UDC 661.616:661.46:797.144(263)”1970.08”
Meteorological and Oceanographic Conditions
During the 1970 Bermuda Yacht Race
FREDERICK SANDERS-Department of Meteorology,
Massachusetts Institute of Technology, Cambridge, Mass.
ABSTRACT-Analysis of conventional data and of infor-
mation provided by a number of the competing skippers
yields an unusually detailed picture of environmental
conditions during the Newport, R.1.-Bermuda Yacht Race
in June 1970. Sea-surface temperature data indicate the
presence of a warm meander of the Gulf Stream just west
of the rhumb-line course, a position intermediate between
that of a warm meander to the west in May disclosed by
bathythermograph observations from €he RMS Franconia
and that of a warm eddy to the east in August found by
a Naval Oceanographic Office survey.
The fleet was harassed by two groups of severe thunder-
squalls during the night of June 21-22, in the vicinity
of the warm meander. Even the anomalously high sea-
surface temperatures, however, were cool relative to the
air in which the thunderstorms were rooted. The storms
originated in the Chesapeake Bay area during the day on
June 21, and they appeared, surprisingly, to gain intensity
over the ocean after being cut off from their surface
source of warmth and moisture. Offshore forecasts for
June 21-22 took no specific account of the presence of
the severe thunderstorm systems.
On June 25, part of the fleet experienced an unexpected
southerly gale just northwest of Bermuda. From the yacht
data, it is found that the gale was attributable to a small
cyclone that formed in an old frontal cloud band and
moved northeastward, remaining undetected by the con-
ventional data network throughout its life history. Analysis
of the surface wind field suggests that baroclinic effects
played only a minor role in the behavior of this cyclone,
which a t least ‘in some respects resembled a tropical
cyclone. Study of the forecasts available a t the time
indicate that in neither case did small-scale convective
activity have a significant direct effect upon the larger
scales of motion.
1. INTRODUCTION
This account of the meteorological and oceanographic
circumstances of the 1970 Newport, R.1.-Bermuda Yacht
Race is given because the presence of a large number of
weather-sensitive platforms in a relatively limited area
offered a rare opportunity for oceanic mesoanalysis and
because the meteorological events seemed to this neophyte
participant to be remarkable, although I am assured by
older hands that the weather was “about average.”
Nonetheless, tt number of boats, including the author’s,
were disabled.
The analyses to be presented are based both on con-
ventipnal data of various types and upon the replies of
fellow skippers to my request for data. These replies
were of varying degrees of precision and quantitativeness,
but the analyses do not knowingly violate anything
recounted.
It is often said that yachtsmen exaggerate. Should this
be true in the present instance, the propensity for ex-
aggeration, being subjective, should vary strongly from
individual to individual. The consistency of the patterns
that emerge from the yachtsmen’s accounts might then
be regarded as a measure of the credibility we should
attach to them. We note, a priori, that only generally
experienced crews are accepted as entries in this race;
panic is not likely to have distorted the information
provided. Navigation error is very d a c u l t to estimate,
as it depends on assiduousness of dead reckoning and
availability of celestial objects for lines of position (use of
Loran in the race is forbidden). We estimate that the
root-mean-square position error of the yachts is between
5 and 10 n.mi.
The course of the 1970 Bermuda Yacht Race is shown
in figure 1. The start was early in the afternoon of June 20.
At any particular time, the fleet covered an area elongated
in the direction of the rhumb line, principally because
the larger boats travel faster, and extended laterally
because of differing track choices of individual skippers in
contending with head winds and with the need to com-
pensate for the eastward set due to the Gulf Stream. The
presence and location of meanders and eddies in the
Gulf Stream are key competitive factors, of course, so
most of the yachts measure sea-surface temperature a t
least occasionally during transit of the presumed zone of
intersection of the Gulf Stream and the track.
Oceanographically, there was not much agreement
between individuals about the details of the currents
encountered although the majority reported help rather
than hindrance by the component of current along the
track. There was good agreement, however, as to the
maximum sea-surface temperature observed in the Gulf
Stream; no yacht found this value less than 78’F nor
more than 8OOF.
Meteorologically, two vigorous bands of in tense thun-
derstorms moved across the fleet early in the evening of
June 21 and again around sunrise June 22. These storms
were widely regarded by the skippers as typical “Gulf
MONTHLY WEATHER REVIEW 1 Vol. 100, No. 8, August 1972 / 597
i
. Davids
Argus Tower
+ 30
65 f t 70 75
Light
FIGURE 1.-Course of the 1970 Newport, R.1.-Bermuda Yacht
Race. The start is at the Brenton Reef Light Station near New-
port, and the finish is off St. David’s Light, Bermuda. The
Argus Tower on Plantagenet Bank southwest of Bermuda is a
mark of the course and must be left to port. The hatched area
represents approximately an envelope of the tracks of individual
yachts.
