Upload
ovidio
View
44
Download
0
Tags:
Embed Size (px)
DESCRIPTION
Can AERONET help with monitoring clouds?. Alexander Marshak NASA/GSFC Thanks to: Y. Knyazikhin, K. Evans, W. Wiscombe, I. Slutsker and B. Holben Supported by: NASA Radiation Science Program. AERONET Aerosol optical depth (from Cordoba-CETT, Argentina) 22 March, 2000. Clouds. Outline. - PowerPoint PPT Presentation
Citation preview
May 10, 2004 Aeronet workshop
Can AERONET help with monitoring clouds?
Alexander MarshakNASA/GSFC
Thanks to:
Y. Knyazikhin, K. Evans, W. Wiscombe,
I. Slutsker and B. Holben
Supported by:
NASA Radiation Science Program
May 10, 2004 Aeronet workshop
AERONETAerosol optical depth (from Cordoba-CETT, Argentina)
22 March, 2000
Clouds
May 10, 2004 Aeronet workshop
Outline
• Overview of cloud optical depth retrievals from ground-based radiometers;
• Main part: • Cloud optical depth retrievals from Cimel’s radiances,
examples, comparison with other surface retrievals at the ARM site;
• Surface inhomogeneity, sensitivity study;• Local climatology of several Aeronet sites• Testing with simulated clouds;
• Summary/Conclusion
May 10, 2004 Aeronet workshop
Ground-based retrieval
downward radiance downward
flux
upward radiance
(flux)
Common approach is to use downward fluxes:
• broadband pyranometers (Leontieva & Stamnes, 1994; Boers, 1997)
• narrowband radiometers (Min and Harrison, 1996, Min et al., 2003)
from Barker and Marshak, JAS 2001
computed by DISORT:g=0.85, 0=1, surf=0.2
May 10, 2004 Aeronet workshop
Clouds with Low Optical (Water) Depth “CLOWD”
– Optical depth is the most fundamental cloud optical property
– Solar radiation is most sensitive to changes in cloud optical depth at low optical depths
– Over 50% of the warm liquid water clouds at the SGP site have LWP < 100 g m-2
– MWR’s uncertainty is 20-30 g m-2 (i.e., errors of 20% to over 100%)
– Aerosol indirect effect research needs accurate measurements of LWP and effective radius
ARM IS UNABLE TO ADEQUATELY OBSERVE OVER HALF OF THE CLOUDS OVERHEAD !!
Courtesy of Dave Turner, PNNLPresentation at the ARM STM
May 10, 2004 Aeronet workshop
Intercomparison between dif. retrievals for 14 March 2000
(Variable Thickness Stratus Case)
Results from 14 Mar in the ARM 2000 Cloud IOP at the ARM SGP site, a day when the cloud was particularly stratiform and uniform
Raman lidar backscatter
Radar reflectivity
comparisons among many volunteered methods for retrieving the low LWP
20:44
courtesy of Dave Turner, PNNLPresentation at the ARM STM
May 10, 2004 Aeronet workshop
Ground-based retrieval;zenith radiance from simulated 3D clouds
0
0.2
0.4
0.6
0.8
1
1.2
0 10 20 30 40 50 60 70
3D NIR1D NIR
optical depth
May 10, 2004 Aeronet workshop
Objectives
from Michalski et al., 2002
Pilewskie’s SSFR within 0.0001 rad (640 m) from the
CART site flying at 300 m on April 5, 2000
X
X NIR
REDX BLUE
to exploit the sharp spectral contrast in vegetated surface reflect. across 0.7 µm to retrieve cloud properties from
the Cimel’s measurements
May 10, 2004 Aeronet workshop
Cimel radiance measurements (GSFC, Bld. 33): four channels (440, 670, 870, and 1020 nm)
0.0
5.0
10.0
15.0
20.0
25.0
30.0
"18:59:39" "19:07:09" "19:14:39"
1020 nm870 nm670 nm440 nm
normalized radiance (arbitrary units)
time (in GMT)
24 May 1999
Clear sky
BrokenClouds
OvercastSky
(c) Transition between (a) and (b):• sharp changes in I around cloud edges;• the “order” of I between all four channels rapidly changes from cloudy to clear and back.
