The Trent Farm Photos Appendix
"On the Possibility that the McMinnville Photos
Show a Distant Unidentified Object (UO)"
[NOTE: Click on any of the figure names to view the figure]
(This was written in 1976-1977 - exact date not recorded - but
was not included in the previous publication. It is published
here for the first time. There have been some clarifying
comments added in April, 2000.)
This appendix is provided to supply certain supplemental information
that will prove useful in evaluating the analysis presented in
the main text, in particular the analysis related to the determination
of the amount and effects of veiling glare. The information is
provided in a series of figures, each of which is described below
. Further information is available from the author.
In the main text the relative brightness of a vertical , white
shaded surface was estimated from the image brightness of the
shadow on the distant house wall. There has been some question
an to whether or not the wall was "truly" white. Therefore
I have made another estimate based upon the image brightness
of the nearby (Trent) house that appears at the right hand side
of photo 2. This house was (according to Mrs. Trent in 1975)
painted white only about a year before the pictures were taken.
An image of the corner of the house just below the eave
appears in the second UO photo at the right hand side. (The
corresponding image in the first UO photo was cut off the original
negative sometime after publication in the Telephone Register
newspaper, which shows the corner of the house in both photos.)
Figure A1 illustrates the calculation of the brightness
of a vertical white surface from the brightness of the image
of the western corner of the south wall of the nearby house.
Although the veiling glare correction is larger (because
it is immediately adjacent to the sky), there is no atmospheric
brightening correction. The brightness of a horizontal,
shaded white surface based on this nearby house image differs
only slightly from the value obtained using the image of the
distant house.
As pointed out in the main text, certain evidence suggests that
12% may be an upper bound on the glare index (defined as the
brightness of a perfectly intrinsically black UO image divided
by the adjacent sky brightness; see the text and see below).
The evidence for this is presented in Figures
A3,
A4 &
A5.
These figures contain data on the relative brightnesses of images
(garage roof, wall) which, if there were no veiling glare, would
have (approximately) constant intrinsic brightness (because
of constant reflectivity) over angular distances of at least
several degrees away from the object/sky boundary. Figure
A2 illustrates the variation of the brightness of the garage
roof in each picture. The angular distances are measured along
the scan directions indicated by the arrows. Figures
A3 and
A4
illustrate the brightness variation of the garage wall at the
level of the rafter ends, and Fig.
A5 illustrates the brightness
variation of the shadow that is just under the edge of the roof
and Just above the rafter ends. The variation in brightness
is mostly caused by veiling glare...light from the adjacent sky
is scattered by camera optics into the image of the darker roof
or wall.
************************************************************************
FIG. A1
THE SOUTHWEST CORNER OF THE WALL AND EAVE
OF THE NEARBY (TRENT) HOUSE APPEARS AT THE RIGHT HAND SIDE OF
PHOTO 2. THIS HOUSE WAS REPORTEDLY PAINTED WHITE WITHIN THE YEAR
BEFORE THE PICTURES WERE TAKEN.
SINCE THE SUN WAS SLIGHTLY NORTH OF WEST (OR IF IN THE
MORNING, SLIGHTLY NORTH OF DUE EAST AT AN ANGULAR ELEVATION OF
ABOUT 25 DEGREES), AND SINCE THE ROOF OF THE HOUSE HAD A SMALL
EAVE , THE SOUTH WALL WAS SHADED FROM THE DIRECT SUN.
IT WAS, HOWEVER, ILLUMINATED BY SKYLIGHT AND GROUND-REFLECTED
LIGHT. THUS THE INTRINSIC BRIGHTNESS OF THE WALL SHOULD
BE THE SAME (OR PERHAPS SLIGHTLY GREATER, SINCE THERE WAS
NO EAVE SHADING IT) AS THAT OF THE SHADED PART OF THE WALL OF
THE DISTANT WHITE HOUSE. FROM THE DENSITY MEASURMENTS
AND TRANSFER CURVE:
Ewall image = 0.021
(1/2) degree from the edge
Esky image = 0.065
adjacent to the wall
WHEN THE GLARE INDEX IS 12%, A DARK AREA NEXT TO
A UNIFORMLY BRIGHT AREA HAS A GLARE OF ABOUT 6% AT A DISTANCE
OF 0.5 degrees INTO THE DARK AREA IMAGE. THE SKY BRIGHTNESS
IS NOT UNIFORMLY BRIGHT. NEVERTHELESS, A GOOD APPROXIMATION
TO THE INTRINSIC RELATIVE BRIGHTNESS OF THE HOUSE WALL IS
Bvertical,white,shaded surface
= Ewall image - Gwall image
= Ewall image - g Bsky image
= 0.21 - (0.06) (0.06565)
= 0.0171
THIS VALUE IS SOMEWHAT LARGER, BUT IN GOOD AGREEMENT WITH
THE VALUE, 0.0014, CALCULATED FROM MEASUREMENT OF THE BRIGHTNESS
OF THE DISTANT HOUSE SHADOW AFTER CORRECTION FOR GLARE AND ATMOSPHERIC
EFFECTS.
USING THIS VALUE DIVIDED BY 2.4 IN THE RANGE
CALCULATION YIELDS 1.3 KM.
