1
1
2
Compact and Inexpensive Kite Apparatus for Geomorphological 3
Field Aerial Photography, with some Remarks on Operations 4
5
6
Ralph D. Lorenz* 7
Space Department, Johns Hopkins University Applied Physics Laboratory, 11100 Johns Hopkins Road, 8
Laurel, MD 20723, USA 9
10
11
Submitted to GeoResJ, 1 October 2013 12
13
* Corresponding author tel. +1 443 778 2903 fax +1 443 778 8939 email:[email protected] 14
15
16
2
17
Abstract 18
Equipment for performing low-altitude aerial photography for geomorphological studies on 10-100m 19
scales is described, with particular reference to study of sand dunes. An automatic digital camera is 20
lofted by a parafoil kite : the arrangement costs around $300, collapses into a volume of ~2 litres and 21
can be deployed in a few minutes, making it convenient for field use when wind conditions (>4 m/s) 22
permit. Some operational considerations are discussed. 23
24
25
3
26
1. Introduction 27
An aerial perspective is often useful in geomorphological studies, either in its own right or as a bridge 28
between the remote sensing view from orbit and that afforded from the ground. Beyond flying an 29
observer in an aircraft in the classic mode of aerial photography , an array of platforms exist for so-called 30
small-format aerial photography (reviewed in detail by Abel et al. 2010) wherein an imaging capability is 31
deployed from the ground on-site. 32
Such approaches are facilitated by the ever-growing capabilities of lightweight digital cameras. Other 33
technologies include solid-state gyroscopes that facilitate flight of slow-moving radio controlled vehicles 34
such as quadcopters, and GPS guidance which allows autonomous survey flights by small airplanes. 35
However, these techniques (and the use of balloons, which typically requires the conveyance of heavy 36
helium cylinders to the field) all entail a certain amount of expense and complexity. 37
Kite aerial photography (KAP) has a rich history, notably from the dramatic view of the destruction of 38
San Francisco by the 1906 earthquake. The field has experienced a renaissance with the availability of 39
lightweight digital cameras, and a web search will reveal many examples. A wide range of systems can 40
be used, from large kites suspending radio-controlled steerable platforms with video viewfinders used 41
in an interactive manner, to simple timelapse devices. This paper focuses on the latter type, with 42
specific application to the geomorphological study of sand dunes, for which kite images were used in a 43
recent book (Lorenz and Zimbelman, 2014). Since sand dunes tend to be found in windy locations, 44
kiteborne photography is particularly well-suited to this application, and modern parafoil kites (figure 45
1) can be readily packed into a small rucksack for convenient deployment in the field. More particularly, 46
the relatively low cost, mass and volume, and quick deployment and retrieval, means that aerial 47
photography can be attempted with little effort, and can be considered an opportunistic augmentation 48
to other field activities, rather than needing to be a premeditated enterprise. 49
50
51
4
52
Figure 1. Experiments with an early kite camera variant - a homebuilt timer driving a keychain camera, 53
at White Sands National Monument, New Mexico, USA. Author is at left, with Jim Zimbelman at right. 54
Note the curved upper surface of the parafoil kite. Photo courtesy of Jason W. Barnes. 55
56
2. Apparatus 57
5
Some initial experiments were carried out with a small parafoil kite and a homebuilt timelapse camera 58
using a simple timing circuit ‘hotwired’ via an optocoupler across the shutter switch of a small keychain 59
digital camera (see Lorenz et al., 2010) attached to a bar or small board suspended a couple of meters 60
below the kite bridle attachment point. Reiche et al. (2012) describe a similar arrangement for 61
suspending meteorological instrumentation from a Parafoil kite. Although a few good results have been 62
obtained (figure 2), this approach was unsatisfactory in a number of ways, most particularly because the 63
majority of frames were motion-smeared. This resulted primarily from the slow readout of the cheap 64
detector in the ‘keychain’ cameras used (such small cameras lack a mechanical shutter), but also 65
because initial experiments tended to attach the camera too close to the kite, leading to higher angular 66
rates. Even taking hundreds of pictures in a flight, not infrequently it was found that no images were 67
useable. 68
69
70
Figure 2. Example results with a ‘keychain’ digital camera. (A) Dumont Dunes, in California. This image 71
looks towards the sun (which has saturated the upper part of the image) and so has strong contrasts. 72
6
(B) Barchanoid dunes on the Chott El Gharsa, Tunisia. Contrasts in this noontime image are rather 73
