þÿ

Satellite Images Show the Movement of Floating Sargassum in the Gulf

of Mexico and Atlantic Ocean

Jim Gower and Stephanie King

Fisheries and Ocean Canada,

Institute of Ocean Sciences, PO Box 6000, Sidney BC, Canada V8L 4B2

Email: gowerj@pac.dfo-mpo.gc.ca

Summary
The question of the origin, distribution and fate of the floating seaweed Sargassum has
fascinated sailors and scientists from the time of Columbus. Observations from ships are
hampered by the large and variable area over which Sargassum is dispersed. Here we use
satellite imagery to present the first mapping of the full distribution and movement of the
population of Sargassum in the Gulf of Mexico and western Atlantic in the years 2002 to
2008. For the first time, we show a seasonal pattern in which Sargassum originates in the
northwest Gulf of Mexico in spring of each year, is advected into the Atlantic in about
July, appearing east of Cape Hatteras as a "Sargassum jet," and ending northeast of the
Bahamas in February of the following year. This pattern appears consistent with
historical surveys. Future satellite observations will show whether this pattern repeats in
all or most years.

Introduction
Although the free-floating pelagic species of Sargassum (natans and fluitans) have been
studied since at least the 1830's and have been acknowledged in marine lore by the
naming of the Sargasso Sea, they have only recently been detected in satellite images

Observations from ships are hampered by the large and variable area over which
Sargassum is dispersed. For satellites, this is not a problem since the area of coverage is
almost global and is regularly repeated. Also, Sargassum should be an ideal target for
optical satellite sensors. It is long-lived, buoyant and has a spectral signature which
contrasts strongly with surrounding water. Aggregations are extensive enough to be
detected by relatively low-resolution sensors. The major reason for its non-detection in
the past has been the lack of a combination of sensor bands that provides a definitive
signal in the presence of cloud, haze and sun glint. This is now rectified by the ESA's
(European Space Agency) MERIS sensor

The MERIS Imager on the Envisat satellite showed extensive areas of long, narrow,
meandering slicks in the north-western Gulf of Mexico in the early summer of 2005

We have now extended the survey to global coverage for the time period June 2002 to
April 2008. Results show an average annual cycle in Sargassum distribution in the Gulf
of Mexico and North Atlantic, with considerable interannual variability. The satellite
data do not show any sign of similar populations of pelagic Sargassum in other oceans of
the world.

Nature Precedings : hdl:10101/npre.2008.1894.1 : Posted 15 May 2008

Satellite image data
The Medium Resolution Imaging Spectrometer (MERIS) was launched on the European
Space Agency's Envisat satellite in March 2002 and has provided systematic global
coverage at 1200 m resolution since June of that year. To be detected, Sargassum
therefore has to be dense enough, and to cover a large enough area, to affect the average
color (visible surface spectral reflectance) of an area of ocean surface 1200 m across.
We make use of an index, MCI (Maximum Chlorophyll Index), which provides good
discrimination of floating and coastal vegetation, as well as intense surface plankton
blooms

. MCI is computed from the above-atmosphere spectral radiances measured for

each pixel of the satellite image data to show excess radiance at 709 nm, above a baseline
defined by linear interpolation between the two neigbouring bands at 681 and 754 nm, as
defined by equation 1.

MCI = L

709

681

(709 681)*(L

754

681

)/(754 681) .... (1)

where L

709

represents radiance at 709 nm, etc.. Pixels containing significant cloud, land

or sun glint are screened out by accepting only pixels for which L

865

is less than a

threshold value.
For large-area tracking of Sargassum, we make use of global, daily composites of the
MERIS data at 5 km spatial resolution. The value of each pixel in the composite is the
maximum MCI of any RR image pixel assigned to that composite pixel. The daily
composites are combined into monthly images in which each pixel shows the maximum
MCI recorded at that pixel on any day of the month. This fills areas missed due to cloud,
sun-glint, and lack of MERIS coverage, while preserving any evidence of Sargassum
occurrence.
The monthly composites of MCI signal at 5 km spatial resolution are analyzed by
computing the frequency distribution (histogram) of MCI values in each one-degree
square, and assuming that all MCI values exceeding the mean ocean background value in
that one-degree square by a threshold amount (here 0.4 mW/(m

.nm.sr)), indicate

presence of Sargassum. Squares that include coastlines and other fixed areas where MCI
is observed to be high, such as coral reefs and areas with frequent coastal plankton
blooms, are masked in all months. We name the number of MCI values above threshold,
multiplied by the amount by which MCI exceeds its background value (in mW/m

.nm.sr)

as "MERIS count." We take this count as being proportional to the total amount of
Sargassum in each one-degree square.

