Migration on Wings

1 Migration on Wings
Animal migration is a fascinating marvel. Millions of years of evolution have con-
ditioned some animals to migrate seasonally for survival of their species. Of all such
migrations, the long distance migration of some birds is truly an amazing phenome-
non, with some birds traveling extremely long distances nonstop. Bar-tailed godwits
(Fig. 1) routinely travel from Alaska and Siberia to Australia and New Zealand and
backeveryyear, a total roundtripdistance of over 22,000 km(13,750 miles), half way
around the world. One female godwit, called E7 for a tag onits upper leg, was captured
in February 2007 and a tiny radio transmitter was implanted in its abdomen. It was
subsequently continuously satellite-tracked by scientists from the U.S. Geological
Survey. The godwit left NewZealand on March 17, 2007 and ewnonstop 10,080 km
(6,300 miles) to reach China on March 25, 2007. The journey took 8 days and the
average ight speed was 14.6 m/s (32.8 mph). The bird stayed in China for 5 weeks,
replenishing its fat reserves before it took off on May 1, 2007 across the Sea of Japan
and North Pacic, and took 6 days to reach the Yukon-Kusokwim delta in Alaska on
May 6, 2007, ying a distance of 7,200 km (4,500 miles) at an average speed of
13.9 m/s (31.3 mph). It summered in Alaska, fattening up on marine worms and clams
that it plucked from the mud with its long beak, and breeding. It took off on its fall
migration on August 29, 2007, and after a nonstop ight of 11,520 km (7200 miles)
lasting8 days, landedinNorthCape, NewZealandonSeptember 7, 2007. The average
ight speed was 16.7 m/s (37.5 mph), augmented by strong tail winds up to 11 m/s
(25 mph) in the North Pacic. During its ight from Alaska, the bird rst headed for
Hawaii, then turned southwest toward Fiji before heading south to New Zealand,
ying at altitudes as high as 15,000 ft. Scientists estimate the bird to have lost 50% of
its body weight, and conjecture that it may have shut off half its brain to nap along the
way, as mallard ducks do (Rozell 2007). Its entire ight path is shown in Fig. 2.
Such phenomenal journeys are not conned to this particular bird. Other birds of
the species also travel remarkable distances in their annual migration. The USGS
team also tracked eight other godwits on what scientists called extreme endurance
L. Kantha, Migration on Wings, SpringerBriefs in Applied Sciences and Technology,
DOI: 10.1007/978-3-642-27925-6_1, The Author(s) 2012
1
ights of between 6,968 km(4,355 miles) and 11,613 km(7,258 miles) depending
on the route. Figure 3 shows the ight paths of eight bar-tailed godwits during their
fall migration from Siberia to Australia in the October of 2008.
Nor are such migrations conned to godwits. Sooty shearwaters are seabirds
slightly over a kilogram in mass (Fig. 4). They are spectacular long-distance
migrants, traveling north up the western side of the Pacic and Atlantic Oceans at the
end of the nesting season in MarchMay, reaching sub Arctic waters in June-July,
where they cross from west to east, then returning south down the eastern side of the
oceans in SeptemberOctober, reaching the breeding colonies in November. They do
not migrate as a ock, but rather as single individuals, Fig. 5 fromProceedings of the
National Academy of Sciences shows the tracks of 19 sooty shearwaters tagged in
early 2005 and tracked for an average of 262 days during their breeding period (light
blue lines) and subsequent migration. Yellowlines showthe shearwaters northward
migration from their breeding sites; orange tracks show the birds activity in three
northern Pacic foraging zones and their return trip southward.
In the Atlantic Ocean, they cover distances in excess of 14,000 km (8,750 miles)
fromtheir breeding colony on the Falkland Islands (52S60W) north to 6070Nin
the North Atlantic Ocean, off north Norway. Distances covered in the Pacic are
similar or larger. Although the Pacic Ocean colonies at 35 to 50Soff NewZealand
are not quite so far south, in moving north to the Aleutian Islands, the longitudinal
width of the ocean makes longer migrations necessary. Recent tagging experiments
have shown that birds breeding in New Zealand may travel 74,000 km in a year,
reaching Japan, Alaska and California, averaging more than 500 km per day.
The Arctic Tern is a small seabird with an average of 110 g in mass and 0.8 m
wing span (Fig. 6). It has a circumpolar distribution, breeding colonially in Arctic
and sub-Arctic regions of Europe, Asia, and North America. The species migrates
from its northern breeding grounds to the oceans around Antarctica and back,
about 38,400 km (24,000 miles) each year. This is one of the longest regular
migrations by any bird. Note however, the journey from one polar region to the
other involves a rest and recharge stop in-between, as seen in Fig. 7. The Arctic
Tern ies as well as glides through the air, performing almost all of its tasks in the
Fig. 1 Bar-tailed godwit
(Wikipedia), a wading bird
with a mass less than 1/2 kg.
It ies annually from its
boreal summer habitat in
Alaska to its winter habitat in
New Zealand and back, a
distance of 11,000 km, each
way
2 Migration on Wings
air. It lands once every one to three years (depending on their mating cycle) to
nest. Once it has nished nesting, it takes to the sky for another long southern
migration. This 19,000 km (about 12,000 miles) journey each way ensures that
this bird sees two summers per year and more daylight than any other creature on
the planet. The average Arctic Tern in its life will travel a distance equal to going
to the moon and backabout 800,000 km (500,000 miles). One example of this
birds remarkable long-distance ying abilities involves an Arctic Tern ringed as
an unedged chick on the Farne Islands, Northumberland, UK, in summer 1982,
Fig. 3 The ight paths of bar-tailed godwits on their winter migration from Siberian Arctic to
Australia in the October of 2008 (left) and from New Zealand to Yellow Sea in March 2007,
which was 11,026 km (nonstop) and took 9 days
Fig. 2 The epic ight path of a female bar-tailed godwit as it ew across the Pacic in 2007 on
its seasonal migration. The bird was tracked by a satellite transmitter implanted in its stomach
1 Migration on Wings 3
Fig. 4 Sooty shearwater
(Wikipedia)
which reached Melbourne, Australia, in October 1982, a sea journey of over
22,000 km (14,000 miles) in just 3 months from edging. Another example is that
of a chick ringed in Labrador, Canada, on 23 July 1928. It was found in South
Africa four months later (Wikepedia). A more recent study by the Arctic Tern
Migration Project found tagged Arctic Terns travelling an average of 71,000 km
(44,000 miles) in their annual migration between Greenland and Antarctica
(Fig. 7), which is 4000 miles more than the Sooty Shearwater and amounts to
2,400,000 km over their 30 year lifetime!
There are numerous other birds that undertake seasonal migration. A tabulation
of many of these can be found in Appendices B and C. Because bird migration has
been a fascinating subject of interest to both scientists and the public, enormous
attention has been lavished on the subject with innumerable scientic articles and
popular articles. National Geographic has published articles, along with compre-
hensive maps of bird migration in the western and eastern hemispheres, which are
reproduced in Figs. 8 and 9.
Jacque Perrins lm on Winged Migration is a real treat for the eyes and is a
must see for bird migration enthusiasts. Its companion book (Perrin 2003) contains
interesting details. A fairly recent book edited by Jonathan Elphick, Atlas of Bird
Migration (Elphick 2009) is an excellent treatment of the subject for the non-scientist
types. Hank Tennekes popular book on The Simple Science of Flight (Tennekes
2009) has just been updated and is still an easily-read description of bird ight.
In the scientic arena, both ornithologists and aerodynamicists have investi-
gated the phenomenon of bird migration. Rapid advances in technology have
enabled radio transmitters to be sufciently miniaturized and implanted in birds so
that they can be tracked continuously by satellites. Examples of this were shown
earlier. Thomas Alerstam is an authority on bird migration and his authoritative
book Bird Migration (Alerstam 1990), while a bit dated is still useful to sci-
entists and researchers. Colin Pennycuick is also an authority on the subject and
has recently published Modelling the Flying Bird (Pennycuick 2008), which
contains a computer code to model bird migration. The technical literature on bird
migration is too voluminous to cite here, but a fairly recent paper by Alerstam and
Hedenstrom (1998) and the references therein are a good start. See also Gree-
newalt (1975), Pennycuick (1969 & 1989) and Shyy et al. (1999).
4 Migration on Wings
Fig. 6 Arctic Tern (Wikipedia)
Fig. 5 Flight paths of 19 shearwaters satellite tracked during 2005. Light blue lines correspond
to the breeding period. Yellow lines show the shearwaters northward migration from their
breeding sites; orange tracks show the birds activity in three northern Pacic foraging zones and
their return trip southward
1 Migration on Wings 5
Fig. 7 Arctic Tern Migration Map (courtesy of Greenland Institute of Natural Resources)
6 Migration on Wings
Fig. 8 National Geographic wall map of migration in the eastern hemisphere (reproduced with
permission)
1 Migration on Wings 7
Fig. 8 (continued)
8 Migration on Wings
Fig. 9 National Geographic wall map of bird migration in the Americas (reproduced with
permission)
1 Migration on Wings 9
Fig. 9 (continued)
10 Migration on Wings
2 Aerodynamics and Energetics of Flight
Ornithologists are interested in physiological and ecological aspects of bird
migration, while aerodynamicists have focused on aerodynamics and energetics
aspects such as: What are optimum speeds for migration? What sort of fat reserves
are needed for carrying out the migration successfully? We will now look at some
aspects of winged migration.
First, power requirements for migratory ight. The power required P is given by
P = D V, where D is the drag and V is the ight speed. Using the fact that weight
W equals the Lift L for level ight, the power per unit body mass can be written as:
P
m