muted, but the presence of personnel and vehicles adds thematic interest to the image. (C) a linear 74
dune in Quattaniya, west of Cairo, Egypt. The image shows the wavy (seif) crestline of the dune, and the 75
fact that the eastern (right) plinth on which the main dune sits is narrower than at the left. Note that the 76
presence of the horizon in the image facilitates interpretation. (D) a downlooking image from relatively 77
low altitude at Kelso, California. Note the smearing in the lower part of the image, due to violent kite 78
motion. This smear is a particular problem with (typically inexpensive) cameras where the longer 79
exposure times increase the chance of scene motion. 80
81
A dramatic improvement is obtained with a more modern camera, optimized for documenting fast-82
moving sports activities such as surfing and mountain-biking. Such cameras, the example being used 83
here is the HERO GoPro2, have optics and detectors that permit short exposure times and thus much 84
reduced susceptibility to blur. They are furthermore marketed with robust protective housings, and 85
convenient controls to set image acquisition at regular intervals. The image format is large enough (e.g. 86
2840x2880 pixels) that images can be cropped substantially to show regions of interest with acceptable 87
resolution. 88
The kite dynamics can also be improved with some care. In effect the camera suspension can be 89
considered as a pendulum. A vital consideration, learned unfortunately by experience, is that the 90
pendulum should have a long period to prevent coupling between the swinging camera and the kite 91
dynamics which can in turn reinforce the swinging. This is avoided by the simple precaution of attaching 92
the camera to the line about 10m away or more from the kite. This effectively lengthens the swing 93
period to a point where resonant interactions are reduced acceptably. 94
More elaborate KAP apparatus uses a Picavet suspension wherein a platform is suspended on 4 pulleys, 95
through which a cord passes to attach at two locations on the kite string. This suspension allows the 96
platform to maintain a constant horizontal orientation even when the suspension points move rapidly in 97
space. Such an arrangement is important to avoid rapid scene motion in video, but is generally 98
unnecessary for terrain photography and is somewhat prone to tangling. Simply passing a loop of cord 99
through four holes in the camera platform (for which the plastic display mount for the GOPRO was 100
used) is adequate to ensure a generally-horizontal orientation. The system used by the author has a 101
~1.5m loop of cord, such that the suspension forms 2 legs of about 40cm length. 102
7
103
Figure 3. Apparatus is compact. (A) the packed Parafoil-10 kite with a hoop of 100lb line (B). Gloves 104
(C) are essential to prevent abrasion injury to the hands. (D) is a small hoop of 300’ of 30lb line. (E) is 105
the GoPro2 camera used for the later images in this paper. (F) is a handheld anemometer, useful to gain 106
familiarity with the wind conditions. (G) is the GoPro camera housing, on a small plate suspended from a 107
loop of heavy string. (H) is a smaller Parafoil-5 kite. 108
109
8
110
The suspension cord is attached at two points onto the kite string - most conveniently after the kite has 111
been launched. Remarkably, the weight of the camera and the tension on the line is enough to hold the 112
system on the line even with only a couple of loops of the line around a rigid element such as a 113
carabiner or swivel hook. These elements allow the rapid attachment and removal of the suspension 114
from the line. 115
The distance between the suspension points determines the azimuthal stiffness of the mount. If a rigid 116
mounting is desired (such that the camera platform always points downwind or upwind, along the kite 117
string direction) then the mounts should be far apart (here, about 60cm), with the suspension forming a 118
right angle. Alternatively, mounting the suspension points close together allows the suspension to 119
swing (or even wrap around) which may be desirable for a wider range of azimuths. 120
121
Figure 4. Suspension arrangement 122
123
124
125
9
3. Operating Considerations 126
General safety considerations pertaining to kites should be followed - e.g. do not fly near power lines, 127
roads etc. and pay attention to applicable regulations (see Aber et al., 2010.) Beyond the camera, kite 128
and line, a few accessories have been found to be useful. First, protective (e.g. leather) gloves are all 129
but essential, as severe abrasion of the hands will occur when winding the kite back in. With suitably 130
close-fitting gloves, the camera can be operated and attached without their removal. A handheld 131
anemometer may be useful in assessing wind conditions, although with practice the wind pressure felt 132