Satellite observations of Sargassum distribution
Since the first detection using satellite imagery of Sargassum in the north-western Gulf of
Mexico in May and June of 2005, we have collected several images of dense
aggregations in the Gulf Stream extension area of the western Atlantic, east of Cape
Hatteras in October and November of 2006 and 2007, such as Figure 1. The difference
spectrum in the inset (blue) shows the "red edge" characteristic of land vegetation, with a
shift in wavelength due to water absorption that results in a value of MCI of about 2.0
mW/(m

.nm.sr). The peak in the visible spectrum (400 to 700 nm) at 620 nm is

consistent with the brown colour of Sargassum. Interpretation as Sargassum is also

Nature Precedings : hdl:10101/npre.2008.1894.1 : Posted 15 May 2008

based on the shape of the patches (especially the lines extending over 100 km in length),
continuity of patterns over a several-month period, and lack of indication in the satellite
data of any high background population of phytoplankton.
Figure 2 shows images of MERIS counts of Sargassum detected by MERIS MCI for the
years 2002 to 2007 (top to bottom) and months July to September (left to right), in pixels
measuring one degree in latitude and longitude. High concentrations of Sargassum are
indicated in the northwest Gulf of Mexico for all years in March to June, except 2002 for
which we have data only after June 12. Sargassum then appears in a broad area of the
Atlantic to the east of Cape Hatteras (35 to 40N, 45 to 75W) in July and August, as best
shown in 2005, 2006 and 2007.
Figure 3 shows plots of estimated total amounts of Sargassum in the Gulf of Mexico (15
to 30N, 80 to 100W) and the western North Atlantic (22 to 40N, 20 to 80W), for each
month from June 2002 to April 2008. MERIS counts are scaled to millions of tons by
comparison with ship observations

4,5

as described below. Highest amounts are in May,

June, July in the Gulf of Mexico, with values increasing through 2003 and 2004 to a
maximum in 2005. Amounts in the Atlantic increase each year after July and drop back
to low values in the spring of the following year.
The data clearly indicate strong growth early in the year in the Gulf of Mexico, with
Sargassum moving from there into the Atlantic each year in July and August. Passive
surface floats

take about a month to be advected by the Loop Current and Gulf Stream

from the north-east Gulf of Mexico near 27N, 85W to east of Cape Hatteras at 36N, 75W.
This is a short enough time to be consistent with our interpretation of the movement of
Sargassum.
The average latitude and longitude of Sargassum detected in the Atlantic were computed
using a simple linearly weighted average over the area 22 to 40N, 20 to 80W. Values are
plotted in Figure 4 for months in which Sargassum amounts in Figure 3 were greater than
700,000 tons. The seasonal variation shows considerable consistency from year to year.
The statistical centre of Sargassum is first at about 37N, 67W in the Atlantic, moves
eastwards until October and then moves southwest until February.
The average spatial pattern of the annual cycle is shown schematically in Figure 5. In
each year, Sargassum is first detected in a small area of the northwest Gulf of Mexico in
March, which expands and spreads eastwards. In July Sargassum is present in both the
Gulf and the Atlantic off Cape Hatteras, spreading eastwards to about 45 W by
September, then drifting south and west. Counts are very low in the Atlantic for the
months of March, April and May, though observations to April of the present year show
for the first time at this season, significant Sargassum northeast of the Bahamas.

Comparison with observations from ships
Many sightings of Sargassum were recorded in the 19

century, especially by the

German merchant marine, summarized by Kruemmel

, perhaps inspired by the early

theory that Sargassum indicated presence of a vast undiscovered reef in mid-Atlantic.
Winge

in a historical summary notes that the botanist Meyen, writing of "A journey

round the world in 1830, 1831 and 1832" was the first to suggest that this Sargassum is

