analysis of ice halo appearances using an all-sky...
TRANSCRIPT
Analysis of Ice Halo Appearances Using an All-Sky Camera
Sylke Boyd, Michelle King, and Morton Greenslit
University of Minnesota – Morris, Division of Science and Mathematics
Abstract Assessment of cirrus cloud coverage belongs to one of the more difficult problems in data collection. Spatial as well as temporal resolution is limited due to their optical properties, the restrictions of terrestrial observations and the evasiveness of cirrus in satellite images. Cirrus clouds often cause ice halos, in particular 22-degree halos. Other halo features, such as parhelia and various other subspecies of complex halo displays can be seen as well. Collecting data on these halos will, as a longitudinal set, allow assessment about the frequency, type and diurnal distribution of cirrus clouds. Data on complex halo features allow to make inferences about the types of ice crystals, which (in the future) may lead to an assessment of their growth conditions. We present our results on the observation of cirrus coverage over the first half of 2015, using an all-sky camera and our own image analysis software for the detection of ice halos. We record several thousand images per day. These images are then analyzed for the presence of ice halos, which allows us to compile statistics of their appearances, duration, intensity and correlation with other weather specifics. We present a halo detection algorithm, which was developed by iterative testing on large sets of images under varying sky conditions. Our goal is, to develop this combined all-sky camera/software system to a point at which it becomes portable to other locations, such as schools. This may allow a spatial resolution on the appearance of ice halos and their implications, in addition to the time resolution.
Acknowledgements The authors wish to extend their gratitude to UMM alumni Stephen Sorenson, Shelby Richard and James Froberg, who laid the ground work for the software development. This work is supported by the UROP program of the University of Minnesota, as well as a grant to the University of Minnesota, Morris from the Howard Hughes Medical Institute through the Precollege and Undergraduate Science Education Program. The acquisition and installation of the camera was supported by the UMM physics discipline, the division for Science and Mathematics and a Faculty Research Enhancement Grant of the University of Minnesota.
• Orion Starshoot Allsky camera with auto iris
• On roof of science building since 7-21-2014
• Saves still frames every 30 s (chosen)
• Resolution 480 by 720 • Location Lat 45.589052 Long -95.902858
Roof
Machine room
5th floor
4th floor lab
Allsky 1
(GMc)
Allsky 2
(SB)
Mounting arm, facing south,
A bundle of 3 cables, Guided into 5th floor machine room
video
wireless transmitter
RS232
Wireless transmitter
3 power plugs
Allsky control panel
Video wireless receiver
RS232
wireless receiver
Data Collection and
Internet streaming
Camera details
Bates College, Lewiston, Maine, July 2015
Halo identification program Dubbed haloloop, the program must process large numbers of images in a reasonable amount of time, about 80 000 for each observation month. Images are saved and can be processed with varying goals. The current version of haloloop: • Is written in C++, using openCV • Removes image distortions due to fisheye projection • Identifies the sun (or moon) position and radius • Analyses the radial intensity distribution • Several markers of the image are included and assigned
probability factors as observed for halo images. • The factors are combined into a final halo probability for each
image.
Original image Distortion removed, mask applied and sun identified
2maskr
1maskr
Radial intensity analysis
Analysis area
Outline of
camera view
Sun position
y = -1.28x + 444.28
100
150
200
250
300
350
400
40 50 60 70 80 90 100 110
colo
r va
lue
pixel distance from sun
Radial Intensity
-100
-80
-60
-40
-20
0
20
40
60
80
100
40 50 60 70 80 90 100 110
rela
tive
co
lor
valu
e
pixel distance from sun
Adjusted Radial total Intensity
y = -2.25x + 506.57
100
150
200
250
300
350
400
40 50 60 70 80 90 100 110
colo
r va
lue
pixel distance from sun
Radial Intensity
-100
-80
-60
-40
-20
0
20
40
60
80
100
40 50 60 70 80 90 100 110
colo
r va
lue
pixel distance from sun
Radial Intensity
The radial color intensities and their standard deviations are analyzed within the gray area , see left. This area is also cleared of bias by subtracting an average plane before analysis. The total radial intensity is used as
2 2 2r r r rI B G R
The adjusted radial total intensity is the result of subtracting the best fit line from the total radial intensity. This allows better contrast of the intensity fluctuations associated with a halo presence. The false-color images of the analysis area show the distribution of the adjusted intensity (red positive, blue negative).
Halo Finding Algorithm The halo probability of an image is a composite of probability factors which assess certain qualities of an image, based on the analysis area. There are seven such factors, comprised of three groups.
PHalo = Pblue PMVA PMVS Pslope PR2 PMPS PMPC PPMMP
Color intensities (Pblue, PMVA, PMVS) Pblue is a Gaussian centered at 0.61 for
b = B2/(RG) This excludes clear-sky images.
0
0.2
0.4
0.6
0.8
1
1.2
0
2
4
6
8
10
12
14
16
18
20
0.1 0.3 0.5 0.7 0.9 1.1 1.3 1.5
pro
bab
ility
fac
tor
per
cen
t o
ccu
ran
ce (
Mar
ch2
01
5)
bluecheck value b
other% halo% p
PMVA and PMVS assess variability of the radial total intensity (error bars in figures above) via its average (MVA) and standard deviation (MVS). This reduces fail signals from Ac and Sc skies.
Line fit to the radial total intensity (PSLOPE, PR2) Pslope assesses the slope of the best-fit line. This reduces signals from overcast days (As, St) as well as very clear days. PR2 is simply the R2 value of the fit. Low fit quality correlates with variable skies, such as Ac or Sc.