on the face or wind noise can become an adequate guide. 133
The GoPro cameras have a variety of operating modes, from individual stills to high-speed movies. Most 134
useful in this application is the timelapse setting, where still images at a cadence of under a second to 135
minutes can be automatically acquired. The best cadence is a compromise between acquiring enough 136
images to ensure the available views are obtained, while not oversampling (i.e. obtaining multiple near-137
identical images, which will fill the memory card on the camera more rapidly and take longer to inspect). 138
With the arrangement described here in somewhat steady conditions, a five-second cadence is usually 139
more than adequate. If the kite is undergoing rapid motions in turbulent conditions, faster sampling 140
might be preferred. 141
The GoPro camera used has a very wide field of view, which leads to fisheye distortion. When the 142
camera views horizontally, with the horizon in the center of the frame, this distortion is not severe 143
144
10
145
Figure 5. A chain of reversing barchans in the eastern United Arab Emirates. The distortion caused by 146
the wide field of view is evident, but the utility of the image in showing the morphology is evident. Note 147
that at this location there is good contrast between the gravel plain and the sand. 148
149
Although post-processing to enhance contrast can improve matters, images in the middle of the day can 150
often be disappointing. However, for high-contrast scenes, such as red barchan dunes on a mud pan 151
(figure 5), good images can be obtained even with high sun. 152
Much more generally appealing results can be obtained within a couple of hours of sunset, when 153
topographic shading becomes much more prominent (figure 6). Images taken facing somewhat 154
towards the sun can be particularly dramatic (e.g. figure 7). 155
11
156
Figure 6. Barchans at the Salton Sea, in southern California, USA, in the late afternoon with winds from 157
the west. (A) with the camera mounted down-sun, the view is to the East, and shows the Salton Sea in 158
the background. Although the low sun generates a shadow behind the slip face, the overall image 159
contrast is quite poor. The author, holding the kite, is just upwind of the barchan slipface (B) Ten 160
minutes later, with the camera mounted to view towards the sun. The surface contrasts are much more 161
dramatic and show the morphology much better. The kite with its two stabilizing streamers is visible at 162
12
the top. Also noticeable are patches of light due to sand grains on the lens cover - viewed down-sun 163
these would not be apparent, but at this phase angle, they scatter light strongly. 164
165
166
167
168
Figure 7. Image from about 200m altitude (1000’ line fully paid-out) at White Sands National 169
Monument, New Mexico, USA. The low sun dramatically illuminates the barchanoid ridges, and 170
highlights the normally rather subtle elemental duneforms on the stoss slopes. At center left there is a 171
vehicle for scale - a flat area bounded by a narrow berm is bulldozed as a road and parking area. 172
173
174
13
The most efficient procedure for launching the kite, which can be accomplished easily by a single 175
operator in the field, is as follows. Lay out the kite string, and the camera mount, on the ground, 176
keeping the kite itself in one’s rucksack. Turn on the camera, adjust its pointing as desired and verify 177
that it is taking pictures every few seconds as desired. Then remove the kite and unfold, attaching any 178
tail or streamers (in fact, it has been found that a streamer tail can be efficiently left attached and 179
wound around the kite envelope after flight) and attach to the swivel on the kite string - the kite can be 180
kept from blowing away by kneeling on it if necessary. Put on gloves and launch the kite, then allow the 181
line to pay out by 10m or more. Then the camera suspension can be attached, and the kite then paid 182
out to its operating altitude (usually as high as the string will allow, subject also to time constraints). 183
Releasing the kite can be rather quick in strong winds, but with a hoop it is practical only to winch in the 184
kite at about 0.3m per second. Thus a deployment to the full altitude permitted by a 300m line may 185
take 10-15 minutes to wind back in. Parafoil kites tend to fly at somewhat shallower angles than some 186
rigid kites, and it will be observed that often (especially in marginal conditions) releasing more kite string 187
does not lead to a substantial increase in altitude, but the kite simply flies further downwind. This is 188
because the weight of (and more importantly, the aerodynamic drag on) the kite string progressively 189
loads the system. Indeed, the fact that a kite string forms a curved (catenary) shape was documented 190