FIGURE 2.-Mean sea-surface isotherms for June 1970 at intervals of
2°C (U.S. Naval Oceanographic Office 1970) with values in OF in
parentheses. The heavy solid line is the rhumb line from Brenton
Reef to the Argus Tower. Circled Xs represent approximate
positions of the warm meander observed in May and of the warm
eddy discovered in August.
Stream weather,” although we shall show that they were
north of the Gulf Stream proper and that the warmth of
the water played a minor role a t best. Finally, on the
afternoon of June 25, when all but the smallest boats had
completed the race and were in harbors in Bermuda, an
unexpected southerly gale harassed the remaining com-
petitors and crippled some of them.
We will first examine the oceanographic situation in as
much detail as possible and will then discuss the salient
meteorological events.
2. THE GULF STREAM
The mean sea-surface temperature pattern for June
1970 is shown in figure 2. Note particularly the northward
bulge of the isotherms between 70’ and 71’W, north of
the Gulf Stream, representing an excess warmth of about
4’F over the long-term June average. This anomaly,
together with an excess coldness of 2’3’F shown in the
southward bulge of the isotherms to the east, if represent-
ative of more than a superficial contrast of surface
temperatures, could account for a southeastward com-
ponent of current as suggested by the accounts of most of
the skippers.
Further evidence for the significance of this feature is
found in the time series of bathythermograph soundings
made from the RMS Franconia (U.S. Naval Oceano-
graphic Office 1970) during its traverse from New York
to Bermuda on May 29-30, which showed a prominent
northward meander of the Gulf Stream a short distance
598 1 Vol. 100, No. 8 / Monthly Weather Review
711 70W
FIGURE 3.-Detailed analysis of sea-surface temperature in the
vicinity of the Gulf Stream on June 21-22, 1970. Tracks of vessels
recording continuously or observing at frequent intervals are
shown by heavy dashed lines. Locations of other more isolated
observations are shown by heavy dots. The leftmost track is that
of the RMS Franconia, along which the arrows represent an
estimate of the current profile of the Gulf Stream.
38'40'N -7?46\N 3 7'5& - 70'5 3 h 37bSh-6G5Ob
1200 1300 1400 1500 1600 1700 1800
TIME IGMT)
FIGURE .i.-Time-section of bathythermograph temperature soundings from the RMS Frunconiu on June 21, 1970. Latitude and longitude
are shown a t top of the diagram for selected times. Isotherms are labeled in "C with approximate Fahrenheit values in parentheses.
west of the mean June 1970 position of the warm bulge.
Finally, a prominent anticyclonic eddy was discovered
in late August moving northeastward near 39'N, 69.4'W
(US. Naval Oceanographic Office 1971). Thus, it seems
likely that the meander or eddy produced an anomalous
southward current during the race, which most aided those
yachts near or a short distance west of the rhumb line.
Further detail appears in figure 3, which is based on
temperatures observed by 12 yachts on June 21-22 and
also by the RMS Franconia, which passed southeastward a
short distance west of the fleet on the 21st. The latitude
at which the yachts first encountered the 78'F isotherm,
probably representing the approximate position of the
north wall of the Gulf Stream, varied only from 37'55'
to 38'25". The north-south oscillations of this isotherm,
with a wavelength of about 40 n.mi. (65 km) and an
amplitude of substantial size, do not seem to be attribut-
able to errors in the ship thermometers or positions.
Whether or not they were associated with significant
current systems is impossible to say.
The warm bulge seen in the mean June pattern is very
prominent in this analysis as well and lies very close to
its location in the mean pattern. In figure 3, however, there
is great detail, the significance of which is unknown.
Nevertheless, all six yachts reporting in the vicinity of the
bulge found at least one pronounced maximum of warmth,
with temperatures ranging between 70' and 75'F, well
to the north of the isotherms associated with the Gulf
The fortuitous passage of t.he RMS Franconia just west
-Stream proper.
of the fleet on June 21 afforded a detailed view of the
water structure at depth, shown in figure 4. Development
of the seasonal thermocline obscures the temperature
gradients associated with the Gulf Stream in the first 50
m, but below this depth, the north wall is plainly visible
inclining toward the south with a slope of abouf, 1:85.