*************************************************************************
The image brightnesses illustrated in all these figures would
be roughly constant if there were no veiling glare. However,
since these images are adjacent to the image of the bright sky,
and since there was VG, the brightnesses increase with decreasing
angular distance to the image of 'the sky. Included in these
figures are image brightness variations predicted from laboratory
simulation data on the glare light distribution for various values
of glare index, guo, which is the glare in an ellipse that simulates
the UFO image in photo 1. Figures
A3 and
A4 show that the brightness
variation of the garage wall is more consistent with guo= 7%
than with guo = 12 % or 20 %. Figures
A6,
A7,
A8,
A9,
A10,
A11,
A12 &
A13
contain glare curves
obtained in laboratory simulations of the luminance distribution
in photos 1 and 2. Figures
A6
A7
A8
illustrate the glare brightness
variation along the image of a synthesized "telephone pole"
for various values of glare index. These pole simulation
curves were used to predict the brightness variation of the image
of the pole in photo 1. The predicted variations, illustrated
in Fig.
A16,
were obtained by fitting the laboratory glare curves
to the measured Image brightness at the horizon.using the formula
Eimage= Bintrinsic + Gimage = Bi + g(x,y) Bs , where
Bi, the intrinsic brightness of the object is an adjustable constant
(it is constant for a particular graph of brightness versus position
on the pole), Bs is the brightness of the sky about 10 degrees
above the horizon and g(x,y) is the glare distribution for a
given 'sky' luminance distribution and for a given image shape
and size as a function of x-y coordinates in the film plane.
If y represents angular displacement in the vertical direction,
then, along the vertical pole image, Epole image = Bpole + g(y)Bs.
The function g(y) for the three glare index values illustrated
was, obtained from Figures
A6,
A7 &
A8.
(NOTE: This formulation of the quantitative estimation
of veiling glare has a theoretical basis in the observed fact
that most of the glare in a image comes from the light sources
immediately adjacent to the image, such as from the sky within
a few degrees of the UO, for example. In other words,
the scattering which produces the glare tends to be a small angle
or "forward scattering" phenomenon. The more
grease there is on a lens the larger this scattering angle becomes.
For typical lenses experiments suggest the angle is a
few degrees.)
For example, for guo = 20 %, curve A shown in Fig.
A16 is given
by Eimage = 0.00151+ g(y)(0.06), where g(y) is the variation
of glare with height (y) in Figure A6 (i.e. the graph in Figure
A6) and 0.06 is the sky brightness above the pole. To obtain
curve 3 in Fig. A16,
I have calculated the expected image brightness
from Emage = 0.00259 + g(y )(O.06), where g(y) is
the variation along the simulated pole in Figure A7. Also
in Figure A16 is the expected brightness variation along the
pole image when the glare index is 7%. Clearly the best
fit to the data (dots) is for 12%.
As can be seen in Fig. A16, all the "theoretical" curves
fit the data at the horizon. However, none of them fit the data
below the horizon; the photo data indicate a very constant image
brightness below 1 degree below the horizon. Thus the data
below the horizon are consistent with a glare curve for a glare
index even lower than 7%. If the glare index were that low, the
increased brightness of the pole image above the horizon in the
photo would have to be explained as a combination of glare and
intrinsic brightness increase with height along the pole. I have
noticed that creosoted poles often become lighter colored near
the top as a result of weathering away of the creosote,
so it is possible that some of the increased brightness of the
pole image with altitude was due to an actual increase in brigtness
of the pole, in which case the glare index should be lower than
12%. Unfortunately there is now no way of measuring the
actual brightness variations, if any, of the telephone pole.
Figures
A8,
A9 &
A10
illustrate the laboratory measurements of VG in
a large dark image adjacent to a large bright area, which is
an approximate simulation of the garage roof. These curves
were used to Calculate the glare curves in Fig. A2.
Figures A11,
A12, and
A13 illustrate the glare variations obtained
in laboratory synthesis of the image of the garage when scanned
along the rafter ends. These curves labelled "B" in
the figures were used to calculate the glare curves illustrated
in Figures
A3,
A4 &
A5.
Figures A14 and
A15 illustrate the variation of VG with the angular
size of an image for various types of simple geometric images
silhouetted against a large, constant brightness field. The VG
increases as the image size shrinks, although for sizes much
smaller than 0.1 degree the VG is expected to remain nearly constant.
As an image increases in size the glare shrinks, but it
does not go to zero since some light is always scattered. The
glare curve can be roughly divided into"short range"and
"long range" regions. The short range
glare decreases rapidly with increasing image size. When a lens
is clean the short range glare is evident for images smaller
than a degree in angular extension (depending upon the shape
), as illustrated in Fig. A14. However, when a lens is very dirty
the"short range"glare may extend for many degrees,
as illustrated in Fig. A15. Also illustrated in Fig.
A15
is the observation that an increase in dirt or grease on the
lens does not substantially change the functional form of the
glare for angular sizes less than 1 degree. Note that the
effects of the"short range"glare are also evident in
the separation between curves A and B in Figures
A11,
A12 &
A13.
Figures A14 and
A15 also illustrate the previously mentioned
fact that the glare in an ellipse comparable to the image of
the bottom of the UO (22 degree aspect ellipse with a major axis
length of 1.6 degree) is about the same as in a 1 degree
disc.
Top of Page
© copyright B. Maccabee, 2000. All rights reserved.
|