Vg
0
L=D
1
where L/D is the lift/drag ratio, also known as aerodynamic efciency. The higher
the aerodynamic efciency, the smaller the power per unit body mass that must be
developed by the ight muscles. The higher the ight speed, the higher the power
requirement. The L/D ratio depends on the efciency of lift generation by the
wings and the overall aerodynamic drag of the bird. In general, long, narrow wings
provide a higher L/D. Such wings have a high aspect ratio, which is dened as
A = b
2
/S, where b is the span and S is the wing area. The drag on the bird consists
of two components, the form and skin friction drag, which increases quadratically
with ight speed, and the induced drag, which decreases quadratically with ight
speed. Induced drag can be thought of as the penalty for lift generation by nite
wings, since the induced drag is zero for innitely long wings b !1 . The
higher the aspect ratio, the smaller the induced drag. The skin friction drag
depends only on the ight Reynolds number, since ight Mach numbers and hence
compressibility effects are negligibly small, and the ow very nearly incom-
pressible. The smoother the surface, the lower the skin friction drag. Form drag
depends on the shape of the object. The form drag is reduced signicantly by
minimizing ow separation by streamlining and retracting appendages that might
stick out in the ow and cause ow separation. Birds have highly streamlined
bodies and wings, and tuck in their feet during ight. Feathers covering their
bodies appear to play an important role in reducing the overall drag. Consequently,
it is the aspect ratio of their wings that is important in determining their L/D ratio.
Song birds have fairly short, wide wings and hence small aspect ratios of around 4,
while albatrosses with long, narrow wings can have aspect ratios as high as 25.
Ocean birds have aspect ratios typically around 1214.
The optimum ight speed of a ying object depends on whether the expended
power needs to be minimized or the ight distance needs to be maximized. To see
this, consider the force balance during level ight.
2 Aerodynamics and Energetics of Flight 11
L
q
2
V
2
SC
L
W
D
q
2
V
2
SC
D
; P DV
2
where C
L
and C
D
are the lift and drag coefcients, which are functions of the wing
angle a and the ight Reynolds number Re
qVb
l
; where q is the air density
(which is a function of altitude) and l is the viscosity. For normal ight speeds, Re
is high enough that its inuence on skin friction drag does not change much with
speed. For an airfoil, the lift coefcient C
L
is a function of mainly the angle of
attack: C
L
0:105 a; where a is in degrees. Bird wings are shaped like airfoils
and therefore obey this rule. The drag coefcient C
D
can be written as:
C
D
D
D0
C
Di
C
D0

C
2
L
pAe
3
where subscript 0 denotes the form drag and subscript i, the induced drag. C
Di

C
2
L
pAe
is the induced drag coefcient, where e is the platform efciency (also called
Oswald efciency), which depends on the lift distribution on the wing. For the best
designed man-made wings, e is typically around 0.85. Since birds ap their wings,
which do not generate much lift during the upstroke, the platform efciency is
smaller. A value of 0.75 is reasonable for birds. The drag on the bird and the power
required can be written as:
D
q
2
V
2
S C
D0

C
2
L
pAe
_ _

qS
2
_ _
C
D0
V
2

W
2
qS
2
_ _
pAeV
2
P
q
2
V
3
S C
D0

C
2
L
pAe
_ _

qS
2
_ _
C
D0
V
3

W
2
qS
2
_ _
pAeV
4
By differentiating with respect to V, it can be shown that minimum drag and
minimum power occur for ight speeds given by
V
2
1
V
2
min D

W
qS
2
_ _
C
D0
pAe
p and V
2
2
V
2
min P

W
qS
2
_ _
3C
D0
pAe
p 5
respectively. The minimum drag condition also corresponds to minimum energy
per unit distance (since E = DR, where E is the energy and R is the distance), and
hence maximum range: V
max R
V
min D
. So birds must travel at V
max R
for making
the best use of their fat reserves. On the other hand, they must travel at a slower
ight speed V
minP
for minimum exertion. Note that V
max R
1:316 V
min P
:
By differentiating with respect to C
L
, it can be shown that minimum drag and
minimum power occur for
12 Migration on Wings
C
D0

C
2
L
pAe
C
Di
and C
D0

C
2
L
3pAe

1
3
C
Di
6
respectively or equivalently
C
D0
C
Di
1;
C
Di
C
D

1
2
; C
D
2C
D0
and
C
D0
C
Di

1
3
;
C
Di
C
D

3
4
; C
D
4C
D0
7
respectively. The condition for minimum drag D corresponds to maximum pos-
sible L/D (since L = W = constant) or maximum aerodynamic efciency of the
ight system, leading naturally to the maximum possible range for the bird. Also
L
D

min D

W
qS
2
_ _
V
2
min D
2C
D0

and
L
D

min P

W
qS
2
_ _
V
2
min P
4C
D0

8
so that (L/D)
minP
= 0.866 (L/D)
minD
.
However, birds have been observed to travel often at ight speeds higher
than those corresponding to minimum drag (maximum range). To explain this,
it is important to realize that the sustained continuous power output from the
ight muscles may permit the bird to travel at a higher speed without tiring it
unduly. If we write C
Di
bC
D0
; b is a decreasing function of ight speed
V and the expressions for the ight speed V, drag D, L/D ratio and power
P can be written as:
V b
1=4
W
1=2
qS
2
_ _
1=2
pAe
1=4
_
_
_
_
C
1=4
D0
9
D 1 b b
1=2
W
pAe
1=2
_ _
C
1=2
D0
;
L
D

b
1=2
1 b
pAe
1=2
C
1=2
D0
10
P 1 b b
3=4
W
3=2
qS
2
_ _
1=2
pAe
3=4
_
_
_
_
C
1=4
D0
and
P
m
1 b b
3=4
W
1=2
g
0
qS
2
_ _
1=2
pAe
3=4
_
_
_
_
C
1=4
D0
11
Since b 3 for minimum power and 1 for maximum range, Using the above
expressions, it is easy to show that
P
max R
1:14P
min P
; V
max R
1:316V
min P
;
L
D

min P
0:866
L
D

max R
12
More importantly, these expressions allow us to determine the power P and L/D
for any given velocity Vin terms of the values at either minimumpower or maximum
range. For example,
2 Aerodynamics and Energetics of Flight 13
L
D

2b
1=2
1 b
_ _
L
D

max R
where b V
max R
=V
4
and P 0:57
1 b
b
3=4
_ _
P
min
where
b V
min P
=V
4
13
These expressions can also be used to determine the ight speed for available
muscle power. The ight muscles of birds can generate about 100 W/kg of con-
tinuous power (twice as much in short bursts) and the mass of the ight muscles is
roughly 2025% of the total body mass of the bird (larger birds have smaller
percentage; some birds grow ight muscles before migration). Therefore, the long
term sustained power output available for ight from ight muscles is therefore
2025 W per kg of total body mass. Using
P
m
_ _
avilable

1 b b
3=4
2
P
m
_ _
max R
14
it is possible to determine the corresponding value of b and hence the ight speed
possible: V
max P
V
max R
b
1=4
:
Figure 10 shows the ight speed of various birds that correspond to the sustained
muscle power output available for ight plotted against the minimum power (red
points), maximum range (blue points) and observed speeds (black points). Both the
red and blue points fall above 45

straight line, whereas the black points cluster

around it. This suggests that migrating birds tend to travel at the sustained power that
their ight muscles can produce without undue fatigue, rather than the ight speed
that produces maximumrange, a condition they are most likely unable to understand
and implement, or that corresponding to the minimum effort (power).
For ight at higher altitudes, one may have to account for the lower air density.
If the bird does not increase its respiratory rate at higher altitudes, the specic
power must be multiplied by the ratio of air density at the ight altitude to that at
sea level, which is r
q
q
SL
, since the power output would decrease due to less
oxygen (per unit volume) available to burn fat (and other tissues). However, birds
could compensate for the lower air density by increased respiration rate and so we
will ignore the density change for simplicity.
Figure 11 shows the weight of the bird plotted against the muscle power per
unit body mass required for ight at ight speeds corresponding to maximum
range as well as the observed ight speeds.
The ight speed of migrating birds (in still air) is of considerable interest.
Tennekes (2009) derives a relationship for ight speed as follows: Using Eq. (2),
V

2
W
S
_ _
qC
L

. However, C
L
6a; where a is in radians or equivalently, C
L

0:105a where a is in degrees. Therefore,
14 Migration on Wings
V

2
W
S
_ _
0:105 a q
SL
q=q
SL

15
The best L/D ratio occurs at around 56
o
and the value of 6
o
is used in Fig. 13,
where the bird weight is plotted against the ight speed V. The ight speed
depends only on the wing loading W/S. This equation however ignores the
inuence of L/D and aspect ratio A on ight speed. These parameters can be taken
into account as follows:
W qC
L
S !
W
S
qC
L
!
W
S
_ _
1
C
D
_ _
q
C
L
C
D
_ _
!
W
S
_ _
1
C
Di
1
C
D0
C
Di
_ _
q
L
D
_ _
Fig. 10 Fatigue-free muscle power limited ight speed plotted against the ight speeds
corresponding to minimum power (red triangles), maximum range (blue circles) and observed
speeds (black squares). The data are from Tennekes (2009) listed in Appendix A
2 Aerodynamics and Energetics of Flight 15
Using D
i
qC
Di
S
1
peq
W
b
_ _
2
;
C
Di

1
peq
2
S
W
b
_ _
2
!
W
S
_ _
peq
2
S
W
b
_ _
2
1
C
D0
C
Di
_ _
q
L
D
_ _
Using AR
b
2
S
!
peq
2
AR
W
S
_ _
1
C
D0
C
Di
_ _
q
L
D
_ _
!q
1
2
qV
2

W
S
_ _
L
D
_ _
1
C
D0
C
Di
_ _
peAR
Therefore; V

2
W
S
_ _
L
D
_ _
1
C
D0
C
Di
_ _
peARq
SL
q
q
SL
_ _

_
16
C
D0
C
Di
1 for minimum drag (maximum range),
1
3
for minimum power. Figure 12
shows also these ight speeds plotted against weight for iers listed in Appendix
A, for whom L/D data are available.
Fig. 11 Power requirement for ight: weight of various birds plotted against the muscle power
required for ight per unit body mass at speeds corresponding to maximum range (blue circles) as
well as observed (black squares) ight speeds. The approximate maximum sustained muscle
power per unit mass available for ight is shown by the vertical red lines
16 Migration on Wings
Note that D
i
qC
Di
S
1
peq
W
b
_ _
2
and
D
i
D
0

1
peq
2
C
D0
AR
W
S
_ _
2
17
This shows that the induced drag depends only on the span loading whereas the
ratio of the induced drag to the form/skin friction drag depends on only the wing
loading.
3 Migration Range
The Breguet equation for the range R of migratory birds can be derived as
follows:
dR
dt
V
P
D