Nature Precedings : hdl:10101/npre.2008.1894.1 : Posted 15 May 2008

truly pelagic. Winge also quotes Meyen as writing (in German) "Some sailors believe
that this weed is collected by the Gulf Stream and that there are huge masses of seaweed
in the Gulf of Mexico, an opinion that however, does not need to be considered further."
In fact, Winge's summary shows that the idea of Sargassum originating in the Gulf of
Mexico, as we propose, was fairly common in the 19

century, but seems to have died

out in the 20

Parr

reported on 194 surface net tows designed to collect Sargassum in the Sargasso Sea,

Caribbean and Gulf of Mexico, in 1933, 1934 and 1935. Of these, 160 were in January,
February or March, and the remaining 34 in April, July and August. Given the
sparseness of his observations and lack of seasonal coverage, he used Kruemmel

estimates of the seasonality and average area of Sargassum. In late January of 1934 and
early February of 1935, Parr measured Sargassum amounts on cruises from Cape Cod to
Bermuda to the Caribbean. His observations were frequent enough, over a wide enough
range of latitude to allow an estimate of the mean latitude (about 27 N) of the Sargassum
at these times (letter P on the left panel of Figure 4).
Between April 1977 and January 1982, Stoner

measured Sargassum density at 266

locations on a series of 15 cruises in the Sargasso Sea and Caribbean. The results were
initially interpreted

to show a drop in Sargassum biomass compared to Parr

's earlier

observations, but were later reinterpreted

as showing a significantly lower value in only

one area near 23 N, 65 W north of Puerto Rico, which Stoner visited in November of
1977 and 1980, and Parr visited in February/March 1933. Butler and Stoner

suggested

that this difference may be due to a seasonal variation. Our satellite data now confirms
this, showing that the density at this location in November should indeed be lower than in
February (Figure 5).
We select areas and months where Parr

and Stoner

measured significant amounts of

Sargassum, and compare these with the satellite observations for the same areas and
months. This means we are comparing ship and satellite observations made 70 and 25
years apart in time, but we know of no more recent surveys which would allow
comparisons with a smaller time interval. The results for 11 different areas and dates
give an average value of 1400 tons per square degree per MERIS count, with r.m.s.
scatter of about a factor two among the 11 estimates of this value. We use this average
value to compute the total amount of Sargassum in Figure 3.
Our interpretation of significant amounts of Sargassum entering the Atlantic from the
Gulf of Mexico in the summer is supported by the observations of Dooley

who

collected Sargassum passing Miami at semi-monthly intervals from April 1966 to May
1967, and noted "very low quantities in spring and winter, while tremendous quantities
were available in summer and fall."
The satellite image data suggest that Sargassum amounts are regularly greater in the Gulf
of Mexico than in the Atlantic (Figure 3). There have been few reported surveys of
Sargassum in the Gulf, and no systematic surveys of seasonal variation, but sightings are
common, especially washed up on beaches. In a single visit to Corpus Christi Texas on
April 1 1935, Parr

noted a "secondary maximum" of Sargassum in the northwest Gulf of

Mexico, but observed it to be of "apparently deteriorating weeds." Wells and Rooker

Nature Precedings : hdl:10101/npre.2008.1894.1 : Posted 15 May 2008

state that Sargassum occurred in the northwest Gulf from May to August 2000, a
seasonal pattern confirmed by Figure 3.
Our interpretation of the satellite data is that Sargassum starts growing each year in the
Gulf of Mexico in about March, and dies about a year later in the Atlantic in the area
northeast of the Bahamas. The idea of new growth of Sargassum in the Gulf in March
and April is contrary to Parr's conclusion that the weed there was "deteriorating," but a
rapid increase in the amount of Sargassum in the northwest Gulf each year in March to
July is very clear in the satellite data.
Our estimates for average total mass of Sargassum, derived by calibrating the satellite
data with ship measurements in the same areas and months, is about 1 million tons in
each of the Gulf of Mexico and the Atlantic, for a total of 2 million tons (averages of the
data plotted in Figure 3). This is less than the 7, 11 and 4 million tons estimated by Parr

for each of the three years of his observations (1933, 1934 and 1935).

