<font color=red>Photos show 65,000 at peak of S.F. rally - Aerial study casts doubt on estimates of 200,000
sfgate.com
Wyatt Buchanan, Chronicle Staff Writer
Friday, February 21, 2003
San Francisco -- A survey using sophisticated aerial photography
of Sunday's anti-war march and rally in San Francisco has
produced results that indicate a far smaller crowd than the
200,000 protesters estimated by police and event organizers.
The results of the independent survey, commissioned by The
Chronicle and SFGate.com, cast doubt on traditional counting
methods and contradict the crowd estimate of 200,000, which was
reported in this newspaper and news media around the world.
Crowd size in a demonstration is important because organizers
tend to use it as evidence of support for their cause.
In a series of detailed, high-resolution photographs, the aerial
survey shows that around 65,000 people were in the area of
Market Street and Civic Center Plaza at 1:45 p.m. Sunday, which
organizers said was when crowd size was at its peak. That number
does not take into account marchers who dropped out before or
arrived after the moment the photo sequence was shot.
Calculating a precise number of protesters for the entire rally is
not possible from this survey, but the result is much more accurate
than the visual scan method most commonly used by police and
organizers.
The Chronicle hired the Santa Clara air photography firm Air
Flight Service, a company with 20 years of experience in taking
photographs for topographical maps for government agencies
and private companies.
Using a fixed camera mounted in the floor of the plane, the crew
made images of the rally from 2,000 feet. The photographs --
taken directly above Market Street and Civic Center Plaza and
enlarged -- provide a perspective that allows a discrete count of
individuals and a view of the spaces between them, a view that is
impossible from ground-level.
Both Air Flight Service and The Chronicle examined the photo
survey and independently arrived at the estimate of 65,000
marchers at the time the photographs were taken, a figure
supported by public transportation statistics.
The flight service says its count is accurate within a range of plus
or minus 10 percent.
This verifiable technique, experts say, could replace the current
politically sensitive count totals, whereby organizers tend to
provide a high number of participants, and police generally
provide a lower number.
'MYTHICAL NUMBER'
When told of The Chronicle's survey, Alex S. Jones, the director
of Harvard University's Joan Shorenstein Center on the Press,
Politics and Public Policy, said, "The number of people (in a
crowd) is a mythical number, and now you're going to turn it into
a fact, and that won't be welcomed."
Jones, a Pulitzer Prize-winning former reporter for the New York
Times, added, "There's an old saying in journalism: People only
see what they believe. This is an emotional issue, not a factual
issue as far as most people are concerned."
Police and event organizers, when told of The Chronicle's
numbers, stood by their estimates of 200,000 marchers, though
both groups based their figures almost entirely on observational
methods and not on a verifiable methodology. (Organizers
originally estimated the crowd at 200,000 to 250,000 but now are
going with the 200,000 figure.)
They said the aerial survey results were flat wrong.
"Oh my word. Come on, that's ridiculous," said Bill Hackwell,
spokesman for International ANSWER, one of the groups that
organized Sunday's march and rally. The organizers have
another planned for March 15 in San Francisco, coinciding with
rallies in Washington, D.C., and elsewhere.
'ENORMOUSLY LOW'
Greg Suhr, the San Francisco deputy police chief who calculated
the police figure, said of The Chronicle's estimate, "I can tell you
for a fact that's an enormously low number. (Pacific Bell Park),
just in the stands, holds 40,000. The crowd at Pac Bell would
pale in comparison to the crowd on Sunday."
The photographs, however, indicate otherwise. The series of
images were shot in a 30-second sweep by the crew from Air
Flight Service over the march and rally areas at 1:45 p.m. Peak
attendance was between 1:30 and 2 p.m., Hackwell said.
At the time the photographs were taken, many people had
gathered in the plaza, and the head of the march was at
McAllister Street. The tail was near Sansome Street, and no side
street was full enough to count, said Jack Barcelona, who runs the
air photo firm.
Overlaying the photographs with a grid, surveyors from Air Flight
Service estimated crowd density in the plaza and along the route.
Each grid was evaluated and assigned a density of people, from
10 percent to 100 percent full. Most were judged at 25 percent or
50 percent full. This is the first time the firm has used its
equipment for crowd estimation.
In comparison, both police and rally organizer figures are based
on estimates of previous crowd sizes and on eye-level
approximations of the event Sunday.
To reach the police and protester count of 200,000, more than
twice as many people as were photographed by Air Flight Service
would have to have left the event before the photos were taken or
joined it afterward.