1
g
0
P
m
_ _
L
D
_ _

1
g
0
_ mh
c
g
m
_ _
L
D
_ _

h
c
g
g
0
L
D
_ _
1
m
dm
dt
_ _
18
so that dR
h
c
g
g
0
L
D
_ _
dm
m
RF
dm
dt
_ _
19
where
Fig. 12 Weight versus ight speed given by Eq. (15) (black squares) and Eq. (16) (blue circles
for maximum range and red triangles for minimum power) for iers listed in Appendix A, for
whom L/D data are available
2 Aerodynamics and Energetics of Flight 17
RF
h
c
g
g
0
L
D
_ _
20
is the range factor; h
c
is the energy available per unit mass of fat (39.3 MJ/kg) and
g the efciency of conversion of this energy to useful mechanical energy by the
ight muscles (*23% according to Pennyquick and Battly 2003). Integrating,
assuming L/D is constant, we get the Breguet range equation of bird migration:
R RF ln
m
i
m
f
_ _
RF) ln
1
1
m
fat
m
i
_
_
_
_
_
_ RF) ln 1
m
fat
m
i
_ _
21
Alternatively, following Tennekes (2009), an approximate equation for range can
be derived as follows:
Work done D R Energy produced m
i
m
f
h
c
g:
Therefore, R
m
i
m
f
h
c
g
D
L
W
_ _

m
i
m
f

mg
0
L
D
_ _
h
c
g
22
Using m m
av
; R
2h
c
g
g
0
L
D
_ _
m
i
m
f

m
i
m
f

2RF)
m
i
m
f
1
_ _
m
i
m
f
1
_ _
:
Therefore; R RF
m
fat
m
i
_ _
1 0:5
m
fat
m
i
_ _ _ _
RF
m
fat
m
f
1
m
fat
2m
f
_
_
_
_
_
_ 23
This equation underestimates R slightly for large values of m
fat
/m
f
. Note that the
usual limit on m
fat
/m
f
is around 1.0, although after exhausting the fat reserves
under adverse ight conditions, the bird may have to use up muscle mass for
power generation, increasing the extreme limit to around 1.2.
Both approaches assume L/D = constant during the migration ight. This is not
correct for birds that undergo signicant mass changes during their migration, i.e.,
for long distance nonstop migrants such as the bar-tailed godwit. A range equation
for this situation can be readily derived by recognizing RF = RF
i

m
i
m
m
i
m
f
_ _
RF
f
RF
i
. Then Eq. (19) becomes
dR RF
i

m
i
m
i
m
f
_ _
RF
f
RF
i

_ _
dM
m
_ _

RF
f
RF
i
m
i
m
f
_ _
dm
Integrating this results in:
18 Migration on Wings
R RF
i

m
i
m
i
m
f
_ _
RF
f
RF
i

_ _
ln
m
i
m
f
_ _
RF
f
RF
i

RF
i

m
i
m
fat
_ _
RF
f
RF
i

_ _
ln
1
1
m
fat
m
i
_
_
_
_
_
_ RF
f
RF
i

RF
i
1
m
f
m
fat
_ _
RF
f
RF
i

_ _
ln 1
m
fat
m
f
_ _
RF
f
RF
i

24
Take the case of a bar-tailed godwit mentioned in the introduction on its
nonstop migration from Alaska to New Zealand, for which the observed value of
m
fat
/m
i
= 1 and therefore R = 0.386 RF
f
? 0.307 RF
i
. Let us suppose that ini-
tially it is so fat that RF
i
= 0.8 RF
f
. Then R = 0.617 RF
f
instead of R = 0.693
RF
f
if it had the same L/D ratio throughout the ight. This is a 11% reduction.
Using Eq. (23) instead and an average value for RF = 0.5(RF
i
? RF
f
), we would
have gotten R = 0.667 RF
f
for constant L/D and R = 0.6 RF
f
for varying L/D. The
error in using the approximate formula is about 3.8 and 2.8%, respectively, which
is quite small in view of the uncertainties involved in estimating the range factor.
Figure 13shows the fuel reserves required for a given migrationdistance for various
values of L/D, assumingmigrationoccurs at altitudes of less than500 mandconversion
efciency gis 23%. Clearly, the upper limit on nonstop migration distance is around
2,000 km for song birds, whereas ocean birds can travel around 10,000 kmnonstop.
Fig. 13 The fuel reserves needed for a given still air range as a function of L/D. The lower red
line corresponds to using up just the fat reserves, whereas the upper one to using up some of the
ight muscles as well
3 Migration Range 19
Irrespective of which of the above three equations are used for computing the
range of the migrating bird, it is clear that the range depends on four factors: 1. The
aerodynamic efciency L/D, 2. The efciency of conversion of fat to mechanical
energy g, 3. The ight altitude, and 4. The ratio of the available fuel reserve mass to
the nal body mass m
fat
/m
f
. Taking the bar-tailed godwit as an example, the longest
distance this bird travels nonstop is 11,500 km during the southward migration to
New Zealand and 7,500 km during its northward migration to China. Godwits
migrating fromSiberia to northern Australia appear to make a refueling stop in either
Korea or China so that the maximum distance they travel nonstop is from Korea to
northern Australia around 6,700 km. Since the journey fromAlaska to NewZealand
appears to get some assist fromstrong tail winds until roughly 30
o
latitude, the still air
range must be greater than 7,500 km and less than 11,500 km. Taking the average,
the still air range must be around 9,500 km. Accounting for some travel at higher
altitudes (this particular bird appears to have reached 15,000 ft at times), we can
round this off to 10,000 km. This requires an average L/D ratio of around 15, a value
this bird appears to achieve, according to the satellite tracking.
The breeding habitat in Yukon delta is at roughly 60
o
N, 162.5
o
W, whereas the
winter habitat in New Zealand is at roughly 35
o
S, 173
o
E. Siberian summer habitat
is around 7075
o
N, 140160
o
E, whereas the Australian wintering habitat is at
15
o
S, 125
o
E (Yalu river delta in Korea is around 40
o
N, 124
o
E and Yangtze delta in
China is 30.5
o
N, 121
o
E). Incidentally, the great circle distance between two points
whose longitudes and latitudes are k
1
; k
2
and /
1
; /
2
(in radians) is
D
c
R
E
cos
1
cos /
1
cos /
2
cos k
1
k
2
sin /
1
sin /
2
25
in kilometers, where R
E
= 6,371 km is the radius of Earth.
Pennycuick and Battley (2003) did a thorough study of great knots migrating from
northwest Australia (18
o
S, 122.4
o
E) to Chongming Island at the mouth of Yangtse
river in China (30.8
o
N, 121.5
o
E) ying 5,420 km nonstop in about 4 days in early
April 1998. They collected physiological data on 10 birds before and after migration.
The wing span b = 0.586 m, the wing area S = 0.0397 m
2
and therefore the aspect
ratio A = 8.65 on the average. The initial mass of the birds was 233.4 g, out of which
89.8 g were fat. The nal mass was 125.0 g, but not all fat was consumed during the
ight. About 10.7 g of fat remained. This means that the birds consumed 30.3 g of
muscle and other tissue! Pennycuick and Battley (2003) used a complex code to
compute the likely range of the birds and showed that the birds had the capability to
y 7,040 km, if they used up all their fat reserves, discounting the notion that tail
winds are essential to such long distance migration. These results are of course
dependent on various assumptions made in the simulation, the most critical of which
is the body drag coefcient, for which they assumed a value of 0.1 instead of values
twice as high indicated by wind tunnel experiments on frozen birds. They argued that
a lower value is more realistic for live birds.
The uncertainty associated with the body drag in the above case study is
essentially equivalent to the uncertainty in the L/D ratio of the bird. Consequently,
the use of a simpler method based on Eq. (21), (23) or (24) is well justied.
20 Migration on Wings
However, it is necessary to account for burning of the body tissue, since the value
of h
c
for protein is 18.3 MJ/kg less than half of the 39 MJ/kg value for fat.
Equation (18) must be modied to:
dR
dt

1
g
0
h
fat
c
_ m
fat
h
prot
c
_ m
prot
_ _
g
m
_ _
L
D
_ _
which is difcult to integrate analytically unless it is assumed that the tissues are
consumed after the fat, which may not be strictly speaking correct (however, the
error is probably within other uncertainties in this type of calculation, e.g., L/D). If
m
fat
is the mass of fat and m
tis
is the mass of body muscle and other tissues
consumed during the journey,
R RF
f
ln
m
i
m
i
m
fat
_ _
RF
t
ln
m
i
m
fat
m
i
m
fat
m
tis
_ _
26
Equivalently,
R RF
f
ln
1
m
fat
m
f

m
tis
m
f
1
m
tis
m
f
_
_
_
_
_
_RF
t
ln 1
m
tis
m
f
_ _
27
where RF
f

h
fat
c
g
g
0
L
D
_ _
; RF
t

h
tis
c
g
g
0
L
D
_ _
28
This equation is useful in computing the extra range attainable by burning body tissue
after the bird has exhausted the fat reserve. Applying this equation to great knots,
assuming an average value of 10 for L/D, the range is 4,440 ? 1,020 = 5,460 km,
close to the observed value.
Is L/D = 10 the correct average value for great knots? The above calculation
assumed sequential burning of fat and tissue. Analytical Breguet range equation
cannot be derived for simultaneous burning. However Eq. (23) may be modied to
account for both cases:
R
2
g
0
L
D
_ _
m
fat
h
fat
c
m
tis
h
tis
c
_ _
g
m
i
m
f

so that
R
RF
f
m
fat
m
i
_ _
RF
t
m
tis
m
i
_ _
1 0:5
m
fat
m
i
_ _
0:5
m
tis
m
i
_ _ _ _

RF
f
m
fat
m
f
_ _
RF
t
m
tis
m
f
_ _
1
m
fat
2m
f

m
tis
2m
f
_ _
29
3 Migration Range 21
Using Eq. (29) results in required average L/Dof 11.4 for the observed 5,420 kmrange.
Alternatively, if the weighted value of RF: RF
W