To obtain an idea of surface currents, we calculated the
vertical average of temperature for each sounding over
the layer from 15 to 705 m, the greatest depth consistently
reached. Horizontal gradients of these averaged tempera-
tures are indicated in figure 3 by arrows emanating from
the track of the ship with length proportional to the in-
tensity of the gradient and with direction along the
expected geostrophic surface current. The precision of a
quantitative calculation is limited principally by the lack
of information on salinity. Nevertheless, we tried. From
Gulf Stream cross-sections shown by Neumann and
Pierson (1966, p. 132), it appears that the differences in
density anomaly across the Gulf Stream are determined
principally by differences in temperature and that the
opposing effect of salinity differences is about 40 percent
of the effect due solely to temperature. On this basis,
neglecting pressure effects upon specific volume anomalies
and with the further assumptions that the current van-
ishes at 705 m and is strictly zonal, we find a maximum
current of about 5.3 kt between 37'59' and 37'57". For
the Gulf Stream as a whole, we obtain an average current
of about 2.7 k t between 38'12' and 37'42". If, as is
likely the case, the current does not vanish at 705 m, then
our estimate is too small, whereas, if the current is from
August1972 1 Sanders 1 599
I BO' 70' I I
FIGURE 5.-NMC analysis for 0000 GMT on June 22, 1970. Solid
lines are sea-level isobars at intervals of 8 mb and dashed lines
are isopleths of thickness of the layer from 1000 to 500 mb, at
intervals of 6 dekameters. The hatched area represents the
approximate envelope of the racing fleet a t this time. The circled
X denotes the location of Atlantic City (ACY).
southwest t80 northeast (also likely), our estimate is then
too large. Thus, we may hope for cancellation of errors.
The presence of the Gulf Stream so far to the south of
the maximum anomalous warmth reported by the yachts-
men suggests that the meander may have been cut off
by this time. The reported surface temperatures in the
meander or eddy were about 4 O F lower than those in
the Gulf Stream from which this water came. Perhaps this
cooling represents a change toward the equiIibrium tem-
perature appropriate to the geographical location and
time of year once systematic advection of warm water has
been interrupted.
The meandering process that produced the warm eddy
also yielded a cold eddy observed by the RMS Franconia
on June 8 near 36ON, 69OW. Subsequent passages failed
to show any trace of this feature, nor were appropriate
current or temperature effects reported by the yachts.
Thus, it seems likely that this feature moved toward the
southwest where a temperature minimum is indicated
(fig. 2).
3. THE THUNDERSTORMS OF JUNE 21-22
The squalls of .June 21-22 were associated with a weak,
nondeveloping wave cyclone, as analyzed by the National
Meteorological Center (NMC). From the innocuous ap-
pearance of the map shown in figure 5, one would hardly
guess that violent squalls had swept through the fleet
about 134 hr earlier (fig. 6) and would again about 9 hr
later (fig. 7).
600 / Vol. 100, No. 8 / Monthly Weather Review
7iw
t 70.W
+ 69'W
f 40"
WINDWARD WS54GE
I
FIGURE 6.-Peak gust in the first series of squalls on June 21. Solid
lines are isotachs at intervals of 20 kt. Dashed lines are isochrones
of the peak gust at intervals of 30 min. Peak velocities reported
by each yacht are plotted in the conventional manner. The
reported time of squalls (EDT) is at the upper right of the station
circle. Reports of thunder or rain (f indicating heavy intensity)
from each yacht are plotted to the left of the station circle as
T or R, respectively.
The reports of the yachts in figure 6 form a reasonable
pattern as to peak windspeed, about which some subjec-
tivity would be expected. There appears to have been a
weak spot in the line of squalls from roughly %ON, 72OW
to 39.5'N, 69.5OW, since the yachts Comet, Enchanted,
Shadow, Seeadler, and Fling (the only ones reporting less
than 40 kt) all lay along this zone. Many yachts, including
the author's, experienced a sharp shift in wind direction
from southeasterly to southwesterly during the period of
squalls. The reported times, about which one might expect
objectivity, do not form a consistent pattern, probably
because the request for information was not sufficiently
specific concerning what event was to be timed: the be-
ginning, peak, or end of the squnll, or the windshift.
Nevertheless, it seems unlikely that the peak gust moved
through the fleet a t a speed much less than that indicated
by the analyzed isochrones, which is about 70 kt. By
analogy with experience in severe convective storms over
land, this speed of advance lends credibility to the reported
gusts in the range from 50 to 80 kt.