very early (Marvin, 1897). At the point of suspension, the angle of the kite string with respect to the 191
horizontal is determined by the tangent of (lift minus weight of the kite) divided by the kite drag, 192
whereas the angle of the string at the operator is defined by the tangent of (lift minus weight of the kite, 193
minus weight of the kite string) divided by the drag of the kite and the string. The kite string defines a 194
smooth curve linking these asymptotes. 195
It will be observed that when applying tension to the kite string (as when winding in) the kite tends to fly 196
higher, i.e. the kite string becomes more vertical (see figure 8). This effect can be exploited to obtain a 197
different set of viewing angles, or by periodic wind and release, to gain altitude. In fact, this 198
phenomenon deserves closer study from a purely dynamical perspective, in that simple models of a kite 199
system can be bi-stable (e.g. Adomaitis, 1989), and real-world flight data may elucidate this bifurcation. 200
It presumably results from the curves of lift, drag and moment coefficients as a function of angle of 201
attack, and the moment applied by the kite suspension bridle, such that two moment equilibria exist. 202
One of these is at a higher lift/drag ratio (L/D) and thus steeper kite string, and corresponds to a higher 203
lift and a higher line tension. 204
205
14
206
Figure 8. Paying out more line after launching a kite in weak winds typically results in it flying further 207
downwind with only modest increase in altitude, with the kite string being near-horizontal at the 208
operator. Winding the string back in, applying higher line tension, often makes the kite fly at a steeper 209
angle, with higher lift and lift/drag ratio. This effect can be exploited to derive a range of positions for 210
photography from a single operator location. 211
212
213
It is tempting to launch the kite from the crest of a dune under study. Such a location is clear of 214
obstructions, and has the aesthetic advantage of placing the observer in the center of the scene under 215
study. However, this temptation should be avoided for two reasons. First, the observer may be 216
deceived into launching into marginal conditions - the streamline compression near the brink of the 217
dune means windspeed is at a local maximum, and thus in marginal winds the kite will fly adequately 218
there, as soon as it attempts to gain height it enters slower-moving air and fails to climb. The operator 219
may spend some time in such an environment hoping in vain for the situation to improve. The other 220
difficulty is that flow separation at the brink can often occur (leading, when saltation is active, to the 221
15
classic ‘smoking dune’ behavior). This causes a recirculating flow region (see figure 9) in the lee of the 222
dune, with a vortex bound to the slip face. This region , which may extend some way above the crest, is 223
characterized by weak and often highly turbulent winds, in which the kite may fly very erratically, or (as 224
the author found at a large dune in Sossusvlei, Nambia) it may lose lift entirely and fall to the ground. 225
What the operator can get away with is best learned by experience, but a rule of thumb may be that the 226
recirculating region extends from the brink at about 45 degrees downwind, up to about one dune height 227
above the crest. The airflow may be disturbed, however, rather further downwind than that (a rule of 228
thumb in the kite literature is that a wake behind a building extends about 7 times the height of the 229
building, although this is a particularly sharp-edged obstacle). It may be easiest to explore these regions 230
(and in fact kites may be a useful means of quantitatively mapping the airflow around a dune - e.g. 231
Walesby and Harrison (2010) show that the kite string tension measured by a strain gauge can be a 232
useful proxy of windspeed) by launching the kite upwind of the dune and marching around, than by 233
attempting to launch the kite at the crest itself. 234
235
236
16
Figure 9. Inexperienced operators should avoid launching at dune crests - winds may decrease aloft and 237
downwind leading to ‘trapped’ flight or loss of lift as the line is paid out, and/or erratic flight in the 238
vortex region. Launch upwind of the dune is recommended, where wind may be slightly weaker, but is 239
steadier and does not decrease with height or distance downwind. 240
241
242
243
4. Interpretation 244
It may simply be desired to obtain oblique aerial views for illustration of papers or talks, and for this 245
application the apparatus described works very well. Repeat visits may permit change detection (e.g. 246
due to dune migration) although the importance of making such observations in illumination conditions 247
as similarly as possible is stressed. 248
Quantitative photogrammetric interpretation can be complicated by the oblique view and camera 249