RF
f
m
fat
RF
t
m
tis
m
fat
m
tis
_ _
is used in
Eq. (21)
R RF
W
ln
m
i
m
f
_ _
RF
W
ln
1
1
m
fat
m
i

m
tis
m
i
_ _
RF
W
ln 1
m
fat
m
f

m
fat
m
f
_ _
30
which leads to a required L/Dof 11.1; Either value is quite plausible for a great knot.
Some What if questions can be posed. How much farther could the great
knot travel, if it used up the remaining fat reserve? This is simply
RF
f
ln
m
f
m
f
m
remaining
fat
_ _
911 kmfor a total of 6,331 km. What if all the stored fat
was used up rst and then the tissue? Using Eq. (27), R = 6,371 roughly similar.
Pennycuick and Battly (2003) also present data for the bar-tailed godwit. The
birds starting from Alaska (wing span b = 0.748 m, the wing area S = 0.0568 m
2
and aspect ratio A = 9.85 on the average) had an initial mass of 367 g, out of which
201 g were fat. Assuming that all the fat is consumed during the winter ight to
New Zealand, Eq. (21) gives a range of 10,880 km for an average L/D ratio of 15,
quite adequate for the migration. If additional range were needed under adverse ight
conditions, burning some 20 g of tissue would extend the range by about 830 km.
Thus, it appears that some tail winds may not be needed for the godwit to undertake
this migration. Of course, birds do have the extraordinary ability to select favorable
wind and weather conditions for departure as well as ight altitudes with favorable
tail winds (Liechti 2006), but they are in no way capable of predicting what wind
conditions they are likely to encounter over a ight lasting many days along a route
thousands of kilometers long. Consequently, if persistently adverse ight conditions
are encountered en route, successful completion of migration may not be possible!
As Pennycuick and Battly (2003) point out, it is the fat fraction, the ratio of the fat
mass to the mass of the bird that is of critical importance to the range. Godwits starting
on their migration from Alaska to New Zealand use their remarkable ability to make
sure this ratio is the highest possible, by actually shrinking the tissues such as their
digestive systemnot needed for the ight. Pennycuick and Battly (2003) found that the
northward bound birds did not do so and while their fat reserves were roughly the same
(191 g fat, 446 g total mass), the ratio was 0.43 compared to 0.55 for the southward
bound birds. This is probably because the northward bound birds tend to make a
refueling stop in northern Australia or China, the latter at a distance of only 7,200 km,
which is well within the 7,670 km range achievable with the smaller fat fraction.
It is clear that for a given fat to body mass ratio, the major parameter that
governs the migration range of birds is their L/D ratio, which while being a
function of the wing aspect ratio for the most part, is also a function of the
aerodynamic drag characteristics of the bird itself. Unfortunately, this is a
parameter that is not available for most birds. Generally speaking, larger wing
spans correspond to higher L/D ratios. Figure 14 shows L/D of birds plotted
22 Migration on Wings
against their aspect ratio, wing span, wing loading and span loading, from data in
Tennekes (2009) listed in Appendix A. Also shown are data points (red triangles)
inferred from Figs. 1 and 9 of Templin (2000), who suggests that the wing span is
the principal determinant of the migration range of birds. Clearly, of all the four
quantities, L/D is best correlated with aspect ratio. However, the correlation is
imperfect, with signicant scatter and hence this result must be used with caution.
L/Dratio is beter correlated with the wetted aspect ratio (A
wet
= b
2
/S
wet
). Decades
of aircraft design experience attest to this. For example, AVROVulcan bomber had an
aspect ratio of just 3, small compared to that of Boeing B-47, which had an aspect ratio
of 9.4. Yet, both had L/Dratio of around 17. This is because Vulcan was a ying wing
with a wetted area of 892 m
2
and the wing area 320 m
2
, compared to B47, which was a
conventional design with a wetted area of 1,050 m
2
and the wing area 132 m
2
. The
ratio of wetted area to wing area was therefore 2.8 for Vulcan and 7.9 for B47, yielding
wetted aspect ratioA
wet
of 1.1 and 1.2 for Vulcan and B47, respectively. Consequently,
the L/Dratios were roughly the same for both (Raymer 2006). Note that a ying wing
would have a wetted area to wing area ratio of slightly more than 2.
Unfortunately, wetted areas are not easily available for birds and bats (let alone
insects). Obtaining the ratio of wetted surface area to the wing area along with a
better and more comprehensive data on L/D of migrating birds is therefore, highly
desirable and should be a priority for future research. Taking two extreme
examples, the albatross and the house wren, the albatross has an aspect ratio of 13
and a L/D ratio of 2024, while house wren has an aspect ratio of about 5 and L/D
Fig. 14 L/D ratio of various birds plotted against their aspect ratio, wing span, wing loading
and span loading from data from Tennekes (2009). In the second panel are also shown data
(red triangles) inferred from Templin (2000). Blue lines show robust least square ts to data
3 Migration Range 23
ratio of about 4.0. The birds appear to have a wetted area to wing area ratio of
approximately 2.32.5. The wetted aspect ratios are 5.6 and 2.0, respectively, for
albatross and wren. Using these two extreme limits and assuming a linear rela-
tionship between the L/D ratio and the wetted aspect ratio A
wet
, we get
L/D = 5A
wet
-6
Until more data become available, the migration range capability of a bird
should be regarded as an approximate function of its wetted aspect ratio and the fat
to body mass ratio at the beginning of migration.
4 Formation Flight
Many migrating birds tend to travel together in a V, J or W shaped formation, or
echelon formation of anywhere from a few birds to over a 100. Canada geese, snow
geese, brants, sandhill cranes, whooping cranes and american white pelicans are
among the birds that y in formation. The question is why do they do that? There have
been two schools of thought on this. One, consisting of mainly ornithologists,
advances the proposition that these formations are simply to promote better visual
contact and communication with each other without collisions among the ock. The
other, consisting of mainly aerodynamicists, holds the viewthat formation ights are
energetically protable, since they lead to signicant reductions in drag, and hence
the power and energy expended by each bird in the formation compared to its solo
ight. It is likely that both are equally good reasons for the birds to engage in
formation ight. The task is then to compute the advantages resulting fromformation
ight of a given shape. To do so, one needs to appeal to Prandtls famous lifting line
theory. In this theory, each wing can be replaced by a horseshoe vortex with the
circulation around the vortex dependant on the lift on the wing.
In a formation ight, every wing ies in the upwash induced by every other
wing of the formation. The total upwash w
m
induced on a given wing M is the sum
of the individual upwashes w
mn
induced by all other wings in the formation and in
general, varies along the span y of that wing. The average value of the upwash
is

W
m


N
W
n1

W
mn
where N
w
is the total number of wings in the formation.
Because of this upwash, the direction of lift us turned forward through an angle
Da
w
m
V
, which causes the drag to be reduced by DD
m
L
m
w
m
V
_ _
: The
corresponding reduction in the power required for ight is DP
m
VDD
m

L
m
w
m
; where L
m
is the lift and V is the ight speed. The fractional reduction in
ight power is
g
m

DP
m
P
m

L
D
_ _
m
w
m
V
31
24 Migration on Wings
Consider wing M. Its semi-span is b
m
and its center is located at (x
m
, y
m
). x is
the coordinate in the ight direction and y is the coordinate in the spanwise
direction. Let N be the inducing wing with semi-span b
n
with its center located at
(x
n
, y
n
). Let b be the normalizing length, the reference span. All the distances are
normalized by refernce span b. The normalized distance between the wing centers
in the ight direction is X
mn

X
m
X
n
b
: The normalized distance between the wing
centers in the spanwise direction is Y
mn

y
m
y
n
b
: The distance between wing tips
is Dy
mn
Y
mn

b
m
b
n
2
: The normalized distance between the wing centers in
the vertical direction is Z
mn

z
m
z
n
b
: The mean upwash w induced on wing M by
wing N is given by
w
mn
V

C
Ln
pA
n

f
mn
, where V is the ight speed, C
L
is the lift
coefcient and A is the aspect ratio. The parameter

f
mn
is a pure geometric
function dependent on the wing sizes and relative location and following Higdon
and Corrsin (1978) can be determined as follows:
Consider the effect of wing N on wing M. Assume wing N can be replaced by a
horse-shoe vortex consisting of two trailing vortices separated by a distance 2a
n
and
a bound vortex. Note that a
n

p
4
b
n
: Choosing to place the origin at the midspan of
wing N, the vertical velocity induced at point P (x, y, z) can be written as:
w x,y,z
C
n
2p
y a
n
y a
n

2
z
2
1
x

1
_ _
_ _

y a
n
y a
n

2
z
2
1
x

2
_ _
_ _ _

x y a
n

1
x
2
z
2

x y a
n

2
x
2
z
2

_ __
for y j j [a
n
w x,y,z
C
n
2p

a
n
y
a
n
y
2
z
2
1
x

1
_ _
_ _

y a
n
y a
n

2
z
2
1
x

2
_ _
_ _ _

x y a
n

1
x
2
+ z
2

x y a
n

2
x
2
z
2

_ __
for y j j\a
n
where the rst term in the curly brackets is due to the left vortex L, the second term
due to the right vortex R and the third term is due to the bound vortex and