(b) Surface and Clouds dominate:I440 ≈ I670 < I870 ≈ I1020
Cloud optical properties can be retrieved
(a) Atmosphere dominates:I440 > I670 > I870 > I1020
Aerosol optical properties can be retrieved
May 10, 2004 Aeronet workshop
Two-channel cloud retrievalsSatellite retrieval of and re
from Nakajima-King, JAS 1990
IREDINIRclearcloudyABCα
Surface retrieval and Ac
from Marshak et al., JAS 2004
points A & B are assumed to have the same cloud optical depth but different cloud fraction Ac
May 10, 2004 Aeronet workshop
2D Look-Up TablesNIR vs. RED plane
IRED= IRED (, Ac)INIR = INIR (, Ac)
is cloud optical depthAc is “effective” cloud fraction
0
0.2
0.4
0.6
0.8
1
0 0.1 0.2 0.3 0.4 0.5 0.6
RED
Ac
0.0
0.2
0.4
1.0
0.6
0.8
10060
40 3020
10
1
3
5
0
0.2
0.4
0.6
0.8
0 0.2 0.4 0.6 0.8 1 1.2 1.4
NIR+RED
Ac
0.0
0.2
0.4
1.0
0.6
0.8
100
60
40
30
20
10
3 51
clear
cloudy
May 10, 2004 Aeronet workshop
Where are Cimel data points?
0
0.02
0.04
0.06
0.08
0.1
0.12
0 0.2 0.4 0.6 0.8 1 1.2
NIR+RED
1.0
0.9
0.8
0.7
0.6
70
4020 15 10
30
2
4
Ac
RED
=0.092
NIR
=0.289
=62SZA 0
July 28, 2002 ARM CART siteCimel measurements are taken around
13:45, 13:58 and 14:11 UTTSI is from 14:00 UT
May 10, 2004 Aeronet workshop
Retrieval examples
August 8, 2002;18:00 UT CART site
Cloud optical depth retrieved from:• Cimel• MWR (Microwave Radiometer) assuming re = 7 m
• MFRSR (Multi-Filter Rotating Shadowband
Radiometer)
May 10, 2004 Aeronet workshop
Retrievals
SZA=16.3Time: 17:50
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0 0.2 0.4 0.6 0.8 1 1.2
LUTCimel
NIR+RED
SZA=530+/-1
0
time:14.5-14.6 UMT
30 25 20 15 10
0.5
35
Ac 0.6
1.0
0.7
( 2)b
0.8
0.9
53
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14
0.16
0 0.5 1 1.5 2 2.5 3
NIR+RED
SZA=170+/-1
0
time:17.7-17.9 UMT
302520
15
10
3
0.5
100
Ac
0.7
1.0
0.8
0.9
0.6
( 2)c
SZA=52.3Time: 14:36
MFRSR data is courtesy of Q. Min
0
20
40
60
80
100
13 14 15 16 17 18 19
July 3, 2002
MWRCimelMFRSRSZA
Time(UT)
May 10, 2004 Aeronet workshop
Independent retrieval from BLUE and RED SZA: 29
Time: 20:36
QuickTime™ and aTIFF (Uncompressed) decompressorare needed to see this picture.
SZA: 19Time: 19:32
SZA: 64Time: 14:00
QuickTime™ and aTIFF (Uncompressed) decompressorare needed to see this picture.
SZA: 30Time: 16:24
QuickTime™ and aTIFF (Uncompressed) decompressorare needed to see this picture.SZA: 47Time: 15:17
0
20
40
60
80
100
0 20 40 60 80 100
7.287.03
BLUE vs NIR
MODIS surface albedo:BLUE: 0.044RED: 0.092NIR: 0.289
Scatter-plot of RED_vs_NIR
againstBLUE_vs_NIR
retrievals
0
20
40
60
80
14 16 18 20 22
- retrievals using both blue and red channels 28, 2002July
MWR_07_44_87tau_07_67_87tau
MFRSRSZA
( )Time UT
May 10, 2004 Aeronet workshop
Local Climatology(broadleaf cropland: Bondville, IL)
0
100
200
300
400
500
0 10 20 30 40 50 60 70 80
Bondville (IL)July-August, 2001
optical depth,
1 2 3 4 5
1
2
3
4
5
1 km pixels
1 km pixels
0.1 0.2 0.3 0.4 0.5 0.6Albedo (670 nm)
A2001_193_670_Bondville
1 2 3 4 5
1
2
3
4
5
1 km pixels
1 km pixels
0.1 0.2 0.3 0.4 0.5 0.6Albedo (870 nm)
A2001_193_870_Bondville
1 2 3 4 5
1
2
3
4
5
1 km pixels
1 k mpixels
0.0 0.1 0.2 0.3 0.4 0.5 0.6Albedo (670 nm)
A2001_225_670_Bondville
1 2 3 4 5
1
2
3
4
5
1 km pixels
1 km pixels
0.1 0.2 0.3 0.4 0.5 0.6Albedo (870 nm)
A2001_225_870_Bondville
MODIS srf. albedo670 nm
MODIS srf. albedo870 nm
May 10, 2004 Aeronet workshop