Errors in satellite estimates of Sargassum amounts
There are several possible sources of error in our satellite estimates of Sargassum
amounts. The satellite covers only about half the earth's surface each day in the tropics,
and the area of useful observations is further reduced by cloud and sun glint. On the
other hand, areas that remain hidden from the satellite on all days of a given month are
relatively rare. Sargassum that is evenly distributed may not exceed our detection
threshold, so that the satellite may be detecting only Sargassum that is to some extent
"aggregated." MERIS can detect Sargassum using MCI, only when it is at or just below
the sea surface. At the wavelength of 709 nm, absorption by water is 1.0 per metre, so
that Sargassum at a depth of 35 cm will give an MCI reduced to about half its value at the
surface. Aging Sargassum loses buoyancy and will be subject to an increasing tendency
to be mixed down by wind, waves or currents

. Woodcock

showed that Parr's counts

tended to be lower at wind speeds above 4 m/s, and a similar effect may be expected for
the satellite observations. Our statistical analysis preserves the largest MCI signal in a
month, so that even if wind is a significant factor, a patch of Sargassum needs only one
low-wind day in the month to be detected.
The above errors are cause for concern, but are mostly unavoidable. In all cases they are
the same for all months and locations, and will distort the relative values only if mixing
and surface aggregation are regionally or seasonally dependent. The conversion of the
MERIS counts to tons of Sargassum makes use of a statistical comparison with available
ship data, which compensates for some errors.
Cosmic ray hits on the sensors of the MERIS instrument will give sporadic high MCI
values which will be preserved in our statistics. These are probably responsible for some
of the isolated MERIS counts in Figure 2 and may be responsible for a significant
fraction of the low count values in Figure 3. Hits on the sensor are much more common
in the South Atlantic Anomaly, which affects an area about 2000km across, centred off
the coast of Brazil.
In areas of the world outside the region shown in Figure 2, high values of MCI

show

shorter-lived patches which we relate to intense surface plankton blooms of the type

Nature Precedings : hdl:10101/npre.2008.1894.1 : Posted 15 May 2008

commonly referred to as "red tides." We do not observe patterns with the longer
temporal continuity of Sargassum in any other ocean areas, confirming that a major
population of pelagic Sargassum is found only in the Gulf of Mexico and western
Atlantic. We note that benthic vegetation and coral reefs in shallow water

can cause

false positive signals, but these are limited to fixed and known locations.

Conclusions
Our observations show a large increase in Sargassum in the northwest Gulf of Mexico
between March and June each year, and low total Sargassum amounts in the Atlantic
before the annual injection from the Gulf of Mexico in July. This suggests that most
Sargassum has a life-time of one year or less, with the major "nursery area" being in the
northwest Gulf of Mexico. If Sargassum were longer-lived, we would expect there to be
some circulation of the Sargassum observed northeast of the Bahamas in February, back
into the Gulf Stream and then to the area northeast of Cape Hatteras. This would be
consistent with the traditional picture of the Sargasso Sea as being the main repository of
this biomass, however, we have not observed this Sargassum in satellite imagery for May
and June.
This observation of a significant average flow of about one million tons of Sargassum out
of the Gulf of Mexico each year implies a carbon flux which needs to be accounted for in
productivity and carbon models.
Satellite images clearly provide greatly improved data coverage compared to ship surveys
of Sargassum, but with limitations due to spatial resolution, cloud cover and sun glint.
We note that the satellite may miss significant quantities of Sargassum if it is too evenly
distributed or mixed beneath the surface by wind. In the future, satellite observations can
continue to provide a lengthening time series of data of the type we present here.
Satellites can also play an important role in selecting the sampling pattern for any future
ship survey.
Future observations of Sargassum depend on maintaining the capability provided by
MERIS. The present US sensors SeaWiFS and MODIS and the planned future sensor
VIIRS lack the band at 709 nm which make possible the computation of MCI, used here
for detection of Sargassum.

Acknowledgements
This work was supported by Fisheries and Oceans Canada, by the Canadian Space
Agency (CSA) under the GRIP (Government Related Initiative Program).

Nature Precedings : hdl:10101/npre.2008.1894.1 : Posted 15 May 2008

References
1. Gower, J.F.R., C. Hu, G.A. Borstad and S. King, 2006, Ocean color satellites show
extensive lines of floating Sargassum in the Gulf of Mexico, IEEE Transactions on
Geoscience and Remote Sensing, 44, 3619-3625.
2. ESA (2008),