1
x
2
z
2
y a
n

2
_ _
1=2
;
2
x
2
z
2
y a
n

2
_ _
1=2
: Positive w denotes
upwash and the negative value, downwash. C
n
is the circulation of the horse-shoe
vortex. Assuming the wings have elliptic lift distribution, C
n
C
max
n
since the
span is taken as 2a
n
and not the actual value 2b
n
. Wing M is located at a distance x
in the streamwise direction, z in the vertical direction and y
mn
in the spanwise
direction. The induced drag due to the vertical velocity w along the span of the
wing M due to wing N is (King and Gopalarathnam 2005):
4 Formation Flight 25
D
mn

q
2
C
m
_
y
mn
b
m
y
mn
b
m
w x; y; z dy: 33
The upper and lower limits correspond to the actual physical span of wing M and
therefore, the circulation C
m
must correspond to the average value C
m
C
av
m
and
not the maximum value C
m
C
max
m
. Substituting Eq. (32) in Eq. (33):
D
mn

qC
m
C
n
4p
1
2
ln
ya
n

2
z
2
ya
n

2
+z
2
_ _

1
2
ln

2
x
1
x

2
x
1
x
_ _

x
2
l
1

x
2
z
2

_ _
_ _
yy
mn
b
m
yy
mn
b
m
34
for both y j j [a
n
and y j j\a
n
: The rst term is the induced drag if the trailing
vortices of wing N were to extend from the spanwise position of wing M (which
they dont), the second term is due to the missing pieces of the vortices of wing N
that would have made them extend to semi-innity from the spanwise position of
the wing M, and the last term is due to the bound vortex of wing N. Note that for
x = x
mn
= 0, both the second and third terms vanish and only the contribution due
to the semi-innite trailing vortices of wing N remain. This comes in handy when
applying Munks stagger theorem, which states that in a formation, lifting surfaces
can be translated in the streamwise direction arbitrarily, without changing the
TOTAL induced drag of the formation (although individual values depend on the
exact location of individual wings in the space). This means all the wings in the
formation can be moved to a single vertical plane so that x
mn
= 0 and the sim-
plied form of Eq. (34) can then be used to calculate the total induced drag of the
formation. Substituting the upper and lower limits:
D
mn

qC
m
C
n
4p
f
mn
f
mn

1
2
ln
z
2
y
mn
a
n
b
m

2
z
2
y
mn
a
n
b
m

2

z
2
y
mn
a
n
b
m

2
z
2
y
mn
a
n
b
m

2
_ _

1
2
ln

2
x
_ _

1
x
_ _

2
x
_ _

1
x
_ _

2
x
_ _

1
x
_ _

2
x
_ _

1
+ x
_ _
_ _

x
x
2
z
2

2
_ _

1
_ _ _
_ _
35
where

2
x
2
z
2
y
mn
a
n
b
m

2
_ _
1=2
;

2
x
2
z
2
y
mn
a
n
b
m

2
_ _
1=2

1
x
2
z
2
y
mn
a
n
b
m

2
_ _
1=2
;

1
x
2
z
2
y
mn
a
n
b
m

2
_ _
1=2
36
26 Migration on Wings
Once again, the rst curly brackets are due to semi-innite vortices, the second due
to missing pieces and the third due to the bound vortex. Both the second and third
terms vanish for x = 0. All the quantities in curly brackets involve distances and
hence can be normalized by the reference span b so that

f
mn

1
2
ln
z
2
mn
Y
mn
a
n
b
m

2
z
2
mn
Y
mn
a
n
b
m

2

z
2
mn
Y
mn
a
n
b
m

2
z
2
mn
Y
mn
a
n
b
m

2
_ _

1
2
ln

2
X
mn
_ _

1
X
mn
_ _

2
X
mn
_ _

1
X
mn
_ _

2
X
mn
_ _

1
X
mn
_ _

2
X
mn
_ _

1
+ x
mn
_ _
_ _

X
mn
x
2
mn
z
2
mn
_ _

2
_ _

1
1

1
_ _ _
_ _

2
X
2
mn
z
2
mn
Y
mn
a
n
b
m

2
_ _
1=2
;

2
X
2
mn
z
2
mn
Y
mn
a
n
b
m

2
_ _
1=2
and

1
X
2
mn
z
2
mn
Y
mn
a
n
b
m

2
_ _
1=2
;

1
X
2
mn
z
2
mn
Y
mn
a
n
b
m

2
_ _
1=2
37
where quantities a
n
, b
m
,

2
;

2
;

1
;

1
are now dimensionless having been nor-
malized by reference span b. Note that
D
mn

qC
m
C
n
4p

f
mn
38
In the case where there is no vertical separation between wings, Z
mn
= 0, and
we get

f
mn
ln
Y
mn
a
n
b
m

Y
mn
a
n
b
m

Y
mn
a
n
b
m

Y
mn
a
n
b
m

_ _

1
2
ln

2
X
mn
_ _

1
X
mn
_ _

2
X
mn
_ _

1
X
mn
_ _

2
X
mn
_ _

1
X
mn
_ _

2
X
mn
_ _

1
+ X
mn
_ _
_ _

1
X
mn

2
_ _

1
_ _ _
_ _
4 Formation Flight 27

2
X
2
mn
Y
mn
a
n
b
m

2
_ _
1=2
;

2
X
2
mn
Y
mn
a
n
b
m

2
_ _
1=2
with

1
X
2
mn
Y
mn
a
n
b
m

2
_ _
1=2
;

1
X
2
mn
Y
mn
a
n
b
m

2
_ _
1=2
39
Note that

f
mn
6 0 for m = n. These expressions are similar to that presented by
Hummel (1983) except for a multiplying factor. The terms due to missing trailing
vortex pieces however do not agree:
p
2
2
B
m
B
n
_ _

f
mn
ln
Y
mn
a
2
Y
mn
a
2

Y
mn
a
1
Y
mn
a
1

_ _
ln
Y
mn
a
2
X
mn

Y
mn
a
2

2
X
2
mn
_
Y
mn
a
2
X
mn

Y
mn
a
2

2
X
2
mn
_
_

_
_

_
ln
Y
mn
a
2
X
mn

Y
mn
a
2

2
X
2
mn
_
Y
mn
a
2
X
mn

Y
mn
a
2

2
X
2
mn
_
_

_
_

_
ln
Y
mn
a
1
X
mn

Y
mn
a
1

2
X
2
mn
_
Y
mn
a
1
X
mn

Y
mn
a
1

2
X
2
mn
_
_

_
_

_
ln
Y
mn
a
1
X
mn

Y
mn
a
1

2
X
2
mn
_
Y
mn
a
1
X
mn

Y
mn
a
1

2
X
2
mn
_
_

_
_

1
X
mn

Y
mn
a
2

2
X
2
mn
_

Y
mn
a
2

2
X
2
mn
_
_

Y
mn
a
1

2
X
2
mn
_

Y
mn
a
1

2
X
2
mn
_
40
where B
m

b
m
b
; B
n

b
n
b
; a
1

p
8
B
m
B
n
and a
2

p
8
B
m
B
n

41
Note that

f
mn
0 for m = n. We will use Eq. (38) instead of Eq. (39) henceforth.
In the case where there is no vertical and streamwise separation between wings,
X
mn
= Z
mn
= 0, and we get
D
mn

qC
m
C
n
4p

f
mn
;

f
mn
ln
Y
2
mn
a
n
b
m

2
_ _
Y
2
mn
a
n
b
m

2
_ _
_
_
_
_
_
_
42
28 Migration on Wings
This expression can be used if one is interested in only the induced drag reduction
of the entire ock and not the individuals. The self-induced drag of wing M can be
deduced by putting Y
mn
= 0 also:
D
mn

qC
max
m
C
av
m
4p

f
mn
;

f
mn
2 ln
b
m
a
m
b
m
a
m
_ _
43
Note that both average and maximum circulation values of wing M are used. This
is simply because the average value obtains if the full span b
m
is considered as it is
in computing the induced drag due to the horse-shoe vortex of wing M, whose
spacing is however 2a
m
and strength is the maximum circulation. The two are
related by C
av
m

p
4
C
max
m
as can be seen from below:
If the lift on a wing is elliptically distributed: C y C
max
1
y
2
b
2
_ _
1=2
; where b is
the semi-span, then the average value of the circulation is C
av

pbC
max
2
1
2b

p
4
C
max
: The lift on the wing is L qV
1
_
b
b
Cydy qV
1
pbC
max
2
_ _

qV
1
2aC
max
qV
1
2bC
av
so that the lifting wing can be represented by a
horse-shoe vortex of strength C
max
with spacing 2a, as is the usual custom with
a
p
4
b. However, for calculating D
mn
, we have a choice of using C
max
and
replacing b
m
in earlier equations by a
m
, in which case D
mn
in Eq. (42) becomes
singular, or using C
av
as we have done. Eq. (43) therefore gives for the self-
induced drag of wing M:
D
I

qC
max
m
C
av
m
2p
ln
4 p
4 p
_ _
0:337qC
max
m
C
av
m
44
which is much smaller than that for an elliptically loaded wing, which can be
deduced as follows:
D
I

1
pq
W
2b
_ _
2
and W qV
1
C
max
2a !
C
max
V
1

2
pq
W
2b
_ _

2W
pqV
2
1
b

SC
L
pb
so that D
I

p
8
q C
max

2

2
p
q C
av

2

1
2
qC
max
C
av
45
The underestimate is due to replacing the bound vortex of varying strength and the
vortex sheet behind the elliptically loaded wing by a horse-shoe vortex of strength
C
max
and spacing 2a, a gross approximation at best. So we will use Eq. (45) for the
induced drag of an isolated wing so that the fractional reduction in induced drag of
wing M due to wing N is:
4 Formation Flight 29
D
mn
D
Im

qC
av
m
C
max
n
4p

f
mn
_ _
qC
av
m
C
max
n
2
_ _

1
2p
C
max
n
C
max
m
_ _

f
mn

1
2p
W
2b
_ _
n
W
2b
_ _
m

f
mn
46
Thus only the span loading of the two wings and the geometrical factor

f
mn
given
by Eqs. (37), (39) or (42) need to be considered. For a formation of N
w
wings, the
total fractional reduction in the induced drag of wing M is
DD
1
D
1
_ _
m

1
D
Im

N
W
n1
D
mn

1
2p
W
2b
_ _
m

N
W
n1
W
2b
_ _
b

f
mn
47
Note that

f
mn
0: The fractional reduction in total drag can be obtained by
multiplying Eq. (47) by the ratio
C
D
i
C
D
_ _
m
so that, since power expended is pro-
portional to the drag (velocity being constant), the fractional power reduction of
wing M due to other wings in the formation is:
g
m

DP
m
P
m

DD
D
_ _
m

C
D
i
C
D
_ _
m
DD
I
D
I
_ _
m

1
2p
W
2b
_ _
m
C
D
i
C
D
_ _
m

N
W
n1
W
2b
_ _
n

f
mn
48
The advantage of writing it this way is that one can substitute for
C
D
i
C
D
_ _
a value
of
1
2
for ight at maximum range and
3
4
for ight at minimum power.
The total reduction in power demand of the entire formation is DP

N
w
m1
DP
m
: The power demand of the wings in the formation if each ew solo is
P

