weather Bureau,
‘Miami, Fl)
Jones, R. F:, 1954: Five fights through a thunderstorm belt,
Quart. IR. Meteor. Soe, 80, 877-381.
—, 1985: Discussion of five lights through a thunderstorm belt
(Quart. FR, Meteor. Soc, 81, 122-124.
‘Malkus, J.S., 1960: Recent developments in the study of penetra-
‘ive convection and as application to hurricane cumulonim-
bbas towers. Cumulus Dynamics, New York, Pergamon Press,
65-84,
—, a al, 1961: Cloud patterns in Hurricane Daisy. Tellus, 12,
‘ean
‘McLean, G. S,, 1961: Observation of severe convection activity in
‘a squall ine, Bull. Amer. Meteor. Sec, 42, 252-264.
Newton, C. W,, 1963: Dynamics of severe’ convective storms.
‘Medeor. Monogr, Boston, Amet. Meteor. Soc, 5, No. 27,
33-58,
Pratt, KG, and W. G. Walker, 1954: A revised gustoad
formula and a re-evaluation of V-G data taken on civil trane-
port aieplanes from 1983 to 1950, NACA Report No, 1206,
5'pp. (Available from NASA, Washington, D.C.)
Richardson, N. R,, 1951: NACA VGH recorder. NACA Technical
‘Note 2265, 16 pp. (Availabe from NASA, Washington, D. C.)
Rich, H., 1950'On the role ofthe tropics in the general circulation
of the atmosphere. Talus, 2, 117.
—, 1954: Tropical Meteorlogy. New Vork, McGraw-Hill Book
o,, 394 pp.
— and J. S""Matkus, 1958: On the heat balance in the equa-
{orialirough zone. Geophysice, 6 903-538,
—,, 1961: Some aspects of Hurricane Day, 1958. Tellus, 13,
181-213,
_ 1968: Some relations between the wind and thermal struc:
ture of steady state hurricanes, J. Atmor. Se, 20, 276-287 pp.
Senn, H. V., and E. F. Low, 1989: Studies of the evolution and
motion of radar echoes’ from hurricanes. Unpublished Re-
search Report of the Marine Laboratory and the Radar Re-
search Laboratory of the Univ, of Miami, 109 pp, (ASTTA
AD-227258.)
—; and H. Hier, 1959: On the Origin of hursicane spiral rain-
bands. JMeter, 16, 419-126,
—; thal, 1962: Radar hurricane research 1 July 1961 to 30 June
4962, ‘Unpublished Research Report of the Marine Labora-
tory and Radar Research Laboratory ofthe Univ. of Miami,
109 pp. (ASTIA AD-268058,)
Steiner, k., and R. H. Rhyne, 1962: Some characteristics of severe
storm turbulence. Nat. Severe Storms Project Report No. 10,
ULS. Weather Bureau, Dept. of Commerce, 17 pp. (Available
from U. 8. Weather Bureau, Washington, D.C)
‘Tatfor, J. W., and J. Wamer, 1962: On the measurement from an
aircraft of bouyancy and vertical air velocity in cloud. J.
Atmos. Sc, 19, $1523.
‘Tolelson, HL, B., 1956: Summary of derived gust velocities ob-
tained feom measurements. within thunderstorms. NACA
Report No. 1285, 7 pp. (Available from NASA, Washington,
D.C)