http://envisat.esa.int/instruments/meris/

3. Gower, J., S. King, G. Borstad and L. Brown. (2005). Detection of intense plankton
blooms using the 709 nm band of the MERIS imaging spectrometer, Int. J. Remote Sens.
26, 2005-2012.
4. Parr, A.E. (1939). Quantitative observations on the pelagic Sargassum vegetation of
the western north Atlantic, Bulletin of the Bingham Oceanographic Collection, Peabody
Museum of Natural History, Yale University, 6 (7), 94pp.
5. Butler, J.N., B.F. Morris, J. Cadwallader and A.W. Stoner, 1983, Studies of
Sargassum and of the Sargassum community, Bermuda Biological Station, Special
Publication No. 22.
6. MEDS (Marine Environmental Data Service, 2008) drifting buoy data available at

http://www.meds-sdmm.dfo-mpo.gc.ca/meds/Databases/DRIBU/drifting_buoys_e.htm

7. Kruemmel, O., (1891) Die nord-atlantische Sargassosee, Petermanns Geographische
Mittheilungen, 37, 129-141.
8. Winge, O. (1923). The Sargasso Sea, its boundaries and vegetation, Report on the
Danish Oceanographical Expeditions 1908-10 to the Mediterranean and Adjacent Seas,
Volume III, Miscellaneous paper number 2, 34pp, Copenhagen.
9. Stoner, A.W., 1983, Pelagic Sargassum: Evidence for a major decrease in biomass,
Deep Sea Research, 30, 469-474.
10. Butler, J.N. and A.W. Stoner. (1984). Pelagic Sargassum: has its biomass changed in
the last 50 years? Deep-Sea Research 31, 1259-1264.
11. Dooley, J.K. 1972, Fishes associated with the pelagic Sargassum complex, with a
discussion of the Sargassum community, Contributions in Marine Science, 16, 1-32.
12. Wells, R.J.D., and J.R. Rooker, (2004), Spatial and temporal patterns of habitat use
by fishes associated with Sargassum mats in the northwestern Gulf of Mexico, Bulletin of
Marine Science, 74, 81-99.
13. Woodcock, A.H., (1993), Winds, subsurface Sargassum and Langmuir circulations,
J. Exp. Mar. Biol. Ecol., 170, 117-125.
14. Gower, J.F.R., R. Doerffer, and G.A. Borstad. (1999). Interpretation of the 685 nm
peak in water-leaving radiance spectra in terms of fluorescence, absorption and
scattering, and its observation by MERIS, Int. J. Remote Sens. 20, 1771-1786.

Nature Precedings : hdl:10101/npre.2008.1894.1 : Posted 15 May 2008

Figure Captions
Figure 1. Floating Sargassum in the Gulf Stream near 63W, 37N imaged as MCI on
October 22, 2007 by ESA's ocean sensor, MERIS at its full spatial resolution of 300 m,
available for some areas. The Sargassum is collected into patches and long lines by
surface convergence and shear. The inset shows top-of-atmosphere radiance spectra for
an area containing Sargassum (plotted in red) and for nearby clear water (plotted in
green) in the region indicated by the arrow. The difference spectrum (blue, values on
right axis) shows the "red edge" characteristic of land vegetation, with the apparent red
edge position shifted to a shorter wavelength.
Figure 2. MERIS counts of Sargassum for the years 2002 to 2007 (top to bottom) and
months June to August (left to right), in pixels measuring one degree in latitude and
longitude. The background dark blue colour corresponds to no detections. Increasing
amounts are indicated by the colour sequence green, yellow, red to white. One-degree
squares where MCI gives strong signals from coral reefs and other benthic vegetation are
masked to black. Land at 0.25 degree spatial resolution is shown in grey. High
concentrations of Sargassum are indicated in the northern Gulf of Mexico in all years
except 2002, with the Sargassum appearing in the Atlantic to the east of Cape Hatteras
along about 37N in August and September, especially in 2005, 2006 and 2007. Little
Sargassum is observed in the Gulf starting in September.
Figure 3. Plots of total amounts of Sargassum for the Gulf of Mexico (15 to 30N, 80 to
100W) and the western Atlantic (22 to 40N, 40 to 80W), for the period June 2002 to
April 2008. Values are based on satellite counts calibrated with a mean value derived
from ship observations. Highest amounts are in May, June, and July in the Gulf of
Mexico, with a maximum Sargassum year in 2005. Amounts in the Atlantic increase
after July and usually drop back to low values by March. Very little Sargassum was
detected in 2002.
Figure 4. Average latitude and longitude of Sargassum in the Atlantic (22 to 40N, 40 to
80W) for months in which amounts plotted in Figure 3 were greater than one million
tons. The average positions show a consistent pattern in which Sargassum is injected
into the Atlantic near 37N, and spreads initially eastwards and then south west. The
spatial patterns are shown schematically in Figure 5. The letter "P" on the left panel
shows the average latitude (27.2 N) deduced from Parr