N
W
m1
P
m
so that the fractional power reduction of the whole formation
g
DP
P

N
W
m1
DP
m

N
W
m1
P
m

N
W
m1
DD
m

N
W
m1
D
m

N
W
m1
g
m
D
m

N
W
m1
D
m
: 49
It can be shown that this is a consequence of Munks stagger theorem, which states
that The total induced drag of a collection of lifting surfaces is unaltered by any
translation of the wings in the ight direction as long as the circulation of every
wing is unchanged. This means that g is independent of X
mn
, even though g
m
is a
strong function of X
mn
of the wings in the formation. Therefore, a ock of birds in
line abreast formation will have the same induced drag (and hence the same total
power saved) if the formation changed to a V formation or an echelon formation.
Thus staggering in the ight direction does not change the total induced drag and
hence the total power reduction or power reduction per bird in the ock. However,
staggering causes redistribution of the induced drag, and hence the power saving
of an individual bird depends on the formation and its location in that formation. In
a V-formation, the leading bird and its immediate followers, and the ones toward
30 Migration on Wings
the ends of the V benet less from formation ight than the ones in the middle of
the arms of V. In a line-abreast formation, the birds toward the ends benet less
and the same holds for echelon formation.
Equation (48) can be used to determine the power saving of each bird in
formation ight of any shapeV, W, J, echelon or abreastformation. There is
no restriction either on the bird characteristics such as span, weight and wing
loading, which can assume arbitrary values.
The special case of just two identical birds in formation ight is instructive.
Figure 15a shows the fractional increase in induced drag of each bird for various
spanwise and vertical separations. The benecial location is to the right of the zero
contour line. The maximum benet occurs when the two wings are slightly
overlapping with Y
mn
* 0.8. To the left of the zero contour line, there is drag
penalty since the induced drag increases over the solo ight value. If the two birds
ew exactly one behind the other, the induced drag would double if the vertical
separation is zero. However, with vertical spacing beyond roughly one span, the
induced drag increase becomes less than 10%. Note that if the lift were to reverse
on wing B (as in a mirrored wing in ground effect), the areas of benecial and
detrimental induced drag change would switch, with the regions to the right of the
zero contour being detrimental instead of benecial. This fact can be used to
compute the induced drag changes, when the formation ight is in ground effect
and each wing is mirrored in the ground (lift vector is reversed).
Fig. 15 Fractional induced drag increase during the formation ight of two identical birds
4 Formation Flight 31
Figure 15b shows the fractional induced drag change as a function of the spanwise
separation for various vertical separations. For ight in the same horizontal plane, a
slight overlapinthe wings gives the maximumreduction. However, ight withone bird
immediately behind the other doubles the induced drag. Exact solution fromKing and
Gopalarathnam(2005) for elliptically loaded wings ying at the same horizontal level
is shown as green lled circles. Clearly, the horse-shoe vortex model tends to work
poorly for 0.7 \Y
mn
\1.1. Another noteworthy aspect is that the vertical separation
of h/b [0.5 cuts down the induced drag increase so that by ying above or below the
leading bird, the drag penalty can be reduced.
One simple sub case is the formation ight of identical birds with uniform
spacing between them. In this case,
g
m

DC
D
C
D
_ _
m

1
2p
C
Di
C
D
_ _
m

N
W
n1

f
mn
and g
1
N
W

N
W
m1
g
m
50
Consider a ock of identical birds with the same spanwise spacing. Figure 16
shows the fractional power saving of the entire ock as a function of the number of
birds for different ratios of spanwise spacing to the span (s/b) for two well known
ight conditions: 1. Minimum power expended (blue solid lines) and 2. Maximum
range (black dotted lines). Recall that the power saving does not depend on the
shape of the formation, although the distribution of fractional power saving among
birds depends very much on the shape. Clearly, the power saving decreases rapidly
with the number of birds and the benet of formation ight does not increase much
beyond ock strength of around 30 or so. The spanwise spacing is rather critical. A
tight formation, no matter what the shape, leads to large power savings. The best
benet is for wing overlap corresponding to
s
b

p
4
1 and the benet can be as
much as 75% for minimum power ight (and 50% for maximum range). At this
point, the induced drag of the ock is equal to that of a single bird! However, this
conguration is hard to maintain without birds drifting to the inside of the
horseshoe vortex of the adjacent bird and hence coming under the inuence of its
downwash instead of its upwash. It is probably easier to maintain an overlap
corresponding to s/b = -0.1 or 0. Even then, the power savings are signicant.
Beyond s/b values of around 1 or so, the power saving becomes rather minimal. As
a consequence, the ock must maintain a tight formation, irrespective of the shape
of the formation, whether abreast, echelon or V. Naturally, from the point of view
of visual contact with the birds ahead, either an echelon or V formation is highly
desirable. These are indeed the most commonly observed formations. Of course,
all the birds in the formation must be at the same level to realize the maximum
benet of formation ight, although again from the point of view of visual contact,
it helps to y slightly above the birds ahead in the formation.
It is also noteworthy that as the ight speed increases, C
D0
/C
Di
increases
reecting the fact that the induced drag decreases with speed while the form and
skin friction drag increase. Therefore, the power saving comes down as ight
32 Migration on Wings
speed increases. This can be seen from Fig. 16, which shows that for the same s/b
values, the black curves corresponding to maximum range lie below the red lines
corresponding to minimum power. The ight speed for maximum range is 31.6%
higher than that for minimum power. The benets of formation ight are best
realized if the ight speeds are low. This may encourage the formation to y at
speeds corresponding to minimum power expenditure by the ock as a whole.
An alternative way to present power saving due to formation ight is to present
the fractional power saving as a function of wing spacing s/b for different number
of birds. Figure 17 does so for ock speeds corresponding to the velocity for
maximum range (black dotted) and minimum power (blue solid) of solo ight.
Clearly, even formation of 23 birds reaps signicant power savings. It is not at all
unusual to see formations of as few as 25. An additional noteworthy fact is that as
the number of birds increases, the fractional power saving tends to asymptote
beyond a value of around 25 birds in the ock.
However, the fractional power saving of individual birds in the formation
depends on its location in the formation and the shape of the formation. This can
Fig. 16 The fractional power saving of the entire ock as a function of number of birds for
different values of s/b. Blue curves correspond to minimum power, for which the ight speed is
lower than that corresponding to maximum range. Black curves correspond to maximum range
4 Formation Flight 33
be seen from Fig. 18, which shows the values for each bird of the formation of 21
birds, for abreast, echelon and V formations and ight speed corresponding to the
minimum power requirement of individual birds. The fractional power saving is
0.283 for the entire ock, but the saving for an individual bird depends on its
position in the formation. The saving is particularly small for the leading bird in
both echelon and V formations. This is simply because the upwash decreases
signicantly ahead of the horseshoe vortex and the leading bird, because of its
location, does not benet from the trailing vortices of any bird in the formation. In
the particular example shown, the fractional power saving of the leading bird is
only 0.08 for V formation and 0.04 for echelon formation.
This does not apply to abreast formation since then every bird benets from the
trailing vortices of the rest of the ock. Noticeably the trailing bird in either the V
or echelon formation does not suffer signicantly (the fractional power savings are
0.270 and 0.271, respectively compared to the overall ock saving of 0.283). The
birds in the middle of both formations benet more. Consequently, it helps to be in
the middle of these formations and it is not unusual for birds joining the formation
to try to do so in the middle! On the other hand, the birds at each end of the abreast
Fig. 17 The fractional power saving of the entire ock as a function of s/b for different values of
number of birds. Blue solid curves correspond to minimumpower, for which the ight speed is lower
than that corresponding to maximum range. Black dotted curves correspond to maximum range
34 Migration on Wings
F
i
g
.
1
8
T
h
e
f
r
a
c
t
i
o
n
a
l
p
o
w
e
r
s
a
v
i
n
g
o
f
i
n
d
i
v
i
d
u
a
l
b
i
r
d
s
i
n
t
h
e
f
o
r
m
a
t
i
o
n
f
o
r
a

o
c
k
c
o
n
s
i
s
t
i
n
g
o
f
2
1
b
i
r
d
s
.
T
h
e
f
r
a
c
t
i
o
n
a
l
p
o
w
e
r
s
a
v
i
n
g
i
s
0
.
2
8
3
f
o
r
t
h
e
e
n
t
i
r
e

o
c
k
.
B
u
t
t
h
e
s
a
v
i
n
g
f
o
r
a
n
i
n
d
i
v
i
d
u
a
l
b
i
r
d
d
e
p
e
n
d
s
o
n
i
t
s
p
o
s
i
t
i
o
n
i
n
t
h
e
f
o
r
m
a
t
i
o
n
4 Formation Flight 35
formation do not derive as much benet (only 0.156 compared to the overall ock
saving of 0.283) as the rest of the ock. Since the benet increases toward the
center of the ock and becomes maximum there (0.305), during long ights, the
formation might tend toward V formation because the birds with lesser benet
might not be able to keep up with the abreast formation. It is in fact interesting to
note that a slight deformation of the abreast formation into a U formation leads to
roughly equal distribution of power saving to all birds in the ock, including the
leading bird (Hummel 1983).
The fact that the leading bird in both echelon and V formations derives the least
(but nonzero) benet from formation ight indicates that during long ights, either
the formation must be led by a strong bird or another bird must switch places with the
leading bird. In fact, switching has been observed to happen. If the formation is
traveling at a speed that is considerably less than that corresponding to the maximum
continuous power available fromthe ight muscles, it is possible that the leading bird
can keep its position in the formation for considerable time as long as it is not too
weak. There are indications that migrating birds travel at ight speeds considerably
higher than those corresponding to the maximumrange, simply because the available
power output of the ight muscles permits them to. However, the fractional power
saving of the ock and hence the individual birds decreases as the ight speed
increases. As shown earlier, the migration distance depends only on the aerodynamic
efciency (L/D) of the bird and the fat (and muscle) fraction of the total body mass
converted into mechanical energy. Thus, there is an incentive to travel at maximum
possible L/Dratio. While the birds are certainly capable of recognizing when they are
expending minimum power and when they are at the limit of their muscular power,
they might not be able to recognize the ight conditions corresponding to maximum
L/D, unless of course evolution has conditioned them. This is unlikely since the
attainable range is such a complex and unknowable function of parameters such as
the wind speed and direction along the entire migration path and the ight altitude.
The ight speed for maximum range for formation ight is less than that
corresponding to an individual bird. However, this ight speed is rather close to
the solo ight speed for minimum power. This is fortuitous since the leader of the
ock derives minimum benet from formation ight and is therefore quite likely to
y at a speed corresponding to solo minimum power condition, forcing others in
the formation to y at that speed, which however is close (but not equal) to the
speed for optimum range condition for the ock as a whole.
The geometric function can be simplied so that one can write, for the whole
ock (Kshatria and Blake 1992),
DC
Di
C
Di