The second group of squalls (fig. 7) was experienced by
relatively few yachts because it was of rather small merid-
ional extent and the larger yachts had sped on far toward
the southeast, driven by brisk but steady southwesterly
winds. The northern boundary of the region of strong
gusts was extraordinarily sharp. Note that Snow Star ob-
served heavy rain and thunder but no wind a t a location
only about 30 n.mi. from gusts of near-hurricane force.
The reported times of this second set of squalls showed
1 i w
+
7dw 6 9-W
+ ', \+Rw$E?
I I I
0900GMT O93OGMT sw3B
FIGURE 7.-Same as figure 6 for the second series of squalls on
June 22.
considerable scatter about any' reasonably smooth pro-
gression. However, the average time of peak gust for the
westernmost seven yachts was 0900 GMT and for the ease
ernmost seven it was 0935 GMT, indicating an extremely
rapid progression.
What was the origin of these violent storms? It seems
natural to attribute them either to the Gulf Stream proper
or to the warm meander, which brought unusually high
sea-surface temperatures far to the north of the usual
position of the Gulf Stream. First, we examine the time
series of hourly observations at the XERB-1 buoy, moored
a t 36.5ON1 73.5OW on the northwestern edge of the Gulf
Stream. This location (fig. 5) is about 200 n.mi. upwind
from the position of the fleet. In this time series, shown
in figure 8, the two periods of squalls are apparent in peak
windspeeds, veering wind directions, pressure rises, and
temperature falls to levels lower than the local sea tem-
perature, due undoubtedly to cold downdrafts from
cumulonimbus cloud sys tems. The two periods of squalls
occurred at about 2030 GMT on June 21 and 0630 GMT on
June 22, in each instance 2 or 3 hr before they were
experienced by the fleet, confirming the great speed of
advance implied by the yacht data
Hourly surface observations at all reporting stations in
and immediately west of the Chesapeake Bay area also
show two distinct periods of showers or thunderstorms,
separated as at XERB-1 by about 10 hr. When these
observations are plotted on a representation of the plan
position indicator (PPI) scope of the WSR-57 radar at
Atlantic City, N.J., a striking picture emerges. A series of
these mesoanalyses at 3-hr intervals is presented in figure
9. The following locations are shown in the figure for
orientation: John F. Kennedy Airport, New York, N.Y.
(JFK), the XERB-1 buoy, Hatteras, N.C. (HAT), Nor-
folk (ORF) and Charlottesville (CHO), Va., Martinsburg,
? 360 -
320 - 280- West
240-
Wind South n \
!LA A
160- Direction
(deg true1 lZo-
80-
40-
East "V
3601 I
28 -
A i r 26 - (Sea Temp)
Temp 24 W \
(dag C) t I I
I
Second squolls First squalls
4- June 22 c- June 21 - I I I I I
24 16 08 00 16 08 00
GMT
FIGURE S.-Time-series of hourly observations from the XERB-1
buoy at 36.5"N, 73.5OW on June 21-22, 1970.
W. Va. (MRB), and Washington National Airport, Wash-
ington, D.C. (DCA).
A meso-Low just west of Washington, D.C., at 1500
GMT (fig. 9), produced a great mass of convective echoes
associated with thunderstorms and severe surface winds
3 hr later. In the surface air over eastern Virginia that fed
these storms, temperatures were in the upper 80s and dew
points in the 70s. Numerous radar echo tops were reported
in excess of 40,000 f t elevation. Deaths, injuries, and ex-
tensive damage to vehicles, buildings, and trees were
reported by the Environmental Science Services Admin-
istration (1970). The most severe conditions (including
tornadoes and waterspouts) were observed in the elon-
gated cold frontlike line of radar echoes in the southern
part of the display. Between 1800 and 2100 GMT, the line
became extremely solid and powerful in appearance, even
at distances more than 100 n.mi. from the radar site.
During this time, the leading edge of the echoes moved
at a speed of about 50 kt. This value is intermediate
between the somewhat slower rate of advance of the line
of peak surface gusts over land and the somewhat faster
rate required to bring the line over the ocean to the fleet
at the reported time of peak wind.