distortion - as discussed at length in Aber and Aber (2010), these can be calibrated with an optical model 250
and/or by using markers with known sizes and/or locations. Tools like Google Earth may be used to 251
compare satellite imaging with aerial views. 252
An intriguing possibility is that digital elevation models can be constructed from sets of images acquired 253
from a range of angles - online tools such as 123DCatch can perform automatically the correlation and 254
registration of such images. Such stereogrammetry requires a diversity of viewing angles, and so a 255
combination of walking the kite around with winding the line in and out may be necessary to have the 256
kite move the viewpoint to a wide enough range of locations. For this application especially, adjusting 257
the view azimuth relative to the wind (or allowing it to change, e.g. by narrow suspension points) may 258
be necessary. 259
260
5. Conclusions 261
A compact and relatively inexpensive setup has been described that can provide aerial views up to 262
altitudes of ~200m suitable for geomorphological studies. Some considerations for safe and efficient 263
operations have been discussed. 264
265
266
267
268
17
269
6. Acknowledgements 270
Field visits to some sites where results are reported in this paper were supported by NASA grant 271
NNX13AH14G ‘Cassini Radar Science Support’. Colleagues Joseph Spitale, Jani Radebaugh, Jason 272
Barnes and Brian Jackson are thanked for assistance in the field. NHK TV of Japan is thanked for the 273
donation of a battened foil kite and picavet suspension, and for alerting the author to the GoPro camera 274
range. 275
276
277
18
278
Appendix - Apparatus Details 279
280
The kite apparatus used here was obtained from Into the Wind (www.intothewind.com) and their part 281
numbers are quoted for convenience, although of course products and specifications (the Imperial units 282
used in their catalog are repeated here) can change and other vendors may offer comparable products. 283
Best results have been obtained with a ‘Parafoil 10’ kite (Product #5105) which inflates to 2’9”x3’7” and 284
will fly with the camera in winds of ~6mph or so. With strong winds this kite can develop a strong pull, 285
and so a 100lb line is recommended (e.g. 1000’ of Dacron line #138 wound onto an an 8” hoop spool 286
#1280).For attachment of the camera suspension lines, small swivel clips (e.g. #4032) or carabiners 287
(#6551) can be used. 288
If one is confident of encountering reasonably strong winds a smaller kite (e.g. ‘Parafoil 5’ #5752 289
inflating to 1’11”x2’7” with a 50lb line on a 6” spool #1265) will be slightly easier to handle and is more 290
compact. 291
The consumer electronics industry is of course fast-moving, with products rapidly being retired in favour 292
of new items, and the GoPro2 cameras used in this paper are no longer offered for sale by the 293
manufacturer directly (although aftermarket vendors may retain some stock). GoPro3 cameras likely 294
offer comparable or superior performance, and the kite suspension described in this paper can likely be 295
constructed similarly, or using one of the manufacturers helmet or surfboard mountings, or with a 296
simple bolt and the tripod mount. An alternative might be to thread a small rope through a PVC tube or 297
similar element, perhaps 20-30cm long, and attach the camera using one of the handlebar mounts sold 298
for mountain biking video. 299
300
301
302
303
19
304
References 305
Aber, J. S., S. W. Aber and F. Paveri, 2002. Unmanned Small-Format Aerial Photography from Kites for 306
Acquiring Large-Sclae, High-Resolution, Multiview-Angle Imagery, FIEOS 2002 Conference Proceedings 307
Aber, J., I. Marzolff and J. Ries, 2010. Small-Format Aerial Photography; Principles, Techniques and 308
Geoscience Applications, Elsevier. 309
Adomaitis, R. A. 1989. Kites and Bifurcation Theory, SIAM Review, 31, 478-483 310
Lorenz, R. D., B. Jackson and J. Barnes, Inexpensive Timelapse Digital Cameras for Studying Transient 311
Meteorological Phenomena : Dust Devils and Playa Flooding, Journal of Atmospheric and Oceanic 312
Technology, 27, 246-256, 2010 313
Lorenz, R. D. and J. Zimbelman. Dune Worlds : How Windblown Sand Shapes Planetary Landscapes, 314
Springer, 2014 315
Marvin, C. F. 1897. A monograph on the Mechanics and Equilibrium of Kites, Weather Bureau W.B. 316
No.122, Department of Agriculture, Washington DC. 317
Reiche, M., R. Funk, Z. Zhang, C. Hoffmann, Y. Li and M. Sommer, 2012. Using a Parafoil Kite for 318
Measurement of Variations in Particulate Matter - A Kite-Based Dust Profiling Approach, Atmospheric 319
and Climate Sciences, 2, 41-51, 2012 320
Walesby, K. T. and R. G. Harrison, 2010. Note: A thermally stable tension meter for atmospheric 321
soundings using kites, Review of Scientific Instruments 81, 076104 322