4
p
2
N
W

N
W
1
n1

N
W
mn1
ln 1
p
4
mn j j 1
s
b
_ _
_
_
_
_
2
_

_
_

_: 51
where s is the spacing between wing tips and b is the wing span. Therefore the total
fractional power saving of the ock is
36 Migration on Wings
DP
P

DC
D
j j
C
D

DC
Di
j j
C
D0
+ C
Di

DC
Di
C
Di

1
C
D0
C
Di
: 52
Note that
C
D0
C
Di
1 for maximum range and 1/3 for minimum power. In the limit
N
W
!1;
DC
Di
C
Di

4
p
2
ln sin
p
2
_
4
1
s
b
_ _
=
p
2
_
4
1
s
b
_ _ _ _
53
For the special case of S =
p
4
1; the trailing vortices of adjacent wings overlap
and cancel so that the induced drag of the formation becomes the induced drag of a
wing of span N
w
b! The induced drag of the ock is equal to the induced drag of a
single bird ying solo. Therefore
DC
Di
j j
C
Di

1
N
w
1 and the fractional power saved is
DP
P

DC
D
j j
C
D

1
1
N
W
_ _
1
C
D0
C
Di
_ _
54
5 Great Flight Diagrams
Finally, heavier-than-air ight, both in nature by birds, bats and insects, and by man-
made aircraft, requires wings to produce the lift needed to sustain ight. Conse-
quently, it is not unrealistic to expect certain universal scaling to exist. Tennekes
(2009) has popularized the concept that the wing loading, dened as the weight of
the ying object W (unit is N) divided by the wing area S (unit is m
2
) sustaining the
ight, is the most universal attribute of ight, whether it is natural ying objects such
as birds and butteries, or man-made airplanes. When the weight of the ying object
W is plotted against the wing loading WL = W/S (unit is N/m
2
), the data points tend
to fall around a universal line for weights ranging from 10
-5
N corresponding to a
fruit y to 5.6 9 10
6
N corresponding to the heaviest civilian airliner Airbus A380
(his Fig. 2). This is nearly 12 orders of magnitude change in weight. Such a cor-
relation is quite impressive, especially given the fact that wing loading appears
prominently in considerations of ight speed of birds and planes, and landing and
takeoff speeds of aircraft. Tennekes (2009) calls the plot the Great Flight Diagram.
Figure 19 shows W plotted against W/S for data tabulated in Appendix A,
which includes those from Tennekes (2009). Unlike in Tennekes (2009), the data
4 Formation Flight 37
F
i
g
.
1
9
T
h
e
G
r
e
a
t
F
l
i
g
h
t
D
i
a
g
r
a
m
w
i
t
h
w
e
i
g
h
t
W
p
l
o
t
t
e
d
a
g
a
i
n
s
t
w
i
n
g
l
o
a
d
i
n
g
W
L
=
W
/
S
,
u
p
d
a
t
e
d
f
r
o
m
T
e
n
n
e
k
e
s
(
2
0
0
9
)
.
C
i
r
c
l
e
s
c
o
r
r
e
s
p
o
n
d
t
o
b
i
r
d
s
,
w
i
t
h
d
i
f
f
e
r
e
n
t
c
o
l
o
r
s
f
o
r
d
i
f
f
e
r
e
n
t
s
o
u
r
c
e
s
o
f
d
a
t
a
,
s
q
u
a
r
e
s
t
o
b
a
t
s
,
d
i
a
m
o
n
d
s
t
o
i
n
s
e
c
t
s
,
a
n
d
v
e
r
t
i
c
a
l
t
r
i
a
n
g
l
e
s
t
o
a
i
r
c
r
a
f
t
.
R
i
g
h
t
-
p
o
i
n
t
i
n
g
t
r
i
a
n
g
l
e
s
d
e
n
o
t
e
h
u
m
a
n
-
p
o
w
e
r
e
d
a
i
r
c
r
a
f
t
a
n
d
l
e
f
t
-
p
o
i
n
t
i
n
g
t
r
i
a
n
g
l
e
s
t
o

g
h
t
e
r
a
i
r
c
r
a
f
t
.
B
l
u
e
l
i
n
e
i
s
t
h
e
r
e
g
r
e
s
s
i
o
n
l
i
n
e
i
n
h
i
s
F
i
g
.
2
38 Migration on Wings
includes bats and a large number of insects, and is more comprehensive. While
there exists an approximate correlation between W and W/S as in Tennekes
(2009), a closer examination shows considerable scatter and signicant departures
from the regression line, especially for insects. The scatter is likely due to the fact
that the data points correspond to ight objects with disparate aspect ratios. Also,
wing loading is a parameter that determines the performance of the wing so far as
its cruise, landing and takeoff speed characteristics are concerned, and not nec-
essarily the overall wing design. Consequently, we see signicant departures of
insect wings (green diamonds) as well as gliders and human-powered aircraft
(right-pointing red triangles) from the regression line. Given the different structural
design and the material of bird and insect wings, this is not surprising. The same
applies to the wings of human-powered aircraft, which are low weight, low speed
designs by necessity. Military ghter aircraft (leftt-pointing red triangles), whose
design emphasis is more on maneuverability and performance in combat, rather
than low landing and takeoff speeds, also depart signicantly from the regression
line, whereas birds (circles) and bats (squares), and transport aircraft (vertical red
triangles) pretty much cluster around the regression line.
There are other parameters that inuence the design of wings, both natural and
man-made. One such parameter is the span loading SL = W/b (unit is N/m), which
determines the induced drag of the wing. As mentioned earlier, the induced drag
can be thought of as the penalty for production of lift by wings of nite span, the
induced drag being zero for a wing of innite span. The induced drag is not a
function of the wing loading; the ratio of the induced drag to the parasitic drag is,
this ratio also being a function of the aspect ratio. This suggests that for ying
objects with widely varying aspect ratios, the parameter that better optimizes the
wing design might be the span loading. In fact, Templin (2000) suggests that it is
the span loading that is more important to ight and not the wing loading or the
aspect ratio, even though wing loading is of importance to takeoff and landing.
However, he uses wing span b normalized by the body length l, which he denes
as l = (W/35)
1/3
for his analyses of the ight of birds and bats. It is therefore
useful to replot the Great Flight Diagram using the span loading instead of the
wing loading as the independent parameter. Figure 20 shows such a plot of W vs
SL, from data tabulated in Appendixes AH, including those from Tennekes
(2009). Note that wing span is a quantity more easily determined for natures yers
than the wing area and consequently, the data on wing span is more abundant and
explains the vastly larger number of data points in Fig. 20 compared to Fig. 19.
The correlation between the weight and span loading is tighter over 8 orders of
magnitude change in SL, than that between the weight and wing loading over 5
orders of magnitude change in WL, as can be seen by comparing Figs. 19 and 20.
The scatter is signicantly diminished. The span loading data for gliders and very
light craft follow a similar slope to that of birds, since the emphasis in the wing
design is on light weight. Birds of course have been evolutionally ne-tuned to be
very light iers. The aircraft data also fall roughly on the same line.
Figure 21 shows a plot of the weight W versus span loading SL for just the
natures iers: birds and bats. Note that the wingspan is readily measured, whereas
5 Great Flight Diagrams 39
F
i
g
.
2
0
T
h
e
s
e
c
o
n
d
G
r
e
a
t
F
l
i
g
h
t
D
i
a
g
r
a
m
r
e
s
u
l
t
i
n
g
f
r
o
m
p
l
o
t
t
i
n
g
w
e
i
g
h
t
W
v
e
r
s
u
s
s
p
a
n
l
o
a
d
i
n
g
S
L
=
W
/
b
f
r
o
m
d
a
t
a
i
n
A
p
p
e
n
d
i
x
A

H
.
C
i
r
c
l
e
s
c
o
r
r
e
s
p
o
n
d
t
o
b
i
r
d
s
,
w
i
t
h
d
i
f
f
e
r
e
n
t
c
o
l
o
r
s
f
o
r
d
i
f
f
e
r
e
n
t
s
o
u
r
c
e
s
o
f
d
a
t
a
,
s
q
u
a
r
e
s
t
o
b
a
t
s
,
d
i
a
m
o
n
d
s
t
o
i
n
s
e
c
t
s
,
a
n
d
v
e
r
t
i
c
a
l
t
r
i
a
n
g
l
e
s
t
o
a
i
r
c
r
a
f
t
.
R
i
g
h
t
-
p
o
i
n
t
i
n
g
t
r
i
a
n
g
l
e
s
d
e
n
o
t
e
h
u
m
a
n
-
p
o
w
e
r
e
d
a
i
r
c
r
a
f
t
a
n
d
l
e
f
t
-
p
o
i
n
t
i
n
g
t
r
i
a
n
g
l
e
s
t
o

g
h
t
e
r
a
i
r
c
r
a
f
t
.
B
l
u
e
l
i
n
e
i
s
t
h
e
l
e
a
s
t
s
q
u
a
r
e

t
l
i
n
e
40 Migration on Wings
F
i
g
.
2
1
T
h
e
s
e
c
o
n
d
G
r
e
a
t
F
l
i
g
h
t
D
i
a
g
r
a
m
f
o
r
b
i
r
d
s
a
n
d
b
a
t
s
.
W
e
i
g
h
t
W
i
s
p
l
o
t
t
e
d
a
g
a
i
n
s
t
s
p
a
n
l
o
a
d
i
n
g
S
L
=
W
/
b
f
o
r
a
l
l
a
v
a
i
l
a
b
l
e
d
a
t
a
,
w
h
i
c
h
a
r
e
t
a
b
u
l
a
t
e
d
i
n
A
p
p
e
n
d
i
c
e
s
A
t
o
H
.
C
i
r
c
l
e
s
c
o
r
r
e
s
p
o
n
d
t
o
b
i
r
d
s
,
w
i
t
h
d
i
f
f
e
r
e
n
t
c
o
l
o
r
s
f
o
r
d
i
f
f
e
r
e
n
t
s
o
u
r
c
e
s
o
f
d
a
t
a
a
n
d
s
q
u
a
r
e
s
t
o
b
a
t
s
.
N
o
t
e
t
h
e
t
i
g
h
t
c
o
r
r
e
l
a
t
i
o
n
o
v
e
r
f
o
u
r
o
r
d
e
r
s
o
f
m
a
g
n
i
t
u
d
e
i
n
w
e
i
g
h
t
!
5 Great Flight Diagrams 41
F
i
g
.
2
2
T
h
e
t
h
i
r
d
G
r
e
a
t
F
l
i
g
h
t
D
i
a
g
r
a
m
r
e
s
u
l
t
i
n
g
f
r
o
m
p
l
o
t
t
i
n
g
w
e
i
g
h
t
W
v
s
t
h
e
w
i
n
g
m
o
m
e
n
t
W
M
=
W
b
f
r
o
m
d
a
t
a
i
n
A
p
p
e
n
d
i
x
A