The origin of the second group of squalls is seen be-
tween 2100 and 0000 GMT. Once again a meso-Low formed
just west of Washington, in response to which a new group
of echoes quickly appeared, again in the region just east
of the Low. The surface air feeding these storms had ap-
proximately the same thermodynamic characteristics as
the air involved in the first storms. This group of storms,
however, appeared on radar to be more scattered and of
more limited north-south extent. Again, the rate of ad-
August1972 J Sanders J 601
/
I I
FIQURE 9.-Mesoanalyses of sea-level pressure at 3-hr intervals on June 21-22,1970, superposed on echoes observed by the WSR-57 radar
at Atlantic City (ACY). The PPI scope photograph nearest the nominal time of the 3-hourly surface observations was used. Thin
solid lines are range circles at intervals of 50 n. mi. Dashed lines are sea-level isobars a t intervals of 1 mb. Circled Xs show approximate
centroid of positions of yachts reporting squalls.
vance of the envelope of echoes from 0300 to 0600 GMT
was about 50 kt, the speed required for arrival of the
storms in the fleet area at about 0900 GMT. Finally, the
small meridional extent of the storms agrees well with
limited north-south extent of strong squalls reported by
the yachts.
S.ynoptic analyses of later conditions at sea (not shown)
indicate that both groups of squalls subsequently slowed
and weakened but that the first was still producing
thunderstorms a t 1200 GMT on June 22.
It is clear that neither set of thunderstorms originated
over the Gulf Stream. Rather, they developed in ex-
tremely warm and moist air over land during the daytime
hours. The air was sufficiently warm in its lower levels
that passage over the Gulf Stream itself, to say nothing
of the warm meander or eddy to the north, would have
produced surface cooling and stabilization. Evidently,
the charge received by the air over land was sufficient to
maintain severe convective storms for 12-18 hr after the
removal of the source of surface heating and moistening.
The first group of storms clearly displayed its greatest
radar intensity (fig. 9) after it passed offshore, and the
highest cumulonimbus towers of each group were still
visible on radar at a distance of almost 250 n.mi. seaward.
Comparable persistence of convection above surface sta-
bility is often observed with severe nocturnal storms in
the eastern portion of the Great Plains States. Perhaps,
as suggested by Eessler and Bumgarner (1971), surface
602 Vol. 100, No. 8 1 Monthly Weather Review
stabilization tends to promote the organization of con-
vective activity into large, hence severe, elements.
The initiation of each group of storms immediately
after the formation of a mesoscale Low and immediately
downwind (in the upper flow) from the Low is not likely
a coincidence. The Lows formed in a region of synoptic
scale ascent associated with a shortwave upper trough, in
a region downwind from the crests of the Appalachian
Mountains (thus favorable for orographic cyclogenesis)
and in a region of strong, localized diurnal heating (thus
favorable for thermal cyclogenesis). The thunderstorms
formed in a region where we might reasonably expect
locally enhanced warm advection, because of the pre-
sumed quasi-geostrophic cbaracter of the flow associated
with the mesoscale Lows, and in a region where we might
expect pronounced frictional boundary-layer convergence
because of enhanced low-level vorticity associated with
the mesoscale Low. Thus, the air rising in the cumulonim-
bus towers might be associated either with moisture
convergence brought about by large-scale motions above
the surface boundary layer or with such convergence pro-
duced within the layer as suggested by Charney and Elias-
sen (1964) or both, In short, there was every qualitative
reason to anticipate development of thunderstorms.
In the problem of forecasting severe convection, the
suggestion by Fawbush et al. (1951) that strong vertical
wind shear and convective instability are necessary in-
gredients has become generally accepted. These aspects are
+
+ 75 + + 65 70
FIQURE 10.-Showalter stability index at intervals of 4°C (solid
lines) and isotachs of wind shear for the layer from 850 to 200 mb
at intervals of 10 kt (dashed lines), for 0000 OMT on June 22, 1970.
The observed index value is plotted above and to the left of the
station circle; the wind shear vector is plotted in the conventional
manner. DIA is Dulles International Airport.
displayed for 0000 GMT in figure 10. Stability is repre-
sented by the Showalter index (ambient 500-mb temper-
ature minus the temperature of the 850-mb parcel lifted
dry adiabatically to its condensation level then moist
adiabatically to 500 mb), while wind shear is shown for
the layer from 850 mb, in the lower part of the clouds, to
200 mb, near the level of maximum wind. Generally
speaking, the region of strong shear lies to the north of the
region of instability. Over central and western Virginia
and North Carolina, however, we find negative Showalter
index together with shear greater than 40 k t (hatched area
in fig. 10). At this time, the second group of squalls was just
becoming organized, with radar echoes in the vicinity of
Dulles International Airport and thunder reported at
nearby stations to the south and west within the hatched
area. Twelve hours earlier, Dulles was the only station
east of the Appalachian Mountains meeting the above
shear and stability criteria; the f i s t group of squalls be-
came organized in this area a few hours after this time.
As nearly as can be judged, both groups of squalls ad-
vanced at speeds somewhat faster than the speed of the
large-scale tropospheric mean flow.