's measurements in late January

1934 and early February 1935 along roughly 68 W.
Figure 5. Simplified outline diagram showing the average extent of Sargassum in March,
May, July, September, November and February, based MERIS count distributions by
month (as shown in Figure 2 for June to August) averaged over the years 2002 to 2007.
In each year, Sargassum is first observed in the Gulf of Mexico in about March, by July it
is still present in the Gulf, but has also appeared off Cape Hatteras. It then moves east
and then south west, as shown. In the period 2002 to 2007, MERIS sees relatively little
Sargassum in the Atlantic between March and June (Figure 3).

Nature Precedings : hdl:10101/npre.2008.1894.1 : Posted 15 May 2008

Figure 1. Floating Sargassum in the Gulf Stream near 63W, 37N imaged as MCI on
October 22, 2007 by ESA's ocean sensor, MERIS at its full spatial resolution of 300 m,
available for some areas. The Sargassum is collected into patches and long lines by
surface convergence and shear. The inset shows top-of-atmosphere radiance spectra for
an area containing Sargassum (plotted in red) and for nearby clear water (plotted in
green) in the region indicated by the arrow. The difference spectrum (blue, values on
right axis) shows the "red edge" characteristic of land vegetation, with the apparent red
edge position shifted to a shorter wavelength.

Nature Precedings : hdl:10101/npre.2008.1894.1 : Posted 15 May 2008

Figure 2. MERIS counts of Sargassum for the years 2002 to 2007 (top to bottom) and
months June to August (left to right), in pixels measuring one degree in latitude and
longitude. The background dark blue colour corresponds to no detections. Increasing
amounts are indicated by the colour sequence green, yellow, red to white. One-degree
squares where MCI gives strong signals from coral reefs and other benthic vegetation are
masked to black. Land at 0.25 degree spatial resolution is shown in grey. High
concentrations of Sargassum are indicated in the northern Gulf of Mexico in all years
except 2002, with the Sargassum appearing in the Atlantic to the east of Cape Hatteras
along about 37N in August and September, especially in 2005, 2006 and 2007. Little
Sargassum is observed in the Gulf starting in September.

Nature Precedings : hdl:10101/npre.2008.1894.1 : Posted 15 May 2008

2002

2003

2004

2005

2006

2007

2008

2009

Start of year

t
a
l

assu

(

illion t

)

Gulf of Mexico
Atlantic

Figure 3. Plots of total amounts of Sargassum for the Gulf of Mexico (15 to 30N, 80 to
100W) and the western Atlantic (22 to 40N, 40 to 80W), for the period June 2002 to
April 2008. Values are based on satellite counts calibrated with a mean value derived
from ship observations. Highest amounts are in May, June, and July in the Gulf of
Mexico, with a maximum Sargassum year in 2005. Amounts in the Atlantic increase
after July and usually drop back to low values by March. Very little Sargassum was
detected in 2002.

Nature Precedings : hdl:10101/npre.2008.1894.1 : Posted 15 May 2008

Jul

Aug

Sep

Oct

Nov

Dec

Jan

Feb

Mar

Apr

Month

a
t
i
t
u

e (

r
ees)

2003/4
2004/5
2005/6
2006/7
2007/8

Jul

Aug

Sep

Oct

Nov

Dec

Jan

Feb

Mar

Apr

Month

i
t
u

e w

est

(
d

r
ees)

2003/4
2004/5
2005/6
2006/7
2007/8

Figure 4. Average latitude and longitude of Sargassum in the Atlantic (22 to 40N, 40 to
80W) for months in which amounts plotted in Figure 3 were greater than one million
tons. The average positions show a consistent pattern in which Sargassum is injected
into the Atlantic near 37N, and spreads initially eastwards and then south west. The
spatial patterns are shown schematically in Figure 5. The letter "P" on the left panel
shows the average latitude (27.2 N) deduced from Parr's (1939) measurements in late
January 1934 and early February 1935 along roughly 68 W.

Nature Precedings : hdl:10101/npre.2008.1894.1 : Posted 15 May 2008