H
.
C
i
r
c
l
e
s
c
o
r
r
e
s
p
o
n
d
t
o
b
i
r
d
s
,
w
i
t
h
d
i
f
f
e
r
e
n
t
c
o
l
o
r
s
f
o
r
d
i
f
f
e
r
e
n
t
s
o
u
r
c
e
s
o
f
d
a
t
a
,
s
q
u
a
r
e
s
t
o
b
a
t
s
,
d
i
a
m
o
n
d
s
t
o
i
n
s
e
c
t
s
,
a
n
d
v
e
r
t
i
c
a
l
t
r
i
a
n
g
l
e
s
t
o
a
i
r
c
r
a
f
t
.
R
i
g
h
t
-
p
o
i
n
t
i
n
g
t
r
i
a
n
g
l
e
s
d
e
n
o
t
e
h
u
m
a
n
-
p
o
w
e
r
e
d
a
i
r
c
r
a
f
t
a
n
d
l
e
f
t
-
p
o
i
n
t
i
n
g
t
r
i
a
n
g
l
e
s
t
o

g
h
t
e
r
a
i
r
c
r
a
f
t
.
B
l
u
e
l
i
n
e
i
s
t
h
e
l
e
a
s
t
s
q
u
a
r
e

t
t
o
a
l
l
t
h
e
d
a
t
a
42 Migration on Wings
F
i
g
.
2
3
T
h
e
t
h
i
r
d
G
r
e
a
t
F
l
i
g
h
t
D
i
a
g
r
a
m
f
o
r
b
i
r
d
s
a
n
d
b
a
t
s
.
W
e
i
g
h
t
W
i
s
p
l
o
t
t
e
d
a
g
a
i
n
s
t
w
i
n
g
m
o
m
e
n
t
W
M
=
W
b
f
o
r
a
l
l
a
v
a
i
l
a
b
l
e
d
a
t
a
,
w
h
i
c
h
a
r
e
t
a
b
u
l
a
t
e
d
i
n
A
p
p
e
n
d
i
c
e
s
A
t
o
H
.
C
i
r
c
l
e
s
c
o
r
r
e
s
p
o
n
d
t
o
b
i
r
d
s
,
w
i
t
h
d
i
f
f
e
r
e
n
t
c
o
l
o
r
s
f
o
r
d
i
f
f
e
r
e
n
t
s
o
u
r
c
e
s
o
f
d
a
t
a
a
n
d
s
q
u
a
r
e
s
t
o
b
a
t
s
.
N
o
t
e
t
h
e
t
i
g
h
t
c
o
r
r
e
l
a
t
i
o
n
o
v
e
r
s
i
x
o
r
d
e
r
s
o
f
m
a
g
n
i
t
u
d
e
i
n
w
i
n
g
m
o
m
e
n
t
!
5 Great Flight Diagrams 43
the wing area is harder to discern for birds, bats and insects. Consequently, there
are far more data on wingspan available for birds and bats than wing area. These
data are tabulated in Appendix B, C and D for birds and E for bats, and plotted in
Fig. 21. It is noteworthy that all these data cluster beautifully around W * SL
3/2
line and the correlation is tight! The black circles correspond to Tennekes data for
birds (Appendix A), blue circles to data for birds listed in Appendices B, C and D,
and the red squares to data for bats (Appendix E).
The third Great Flight Diagram parameter is the moment at the wing root
equal to M
_
b=2
0
L(y)y dy; where L(y) is the local lift on the wing and y is
the spanwise distance. This moment determines how strong the wing has to be
and therefore its weight. The higher this moment, the heavier the wing has
to be. The exact value of this moment depends on the lift distribution. For
the canonical elliptic lift distribution L(y) L
c

1
2y
b
_ _
2
_
; M
L
C
b
2
12
. Since
W
_
b=2
b=2
L(y)dy
pbL
c
4
; M Wb/ 3p 0:106 Wb: For uniform lift distribu-
tion, M = 0.125 Wb. Thus no matter what the lift distribution, the wing root
moment is proportional to Wb. Thus the third Great Flight Diagram can be
based on wing root moment WM = Wb. Figure 22 shows the weight W plotted
against the wing root moment Wb for natures yers as well as man-made
aircraft. The correlation is even tighter than the W versus SL plot over 16
orders of magnitude change in WM from the tiniest insect to the superjumbo
A380 (compare Figs. 21 and 22)!
Figure 23 shows the third ight diagram for just the natures yers. Once
again, the tight correlation is self-evident. The correlation is better than in the
second Great Flight Diagram, which is in turn better than the rst Great Flight
Diagram.
6 Concluding Remarks
While migration on wings has always been of great interest to humanity, it has
assumed increased importance lately. Migrating birds have the potential to
quickly spread harmful pathogens worldwide. The possibility of H5N1 avian u
virus becoming a worldwide pandemic (once the virus has mutated to infect
humans) with the potential to inict huge causalities has underscored the need
to better understand and predict bird migration patterns and behavior, so that
effective countermeasures can be put in place. On another front, the eventual
depletion of fossil fuels, and the phenomenal changes in climate the burning of
fossil fuels can cause, has focused mankinds attention on renewable sources of
energy such as winds. However, care must be taken to make sure that wind
turbine farms are located so as not to interfere with, hurt or kill migrating
birds.
44 Migration on Wings
Fortunately, modern technology gives us the capability to study migration much
better than has been possible in the past, as the example of the bar-tailed godwit
cited above shows. New technologies are coming on stream for monitoring bird
migration (Fiedler 2009) and proven techniques such as radar tracking (Nebuloni
et al. 2008) and satellite tracking are being increasingly useful. Hopefully, in the
coming decades, we will be able to completely decipher this fascinating phe-
nomenon of migration on wings.
Finally, it is important for us to realize that migration is in itself an arduous task
and not all the birds, which embark on seasonal migrations complete it (Obviously,
enough do to preserve and propagate the species). Migrating birds encounter
numerous perils along the way: adverse weather, hunters and predators, to name a
few. At the end of their journey, the birds are not only tired, but also very fam-
ished, especially if they have encountered adverse weather conditions along the
way. This highlights the importance of establishing and preserving bird sanctuaries
along migration paths, where the birds can rest and rebuild their fat reserves
undisturbed by hunters before taking off on their journey again. This is the least we
can do to preserve this epic phenomenon of migration on wings for future gen-
erations to observe and enjoy!
References
T. Alerstam, Bird Migration (Cambridge University Press, Cambridge, 1990)
T. Alerstam, A. Hedenstrom, The development of bird migration theory. J. Avian Biol. 29,
343369 (1998)
J. Elphick (ed.), Atlas of Bird Migration. (Firey Books, Buffalo, 2009), p. 176
W. Fiedler, New technologies for monitoring bird migration and behaviour. Ringing Migr. 24,
175179 (2009)
C.H. Greenewalt, The ight of birds. Trans Am Philos Soc New Ser. 65, 167 (1975)
J.J.L. Higdon, S. Corrsin, Induced drag of a bird ock. Am. Nat. 112, 727744 (1978)
D. Hummel, Aerodynamic aspects of formation ight in birds. J. Theor. Biol. 104, 321347
(1983)
R.M. King, A. Gopalarathnam, Ideal aerodynamics of ground effect and formation ight. J. Aircr.
42, 11881199 (2005)
M. Kshatriya, R.W. Blake, Theoretical model of the optimum ock size of birds ying in
formation. J. Theor. Biol. 157, 135174 (1992)
F. Liechti, Birds: blowinby the wind? J. Ornithol. 147, 202211 (2006)
R. Nebuloni, C. Capsoni, V. Vigorita, Quantifying bird migration by a high-resolution weather
radar. IEEE J. Geosci. Remote Sens. 46, 18671875 (2008)
C.J. Pennycuick, The mechanics of bird migration. Ibis 111, 525556 (1969)
C.J. Pennycuick, Bird Flight Performance: A Practical Calculation Manual (Oxford University
Press, New York, 1989)
C.J. Pennycuick, Modelling the Flying Bird (Academic Press, Amsterdam, 2008), p. 480
C.J. Pennycuick, P.F. Battley, Burning the engine: a time-marching computation of fat and
protein consumption in a 5420 km nonstop ight by great knots, Calidris tenuirostris. OIKOS
103, 323332 (2003)
J. Perrin, Winged Migration (Chronicle Books, California, 2003), p. 272
6 Concluding Remarks 45
D.P. Raymer, Aircraft Design: A Conceptual Approach, 4th edn. (AIAA Press, Washington,
2006), p. 838
Rozell N. Bar-tailed godwit goes the distance. Article #1876, Alaska Science Forum, 10 Oct 2007
W. Shyy, M. Berg, D. Ljungqvist, Glapping ight and exible wings for biological and micro air
vehicles. Prog. Aerosp. Sci. 35, 455505 (1999)
R.J. Templin, The spectrum of animal ight: insects to pterosaurs. Prog. Aerosp. Sci. 36, 393436
(2000)
H. Tennekes, The Simple Science of Flight (MIT Press, London, 2009), p. 201
46 Migration on Wings