4. THE GALE OF JUNE 25
The gale of June 25 was of quite different character.
None of the yachts mentioned thunder and only a few
reported rain. The NMC surface analysis for 1200 GMT
(fig. 11) shows a weak frontal wave with a Low center
near 36'N, 67OW. The paucity of thickness lines in the
vicinity of this system suggests horizontal temperature
gradients in the lower and middle troposphere of less than
0.5'C/100 km. The only indication of more than gentle
FIQURE 11.-NMC surface analysis for 1200 OMT on June 25, 1970.
Format is the same as in figure 5. The hatched area is the
approximate envelope of the racing fleet at this time.
zephyrs is the 25-kt geostrophic wind implied by the
spacing of isobars east of the front between 30' and 35'N.
Nevertheless, a Nimbus satellite view about 8 hr before
the time of the map (fig. 12A) shows abundant cloudiness
alined along the analyzed frontal position. Another
view about 4 hr after the time of the map (fig. 12B)
shows dissipation of much of the cloudiness south of 30'N
and a more chaotic pattern to the north. Nothing suggests
the presence of a gale.
Thirteen yachts provided accounts of their experiences
on this date in various formats. For each of them, how-
ever, it was possible to construct a time history of latitude,
longitude, and wind. From these and from the surface
observations a t Kindley Field, Bermuda, a series of de-
tailed analyses a t 3-hr intervals were constructed. These
are shown in figure 13. We see clearly the northeastward
passage of a cyclone through the fleet a t a rate of about
23 kt, being about 85 n.mi. northwest of Bermuda a t its
closest approach. Between the island and the Low center
lay a narrow zone of gale-force winds never more than
50 n.mi. wide that blasted some of the yachts but left
others, farther to the northwest, with either a good
sailing breeze or With light air. The position of the Low
on the 1200 GMT NMC map was evidently in error by
some 300 n.mL or so.
The difference in wind speeds between the southeast
and northwest sides of the cyclone was about 40 kt.
If we imagine the cyclone as an axisymmetric vortex
superposed upon a uniform basic current, then the advec-
tion of relative vorticity (we can neglect the advection
of earth vorticity for a system this small) would indicate
a rate of advance of half the speed difference, or 20 kt.
August1972 1 Sanders 1 603
FIQIJRE 12.-Nimbus satellite (A) high resolution infrared radiometer picture at about 0400 QMT and (B) TV picture at about 1600 QMT
on June 25, 1970. Grid dots represent latitude and longitude at intervals of 2". The fiducial mark is at 30°N, 67" W in both pictures.
i-
FIQURE 13.-Detailed surface analyses at 3-hr intervals from 0900 to 2400 QMT on June 25. Wind is plotted for all yachts and for Kindley
Field, Bermuda, with annotations where available concerning sky conditions and precipitation. Analyses include streamlines of wind
flow (solid lines), isotachs at intervals of 10 kt (dashed lines), and the estimated track of the cyclone with hourly positions indicated
(heavy dot-dash line).
Since the actual speed is only slightly faster, we can con-
clude that convergence and divergence effects associated
with baroclinicity, the predominant mechanism in the
604 1 Vol. 100, No. 8 1 Monthly Weather Review
motion of most extratropical cyclones, were quite un-
important in this case and that the cyclone moved rather
in the manner of a tropical storm.
FIGURE 14.-Twenty-four hr forecast sea-level isobars and thickness lines for the layer from 1000 to 500 mb, derived from the PE model
for (A) 0000 GMT, June 22, and (B) 1200 QYT on June 25, 1970. Format is the same as in figure 5.
It is difEcult to argue that the cyclone was maintained
by latent heat release as tropical cyclones are, since
precipitation was sparse. The only heavy rainfall reported
was in the bright cloudband on the eastern periphery
of the gale, shown in figure 12B, from which Kindley
Field received 0.60 in. of rain in showers. It is possible,
however, that the storm was dissipating at the time it
passed through the fleet, for later NMC maps omitted
the weak Low and frontal system altogether. However
the storm was initiated, its origin was doubtless within
the massive cloud system shown in figure 12A. We, there-
fore, tentatively conclude that the storm was principally
tropical in character, with weak baroclinic characteristics;
and we wonder how many similarly small and short-lived
systems go unobserved across the expanses of the global
ocean. Careful study of an ATS 3 film loop for June 25
does, in fact, disclose rotation of cloud around a small
central hub that can be identified with the small bright
cloud area near 33"N, 67"W in the Nimbus photograph in
figure 12B, but no identifiable cloud targets gave a hint
of maximum strength of the surface winds.