Local Climatology(Santa Barbara, CA: 2003) Version 0 cloud optical depth product
0
100
200
300
400
500
0 20 40 60 80 100
UCSB
tau
0
500
1000
1500
2000
2500
3000
3500
4000
0 0.2 0.4 0.6 0.8 1
UCSB
Ac
0
20
40
60
80
100
0 20 40 60 80 100
UCSB, 2003
y = 0.7 + 0.98x R= 0.996
RED/NIR
Cloud optical depth
“Effective”cloud fraction
RED&NIR vs. BLUE&NIR retrievals
May 10, 2004 Aeronet workshop
Local climatologyARM CART cite
Version 0 optical depth product
0
200
400
600
800
1000
1200
20 40 60 80 100
CART site2001-2003
tau
Cloud optical depth
0
500
1000
1500
2000
2500
3000
0 0.2 0.4 0.6 0.8 1
CART site2001-2003
Ac
"Effective" cloud fraction
0
20
40
60
80
100
0 20 40 60 80 100
ARM CART site2001-2003
y = 0.19 + 0.90x R= 0.986
RED/NIR
RED&NIR vs. BLUE&NIR retrievals
May 10, 2004 Aeronet workshop
Seasonal applicability
0
0.2
0.4
0.6
0.8
1
2001.5 2002 2002.5 2003 2003.5 2004
ARM CART
1 pixel3x3 avg
Date (decimal years)
0
0.1
0.2
0.3
0.4
0.5
0.6
2001.5 2002 2002.5 2003 2003.5 2004
AlbedoBondville, IL
470nm
648nm
858nm
470av
648av
858av
Date (years)
0
0.2
0.4
0.6
0.8
1
2001.5 2002 2002.5 2003 2003.5 2004
Bondville, IL
1 pixel3x3 avg
Date (decimal years)
0
0.1
0.2
0.3
0.4
0.5
2001.5 2002 2002.5 2003 2003.5 2004
AlbedoARM site470nm
648nm
858nm
470av
648av
858av
Date (years)
ARM CART site, OK Bondville, IL
Surface albedo
NDVI
May 10, 2004 Aeronet workshop
How good is the retrieval?Sensitivity to Surface Albedo
July 3, 2002, ARM CART site
0.05 0.10 0.15
0.25
0.30
0.35
RED albedo
NIR albedo
-10 -5 0 5 10 15 20 25 30delta tau in % (0.1 RED, 0.3 NIR ref.)
• If the uncertainties in surface albedo have the same sign, the algorithm performs well.
• If the NIR albedo is overestim. but the RED albedo is underestim., errors in the retrieved are not severe.
• In the opp. case, the algorithm underestim. multiple refl. in the bright band and greatly overestim. .
0.05 0.10 0.15
0.35
0.30
0.25
RED albedo
NIR albedo
-5 -3 0 3 5 8 10delta tau (0.1 RED, 0.3 NIR ref.)
+ +
<-*> <(-*)/*> in %
0
20
40
60
80
100
13 14 15 16 17 18 19
July 3, 2002
MWRCimelMFRSRSZA
Time(UT)
courtesy of A. Trishchenko
from LANDSAT
May 10, 2004 Aeronet workshop
How good is the retrieval?Simulations using stochastic cloud models
0
0.2
0.4
0.6
0.8
1
1.2
0 10 20 30 40 50 60 70
3D (red)
1D (red)
optical depth
3D (nir)
1D (nir)
May 10, 2004 Aeronet workshop
Resultsscatter plots of “retr.” vs. “true”
Total STATISTICS (Ac=81%)
(true) (retr)mean 13.0 13.1std 10.5 10.8
pixel-by-pixel error: 3.0
May 10, 2004 Aeronet workshop
Cumulative error histograms
0
0.2
0.4
0.6
0.8
1
0 5 10 15 20
errors in 3
75% pixels have errors
< 3
May 10, 2004 Aeronet workshop
SummaryThe Main Ideas
• Cimel measures zenith radiance when the Sun is blocked by clouds;
• use two wavelengths:
one in RED (670 nm) [or in BLUE (440 nm)], where vegetation albedo is low, and
one in NIR (870 nm), where vegetation albedo is high
• retrieve cloud optical depth (and cloud fraction) using the NIR vs. RED plane
The Results (so far)
• looks promising; it largely removes 3D effects;
• it is not the final answer but a big improvement against single-wavelength retrievals;
• can fill (cloud) gaps in AERONET aerosol optical depth retrievals and estimate (effective) cloud fraction;
• version 0 cloud optical depth product will be soon available for distribution
May 10, 2004 Aeronet workshop
Can AERONET help with monitoring clouds?