PLAZA CAPACITY ESTIMATE
Sunday's police estimates started with a calculation of Civic
Center Plaza capacity made by the San Francisco Recreation
and Park Department in the mid- 1980s, which found the plaza
holds a crowd of 43,000 people, said Deputy Chief Suhr.
The plaza stayed full throughout the day, even while marchers
still filled Market Street, Suhr said. Based on previous estimates
of crowd sizes on Market Street, he concluded the roadway and
sidewalks had about 100,000 people on them.
Other tributary marches and people in side streets pushed the
number to between 150,000 and 200,000, he said. The crowd
also had pushed the front of the march from First Street to Third
Street before it began.
No officers were assigned to count the crowd.
"It's pretty much me," Suhr said.
In the aerial photograph of Civic Center Plaza, in which the area
appears full, a count by Air Flight Service shows only about
20,000 people in the plaza at 1:45 p.m. However, it did become
more densely packed later as marchers continued to arrive.
Peace groups based their number on a comparison with a rally
Jan. 18 that they estimated at 150,000 to 200,000. Sunday's
protest was similar in size to the march last month, Hackwell said.
Police had initially counted the January gathering at 50,000
participants but revised that figure to 150,000 by saying the
original number was a count of Civic Center Plaza, not the whole
march area.
At Sunday's march, Hackwell said, a man stood at Hallidie Plaza
to videotape passing marchers on behalf of event organizers.
Results from that count were not final Thursday, he said.
A Chronicle reporter standing at Market and Montgomery streets
estimated that 50,400 marchers had passed by from 11:45 a.m. to
2:05 p.m. That number did not account for people who went
directly to Civic Center, bypassing Market Street.
Over the decades, counting large crowds of demonstrators has
been unreliable. During the Vietnam era protests in the 1970s,
newspaper editors often would split the difference between
protesters' high numbers and police agencies' low estimates and
print that figure as the crowd count, according to Todd Gitlin, a
professor at Columbia University's journalism school.
Recent media practice, however, has been to print both
estimates, Gitlin said.
Chronicle Executive Editor Phil Bronstein explained why the
paper undertook the project: "After hearing concerns from our
readers about our accuracy in reporting crowd size in
demonstrations, we were determined to come up with a better
method to calculate the number of people who turn out for such
events.
"This innovative approach, however controversial its conclusions,
is nonetheless a far more exact way to provide readers with
critically important information," he added.
COUNT OF BART RIDERS
The Chronicle figure was supported by BART and Golden Gate
Transit rider numbers.
On Sunday, BART officials estimated that 150,000 people exited
the BART system, a figure they later updated to 190,000. (A
rider's round-trip would involve two exits.) But on Sunday, BART
did not have specific breakdown for how many people
disembarked at the four main downtown San Francisco stations.
Asked later in the week for specific numbers, BART spokesman
Mike Healy said that during the entire day Sunday -- from about 8
a.m. to midnight -- 66, 254 people had exited the gates at the
Embarcadero, Montgomery Street, Powell Street and Civic
Center stations.
By comparison, 23,406 people had exited those same gates the
previous Sunday, Feb. 9, a difference of about 43,000 riders, he
said. Not all riders attended the march and not all marchers rode
BART.
Figures from Muni, whose trains were full heading into the city,
are not available, said Maggie Lynch, Muni spokeswoman.
Ferries from Sausalito and Larkspur transported 7,000 people to
San Francisco on Sunday. On an average Sunday, 1,500 people
make the trip, said Mary Currie, spokeswoman for Golden Gate
Transit.
Despite police and rally organizer doubts of the aerial photo
results, the methodology used in The Chronicle survey was dead
on, said one researcher who led a team of students in a recount
of the 1995 Million Man March in Washington, D.C., based on
aerial photography and television images.
"This methodology gives you the best results under the
circumstances," said Farouk El-Baz, director of the Center for
Remote Sensing at Boston University, referring to fixed-camera
aerial photos. The center specializes in photographing large
tracts of land from space satellites for various projects such as
looking for groundwater in the Sahara.
HISTORY OF MISCOUNTS
Miscounts of large crowds are common, El-Baz said, even such
disparity as calling a crowd of 50,000 a crowd of 200,000.
"It's unbelievable, but it happens consistently," he said. "If you
are in a demonstration yourself, you can easily be misled
because you see so many people."