5. THE FORECASTS
Two questions occur concerning the forecasts for June 22
and 25. First, what effect did the convective storms have
on the predictions for the larger synoptic scales of motion?
Relevant 24-hr predictions of the patterns of sea-level
pressure and of thickness of the layer from 1000 to 500
mb, derived from the NMC primitive-equation (PE)
model (Shuman and Hovermale 1968), are displayed in
figure 14. Comparison of figure 14A with figure 5 indicates
that the actual position of the complex low-pressure system
along the mid-Atlantic coast was east of the predicted
position. It does not seem reasonable to blame this dis-
crepancy on the convective storms, however, because
virtually all the Highs and Lows shared this prognostic
defect whether or not they were associated with cumulus
activity. The slowness of predicted eastward displacement
of mobile Highs and Lows is a generally recognized bias
in the PE forecasts. Comparison of figures 14B and 11
shows that the forecast in the vicinity of Bermuda on
June 25 was as good as the analysis. The paucity of
observations conspired with the sparsity of computational
gridpoints to conceal the existence of the gale before,
during, and after the fact. In neither case was there
evidence that the tropospheric mean temperature, as
measured by the thickness, was significantly influenced on
the large scale-by the release of latent heat in the cumulus.
Nor, in either case, was there predicted or observed inten-
sification of the relevant synoptic scale features subsequent
to the verification times shown in figure 14. We conclude
then that, important as the small-scale storms were to
the participants in the race, they were (so far as immediate
and direct effects were concerned) mere embroidery on the
large-scale fabric of meteorological events.
The second question is how well the all-important
smaller features were predicted in the forecasts available
to the racing crews. We have no documentation concerning
the forecasts around Bermuda on June 25. For the thunder-
storms on June 21-22, however, we can refer to the
National Weather Service West Central North Atlantic
Offshore Forecast for the area between 35" and 41"N and
west of 65"W, the product probably most widely used by
the crews at this time. This forecast is issued at 0400
GMT each day and at 6-hr intervals thereafter.
The forecast issued a t 1000 GMT on June 21 (12 hr
before the first squalls struck the fleet) stated that "a
low [in] western Indiana will move . . . to northeastern
Pennsylvania by Monday [June 221 morning." This
August1972 1 Sanders 1 605
473-110 0 - 72 - 2
projection reflected the slowness of the PE forecast
described above and did not entirely compensate for it.
“Winds will veer on Sunday [June 211 becoming generally
southeast 5 to 15 knots over the southwest portion and
northeast to east 5 to 15 knots elsewhere by Sunday
afternoon.” Rain was forecast over the southwest portion,
but there was no mention of thunderstorms in the vicinity
of the fleet.
The forecast issued at 1600 GMT (6 hr before the arrival
of squalls and at the time of initial thunderstorm forma-
tion in the vicinity of Washington) refers to “low pressure
over the Middle West [moving] eastward 20 knots to
southeastern New York by Monday morning.” The
revised forecast position of the Low was an improvement.
Another change was the introduction of a warm front
crossing the rhumb line at 36’N and predicted to move
northward to 39’N by 1200 GMT. “South of and near warm
front winds south to southwest 15 to 25 knots though
locally stronger in gusts near thunderstorms through
Monday morning.”
The forecast issued at 2200 GMT (as the first squalls
struck) described a “complex low with several centers
from the upper Ohio Valley to Delaware Bay [that will]
become better organized in vicinity of Nantucket by
daybreak. . . .” Again there was an eastward shift of the
low-pressure system. The warm front again served as the
basis for describing the weather; thus “south of frontal
system winds mostly southwest 15 to 25 knots but briefly
higher in thundersqualls.”
The forecast issued at 0400 OMT on June 22 (between
the two squall episodes in the fleet) spoke of a “complex
low over middle Atlantic states with several centers . . .
[becoming] better organized during the night and [moving]
to 41N 67W by 8am [1200 GMT].” This was the final
eastward modification of the predicted position of the
Low center. Over the fleet, the forecast was for “winds
southwest 15 to 25 knots but briefly higher in thunder-
squalls.” Virtually the same language was used to describe
the winds in each ensuing forecast through at least
0400 GMT on June 23.
Note that these forecosts were quite detailed and
specific so far as the location and position of the low-
pressure center were concerned. In contrast, the reference
to the associated weather, in which the seaman is
interested, was very general (i.e., “southwest 15 to 25
knots but briefly higher in thundersqualls” was used
with only minor variation throughout a period of 36 hr,
during which the description of the Low center changed
each 6 hr). Indeed, the average weather experienced by
the fleet was close to that predicted. But the important
events were on the mesoscale, as is so often the case, and
the land-based forecaster generally lacks the data neces-
sary to cope with such occurrences at sea. In the present
case, however, i t is interesting to speculate whether, for
example, the consolidation and rapid eastward motion
of the echoes seen in figures 9B and 9C, with reported
tops as high as 50,000 f t , might have served as the basis
for a specific short-range forecast of gusts in the vicinity
of 50 kt, had sufficient and timely radar information been
available to the forecaster.
