03202014(cirrus no cloud_somecloud_highcloud)-4
DESCRIPTION
I took great efforts in the short time available to me today to understand what the histograms of Cirrus Band 9 images are telling me; by identifying the specific parts of the scene and looking for patterns in the histograms. I took a look at three scenes, one fully covered with clouds one partially covered and one clear scene. Sincerely, -- Jordan D. Ulmer South Dakota State University Image Processing Lab http://iplab2out.sdstate.edu/TRANSCRIPT
CirrusThree Stages of Cloud Cover
Landsat 8, OLIBrookings ROI (P029/R029)
Jordan D. Ulmer20th March, 2014
Focus And Intent
• Present Purpose: • To characterize the Cirrus band, and evaluate
its effective cloud detection abilities relative to TIRS, in combination with RGB.
• Future Intent: • To understand the uses and limitations of
Cirrus. • To improve upon the current techniques and
develop more veracious and versatile algorithms for cloud detection and correction.
2013 Imagery
No Cloud Cover
Landsat 8, OLI – Brookings ROI (P029/R029)(289)No clouds
October16 2013 17:13 UTC RGB
7564;8990;
DNStretch
This is a clear scene RED
(Band#2)
Histogram
Landsat 8, OLI – Brookings ROI (P029/R029)(289)No clouds
October16 2013 17:13 UTC Cirrus
4935;5124;
DNStretch
This is all
sensor NOISE
Not cloudy
Histogram
Count(DN)
Landsat 8, OLI – Brookings ROI (P029/R029)(177)A Good Example of What RGB Can Do too…
June 26 2013 17:13 UTC RGB Bands(4,3,2)
This peak is the dark current of the Landsat 8 detector array at this specific time and over brookings
This peak is present in all the histograms, is it also a property of the instrument?
2013 Imagery
Some Cloud Cover
Landsat 8, OLI – Brookings ROI (P029/R029)(177)A Good Example of What Cirrus Can Do
June 26 2013 17:13 UTC Cirrus Band-9
Dave Says:1. 27-Jun-13
1. Formosat
2. 17:13:31
3. ? (Good from Sat view)
2. 28-Jun-131. Formo
sat2. 17:13:
573. Fair+
3. 28-Jun-131. QB2. 16:37:
113. Fair+
Cirrus Says:
1. Splattered with dense clouds high above the earth’s surface
Landsat 8, OLI – Brookings ROI (P029/R029)(177)A Good Example of What RGB Can Do too…
June 26 2013 17:13 UTC RGB Bands(4,3,2)
Dave Says:1. 27-Jun-13
1. Formosat
2. 17:13:31
3. ? (Good from Sat view)
2. 28-Jun-131. Formo
sat2. 17:13:
573. Fair+
3. 28-Jun-131. QB2. 16:37:
113. Fair+
RGB Says:1. Same
as Cirrus with more cumulous clouds
Landsat 8, OLI – Brookings ROI (P029/R029)(177)A Good Example of What Cirrus Can Do
June 26 2013 17:13 UTC Cirrus
4686;8646;
DNStretch
Landsat 8, OLI – Brookings ROI (P029/R029)(177)A Good Example of What RGB Can Do too…
June 26 2013 17:13 UTC RGB Bands(4,3,2)
6713;13700;
DNStretch
Partly cloudy
Histogram
Count(DN)
Landsat 8, OLI – Brookings ROI (P029/R029)(177)A Good Example of What RGB Can Do too…
June 26 2013 17:13 UTC RGB Bands(4,3,2)
The dark current has a
wider spread and extends further
below 5000 DN
This structure seems to repeat;
is this due to periodic resonant of water vapor
(ie: is reflected)?
Real Clouds
2014 Imagery
Full Cloud Cover
Landsat 8, OLI – Brookings ROI (P029/R029)(068-14)Whispy Clouds
March 9 2013 17:12 UTC RGB Bands(4,3,2)
3813;8241;
DNStretch
Landsat 8, OLI – Brookings ROI (P029/R029)(068-14)Whispy Clouds
March 9 2013 17:12 UTC RGB Bands(4,3,2)
Fully cloudy
Histogram
Count(DN)
Different domain, but similar peaks as the partly cloudy scene ~ 6000 DN – 9000 DN
The count of the noise peak increased
There whole scene is cloudy so there are a
greater number of higher ~8000
DN counts
This DN Signiture must be due to
the NON data filler
pixels