Improved Algorithms for Combining Satellite Imagery
and Geographic Basemaps
Rick Kohrs
Space Science & Engineering Center
University of Wisconsin - Madison
2005 McIDAS Users’ Group Meeting
Madison, WI
…clouds continue to cover the
Northern half of the US and
Canada…
November
29
Current Satellite Image
o
10 F
In Reality …
Current Algorithm
¾ Infrared Brightness Temperature (Tir) based cloud/no-cloud threshold
¾ Basemap is shown when pixels are warmer than Tir threshold
¾ Temperature based transparency is applied to “cloudy” pixels, where
warm clouds are semi-transparent and cold clouds are opaque
Choosing a Brightness Temperature Threshold
260K
265K
270K
275K
280K
285K
290K
295K
300K
Global Scale
300 K
295
290
285
280
275
270
265
260
Current Algorithm - Problems
¾ Exaggerates cloud coverage in higher latitudes
¾ Underestimates low-level cloud coverage in lower latitudes
¾ Seasonal variability
¾ Global Scale
New Algorithm
¾ Incorporates Surface Temperatures (Tsfc)
•
6-hourly global products use surface observations over land and
NCEP sea surface temperatures over oceans
•
3-hourly global products use GFS surface temperatures over land and
NCEP sea surface temperatures over oceans
GFS Tsfc Land Temperatures (320-200K Black-White)
NCEP Tsst Ocean Temperatures (320-200K Black-White)
Combined Tsfc Image (320-200K Black-White)
New Algorithm
¾ Incorporates Surface Temperatures (Tsfc)
¾ Temperature difference (Tsfc-Tir) based cloud/no-cloud threshold
Tsfc
Tir
-
=
Tsfc –Tir (-10 – 100 Black-White)
Choosing a Temperature Difference Threshold
Tsfc – Tir =
4
12
-4
8
0
Tir =
285K
275K
265K
Globally
Tsfc – Tir =
0
4
8
New Algorithm
¾ Incorporates Surface Temperatures (Tsfc)
¾ Temperature difference (Tsfc-Tir) based cloud/no-cloud threshold
¾ Transparency based on Tsfc-Tir
Cloud Percentage vs Temperature Difference
100
90
80
% = sin((π/2 * (Tdiff-Tthresh-1)/110.))-6
Percentage
70
60
50
40
30
20
% = tan-1(((Tdiff * (20+Const)/Tthresh)-20) + π/2) / π)
10
0
-10
0
10
20
30
40
50
Tsfc - Tir
60
70
80
90
100
110
Tsfc - Tir
Tir
Basemap
Movie
Current Algorithm - Problems
¾ Rapid surface cooling after sunset produces erroneous clouds
T
85F
70F
Tsfc
= 90F
sfc =
10
Tsfc-Tir = 0
T
75F
70F
Tirir == 90F
Current Algorithm - Problems
¾ Rapid surface cooling after sunset produces erroneous clouds
• Possible Solution – Adjust transparency curve based on date & time
¾ Values of Tsfc – Tir can be small for fog
• Possible Solution – Adjust transparency curve based on relative humidity
Algorithm for Combining Visible and Infrared Data
¾ Visible and Infrared percentages are determined by day and time
Visible(%) = 80 * (sin(Daypct * π/2))2
Infrared(%) = 100 – Visible(%)
where: Daypct = abs(Timeimage-Timenoon)/(Lengthday/2)
Visible and Infrared Pecentages
Sunrise
100
Local Noon
Sunset
90
80
Percentage
70
60
50
40
Infrared
30
Visible
20
10
0
0
1
2
4
5
6
7
8
10 11 12 13 14 16 17 18 19 20 22 23
Time
Algorithm for Combining Visible and Infrared Data
¾ Visible and Infrared percentages are determined by day and time
¾ Transparency based on Visible Albedo and Infrared Brightness Temperature
Transparency
100
90
80
Visible Albedo
70
60
50
40
30
20
10
315
300
285
270
255
240
225
Infrared Brightness Temperature
210
195
180
165
Image Time: 20:45
Sunrise:
11:34
Sunset:
00:20
Visible:
Infrared:
48%
52%
Movie
Availability of Applications and Basemaps
¾ New core supported application available in 2006
• Similar to IMGFILT using FILTER=VISIR, VIS, IR, TDIFF …
• Work with NASA Big Blue Marble or Topography and Enhancement
¾ Global Basemaps available via ftp and possibly ADDE
• 1 km NASA Big Blue Marble – RGB areas – 1 GB/area
• 1 km Topography – 2 byte data – 2 GB/area
• 1 km Topography with Lakes – 2 byte data – 2 GB/area
• 1 km Land Sea Mask – 1 GB
• 10 km Topography and Bathymetry – 9 MB
• 20 km NASA Big Blue Marble and Bathymetry – RGB areas - 2 MB/area
Download

Improved Algorithms for Combining Satellite Imagery and Geographic Basemaps Rick Kohrs