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
0
5
10
15
20
25
-5 -4 -3 -2 -1 0 1
pro
bab
ility
fac
tor
per
cen
tage
Mar
ch 2
01
5
radial best-fit slope
halo%
other%
p(slope)
Markers for the adjusted radial intensity (PMPS, PMPC, PMPPM)
Ad
just
ed r
adia
l In
ten
sity
r
MPC
MPPM
# = MPS
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
0
2
4
6
8
10
12
14
70 75 80 85 90 95 100
pro
bab
ility
fac
tor
per
cen
tage
Mar
ch 2
01
5
high-impact mp crossing MPC
halo%
other%
p(MPC)
12
/31
1/1
1/2
1/3
1/4
1/5
1/6
1/7
1/8
1/9
1/1
0
1/1
1
1/1
2
1/1
3
1/1
4
1/1
5
1/1
6
1/1
7
1/1
8
1/1
9
1/2
0
1/2
1
1/2
2
1/2
3
1/2
4
1/2
5
1/2
6
1/2
7
1/2
8
1/2
9
1/3
0
1/3
1
2/1
January 2015
No data
1A
1C
1B
1E
1F
1D
1G
No data
1H
1I
1J
1K
1L 1M
1N
1O
2/1
2/2
2/3
2/4
2/5
2/6
2/7
2/8
2/9
2/1
0
2/1
1
2/1
2
2/1
3
2/1
4
2/1
5
2/1
6
2/1
7
2/1
8
2/1
9
2/2
0
2/2
1
2/2
2
2/2
3
2/2
4
2/2
5
2/2
6
2/2
7
2/2
8
3/1
February 2015 2A
2D
2C
2B
2E 2F
2G 2H
2I 2K
2L
2M
2N
2O
2P
2J
2Q
2R
2S
2T 2U
2V
2X
2Y
2Z
2ZA
2ZB
2ZC
3/1
3/2
3/3
3/4
3/5
3/6
3/7
3/8
3/9
3/1
0
3/1
1
3/1
2
3/1
3
3/1
4
3/1
5
3/1
6
3/1
7
3/1
8
3/1
9
3/2
0
3/2
1
3/2
2
3/2
3
3/2
4
3/2
5
3/2
6
3/2
7
3/2
8
3/2
9
3/3
0
3/3
1
4/1
March 2015
3A
3D
3C
3B 3E
3F 3G
3H 3I
3K
3L
3M 3N
3O
3P
3J
3Q
3R 3S
3T
3U
3V 3X
4/1
4/2
4/3
4/4
4/5
4/6
4/7
4/8
4/9
4/1
0
4/1
1
4/1
2
4/1
3
4/1
4
4/1
5
4/1
6
4/1
7
4/1
8
4/1
9
4/2
0
4/2
1
4/2
2
4/2
3
4/2
4
4/2
5
4/2
6
4/2
7
4/2
8
4/2
9
4/3
0
5/1
April 2015
4A
4B
4C
4D
4E
4F
4G
4H
4I
4J
4K
4L
4M
4N 4O
4P
4Q
5/1
5/2
5/3
5/4
5/5
5/6
5/7
5/8
5/9
5/1
0
5/1
1
5/1
2
5/1
3
5/1
4
5/1
5
5/1
6
5/1
7
5/1
8
5/1
9
5/2
0
5/2
1
5/2
2
5/2
3
5/2
4
5/2
5
5/2
6
5/2
7
5/2
8
5/2
9
5/3
0
5/3
1
6/1
May 2015
5A
5B 5C
5D
5E
5F 5G
5H 5I 5J
5K 5L
5M
5N
5O
5P 5Q
5R
6/1
6/2
6/3
6/4
6/5
6/6
6/7
6/8
6/9
6/1
0
6/1
1
6/1
2
6/1
3
6/1
4
6/1
5
6/1
6
6/1
7
6/1
8
6/1
9
6/2
0
part of June 2015
6A
6B
6C
6D
6E
6F
6G
6H 6I
6J
6K
6L
6M
6S
6O
6P
6Q
No data (yet)
6R
Halo Observations Shown is log(Phalo) versus time for the first several months of 2015. The data have been Gaussian –broadened to enhance the influence of sequential halo signals.
Solved Problems: • Images from a commercial all-sky camera are being continuously recorded every 30
seconds. • A program named haloloop has been developed to search for a halo signal in
individual images. Currently, the focus is on the 22-degree ice halo. • The program detects the sun, removes the lens distortion and any intensity bias in
the analysis area. • The halo algorithm has been iteratively tested on large numbers of images and
refined to maximize fraction of recognized halos and minimize the false halo signals.
• The program was tested on the first several months of 2015. • The tests demonstrate that halos are a frequent enough phenomenon to consider
them as a factor in cirrus observations.
Unsolved Problems and challenges: • There are still halo signals for some images that do not actually contain a halo. In
particular, altocumulus skies are difficult, as well as the occasional cloud constellation that triggers a halo signal.
• There also are halo images that do not trigger a halo probability, in particular during transitions between Cs and As, while the cloud layer is thickening and the halo begins to fade.
• A good low-cloud subtraction mechanism needs to be implemented. • Camera downtime due to water, computer storage problems, and focus drift
need to be addressed. • A higher-resolution camera, lower horizon and perhaps a conventional-camera
array would be desirable.
Outlook and questions • How frequent are halos, really?
• How can they be used to assess temporal and spatial cirrus coverage?
• Can halo observations be used to assess the characteristics of cirrus clouds?
• Will long-term observations allow to observe changes in cirrus coverage?
• With better camera arrangements, will it be feasible to assess detailed information on crystal distributions, optical thickness, time evolution?
• Can this information be used to make inferences about the growth conditions at cirrus altitude?
We are also working on a Lidar project to measure the clouds in the third
dimension.