The focus of. the forecasting effort needs to be changed
from the synoptic scale to the mesoscale. The forecasts
in this case were not inaccurate; they were irrelevant.
ACKNOWLEDGMENTS
This study would not have been poesible without the information
and advice provided by the following skippers and navigators, to
whom the author is especially grateful: Armando Grandi, Atrevido;
A. Wallace Barr, Brambers; Wells Moms, Carrillon; Robert
J. Brockhurst, Comet; James M. Clark and Ross Pilling,
Cosette; Elihu E. Allinson, Dragon Lady; Randolph P. Barton and
Everett MOWS, Jr., Enchanted; Henry M. Chance 11, Fling; Mark
C. Ewing, Harpoon; Charles L. Ill and William 0. Price, Ill Wind;
Anthony W. Parker, Jubilee I I I ; Robert W. Hubner and A. Peter
Quinn, Jr., Kate; Humphrey B. Simpson, Kittiwake; R. C. McCurdy,
Mah Jong; John H. Page and Harvey Loomis, Pageant; Richard T.
duMoulin, Rage; T. Vincent Learson, Nepenthe; Benjamin B.
duPont, Rhubarb; Frederick E. Hood and Joseph Chase, Robin;
Walter S. Frank, SaUy Tiger; Howard W. Read, SeeadEer; Charles
T. Sturgess, Shadow; John C. Kiley, Jr., Snow Star; W. Dorwin
Teague, Soufle; W. Van Alan Clark, Jr., and Irving C. Sheldon,
Swamp Yankee; James R. Shepley, Syriene; Stuyvesant Wainwright
111, Wainscott Flame; E. M. Gosling, Whistler of Paget; and Harvey
White, White Caps.
In addition, I express my thanks to John J. Schule, Jr., U.S.
Naval Oceanographic Office, for providing detailed data from the
RMS Franconiu; David B. Spiegler of Allied Research Associates,
and Donald C. Gaby of the NOAA National Hurricane Center for
providing satellite pictures; Charles E. Mason I11 of Sail magazine
for making available his collection of data from skippers; and the
Bermuda Race Committee, Cruising Club of America, for its co-
operation and assistance.
The investigation was supported by the Atmospheric Sciences
Section, National Science Foundation, under NSF Grant GA-
27908X.
REFERENCES
Charney, Jule G., and Eliassen, Amt, “On the Growth of the
Hurricane Depression,” Journal of the Atmospheric Sciences,
Vol. 21, No. 1, Jan. 1964, pp. 68-75.
Environmental Science Services Administration, Storm Data, Vol.
12, No. 6, Environmental Data Service, Asheville, N.C., June
1970, pp. 68-94.
Fawbush, Ernest J., Miller, Robert C., and Starrett, Lloyd G.,
“An Empirical Method of Forecasting Tornado Development,’’
Bulletin of the American Meteorological Society, Vol. 32, No. 1,
Jan. 1951, pp. 1-9.
Kessler, Edwin, and Bumgarner, William C., “Model of Precipita-
tion and Vertical Air Currents,” NOAA Technical Memorandum
ERL NSSL-54, U.S. Department of Commerce, National
Severe Storms Laboratory, Norman, Okla., June 1971, 93 pp.
Neumann, Gerhard, and Pierson, Willard J., Jr., Principles of
Physical Oceanography, Prentice-Hall, Inc., Englewood Cliffs,
N.J., 1966, 545 pp.
Shuman, Frederick G., and Hovermale, John B., “An Operational
Six-Layer Primitive Equation Model,” Journal of Applied
Meteorology, Vol. 7 , No. 4, Aug. 1968, pp. 525-547.
U.S. Naval Oceanographic Office, The Gulf Stream Monthly Sum-
mary, Vol. 5, No. 6, Washington, D.C., June. 1970.
U.S. Naval Oceanographic Office, The Gulf Stream Monthly Sum-
mary, Vol. 6, No. 1, Washington, D.C., Jan. 1971.
[Received November 29, 1971; revised March 13, 19721
606 1 Vol. 100, No. 8 1 Monthly Weather Review