Counting crowds based on physically counting each individual
seen in an aerial photo instead of relying on ballpark estimates,
"is going to change the whole dynamic" of crowd estimates,
El-Baz said.
In cities around the world, police estimates of crowd sizes from
Saturday's rallies were based on the average number of people in
a small area and then expanding that figure to cover the whole
route.
In London, Metropolitan Police in helicopters counted protesters
in the streets based on the density of an average 10-by-10-yard
square. That figure was then applied to the length of the march,
said Alastair Campbell, police spokesman, in a Chronicle
interview.
Police estimated 750,000 marchers in the streets and more than
1 million at Hyde Park.
In Rome, the Carabinieri, Italy's national police force, said
700,000 people had protested, basing that number on the
capacity of piazzas where the demonstration happened and
calculating four people per square meter during the march,
according to an e-mail from a police spokesman in response to a
reporter's questions.
French officials said a Paris rally had drawn 100,000 people,
while organizers put that number at 300,000, Emmanuel
Gagniarre, press officiate with the French Embassy in
Washington, D.C., said in an interview.
Details of how French officials made their estimate were not
available.
HOW THE CROWD WAS COUNTED
High-resolution graphics show how we came up with the number.
Chronicle staff members Tyra Lucile Mead, Carl Nolte and Laura
Thomas, and SFGate.com staff members James Irwin and
Rochelle Aquino contributed to this report. / E-mail Wyatt
Buchanan at wbuchanan@sfchronicle.com.
=================================================================================
Counting crowds
Using aerial photography to estimate the
size of Sunday's peace march in S.F.
Friday, February 21, 2003
From above, the world looks different from the way it appears at
street level, as aerial photos of last Sunday's S.F. anti-war march
show.
Click to View Full Size
THE ROUTE
The anti-war demonstration began with an 11 a.m. rally at Justin
Herman Plaza at the Embarcadero in San Francisco. At around
noon, the march began up Market Street toward the Civic Center
Plaza. The march route was approximately 1.75 miles. In the
photos on these pages, taken at 1:45 p.m. of the entire length of
the march, the crowd had moved far up Market Street. The
Embarcadero area is nearly empty. The march was densest near
the Powell Street cable car turn-around. The front of the main
march had reached McAllister Street, as another mass of
protesters was gathering in the Civic Center Plaza area. That day,
San Francisco police and march organizers estimated the total
crowd at 200, 000; organizers said the crowd was largest between
1:30 and 2 p.m. The Chronicle's estimate is 65,000 at the
moment the photos were made. This figure does not account for
marchers who left the group before that time or who joined the
rally later.
THE METHODOLOGY
The aerial photographs of Sunday's anti-war march were taken at
a time believed by organizers to be close to the peak of the
march. The plane made a pass over the length of Market Street
and the Civic Center area, taking contiguous photographs.
The lens in the camera looked directly down from 2,000 feet.
This process was repeated once to ensure that there would be a
complete set of usable photos. An additional pass was made to
allow a Chronicle staff photographer to shoot out the window of
the plane.
Air Flight Service, a company that specializes in high-resolution
aerial photography, produced 9-by-9-inch black and white
negatives, which were enlarged into prints with enough clarity to
define individuals on the street.
Air Flight Service, under contract to The Chronicle, then applied
a grid pattern to the prints to provide units in which the marchers
could be counted. The entire length of the 1.75-mile route is
visible in the images.
Because the density of the crowd varied over the route, the grids
were sorted into five categories - 100 percent, 75 percent, 50
percent, 25 percent and 10 percent filled with people. This was
done by visual estimates. Each of the nearly 300 grids was given
a colored dot representing these percentages.
At this point, the hand counting began. A representative sample
of grids for each density was selected. For instance, eight 50
percent grids were counted, eight 25 percent grids were counted
and so forth. The totals for each group were examined to ensure
there were not extreme deviations. This tested the original visual
estimate that had assigned grids to density groups. If necessary, a
sample grid's category was changed to reflect the count.
An average number of people for each density grid was
established using the sample groups. Air Flight Service then
tallied the grids to come up with a total of nearly 65,000 people
visible from the air at the moment the photographs were shot.
The company said the margin of error is 10 percent.
Chronicle staff members double-checked the calculations
counting multiple grids, confirming that the density assignments
were correct. The Chronicle then counted individuals in a
sampling of the grids. The newspaper's count in some cases was
higher, in some cases lower than Air Flight Service's count. But
overall the newspaper's total confirmed the Air Flight Service
count.
sfgate.com |