Alexander MarshakNASA/GSFC
Thanks to:
Y. Knyazikhin, K. Evans, W. Wiscombe,
I. Slutsker and B. Holben
Supported by:
NASA Radiation Science Program
May 10, 2004 Aeronet workshop
Dates with the cloud mode data
Bondville: July 2001 - presentCartel: July 2001 - May 2002Egbert: July 2001 - Jan. 2002GSFC: Dec. 2001 - Dec. 2002Harvard Forest: July 2001 - Dec. 2001SGP CART: Oct. 2001 - March 2004Shirahama: Oct. 2001 - May 2002UCSB: May 2003 - present
May 10, 2004 Aeronet workshop
NDCINormalized Difference Cloud Index
-0.2
-0.15
-0.1
-0.05
0
0.05
0.1
0.15
0.2
"18:59:39" "19:07:09" "19:14:39"
NDCI
NDCI
time
24 May 1999
Clear skyBroken cloudsCloud
€
NDCI =I(nir) − I(red)
I(nir) + I(red)
0.0
5.0
10.0
15.0
20.0
25.0
30.0
"18:59:39" "19:07:09" "19:14:39"
1020 nm870 nm670 nm440 nm
normalized radiance (arbitrary units)
time (in GMT)
24 May 1999
Clear sky
BrokenClouds
OvercastSky
May 10, 2004 Aeronet workshop
1D Calculations
€
NDCI =I(nir) - I(red)
I(nir) +I(red)=
1- tanα
1+tanα
where
tanα = I(red) I(nir) on NIR vs.RED plane
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.05
0.1
0.15
0.2
0.25
0 5 10 15 20 25 30 35 40
zenith radiance NDCI
optical depth
I(vis)
I(nir)NDCI
€
vis ≡ red = 0.67μm nir = 0.87μm
May 10, 2004 Aeronet workshop
0.0
5.0
10.0
15.0
20.0
25.0
11:06:40 16:40:00 22:13:20
Argentine (Cordoba-CETT)3/22/00
1020870670440
normalized radiance
time(GMT)
7.0
8.0
9.0
10.0
11.0
12.0
13.0
14.0
13:59:40 14:01:20
Argentine (Cordoba-CETT)3/22/00
1020870670440
normalized radiance
time (GMT)
Cloud observationsNormalized radiance, NDCI and cloud optical depth
zoom
1D retrieval
0.03
0.04
0.05
0.06
0.07
0.08
0.09
0.1
0.11
13:59:20 14:00:40
NDCI
ndci
time (GMT)
0
5
10
15
14:00:00 14:01:00
cloud optical depthretrieved from NDCI
optical depth
time (GMT)
May 10, 2004 Aeronet workshop
How good is the retrieval?Surface Albedo at the ARM SGP site
from Li et al., JGR 2002
+
from Michalski et al., 2002
Pilewskie’s SSFR within 0.0001 rad (640 m) from the
CART site flying at 300 m on April 5,
2000
courtesy of A. Trishchenko
from LANDSAT
May 10, 2004 Aeronet workshop
Local Climatology( a few other locations)
0
200
400
600
800
1000
0 20 40 60 80 100
Egbert, CAN
tau
0
200
400
600
800
1000
1200
1400
0 20 40 60 80 100
Cartel, CAN
tau 0
500
1000
1500
2000
2500
3000
3500
0 0.2 0.4 0.6 0.8 1
Cartel
tau
0
500
1000
1500
2000
2500
3000
3500
0 0.2 0.4 0.6 0.8 1
Egbert, CAN
Ac
optical depth(effective)
cloud fraction
Egbert (2001)
Cartel (2001-2002)
May 10, 2004 Aeronet workshop
Encouraging?
LWP comparisons for March 14th and 15th
Comparison of the Miller microbase algorithm for LWP (based on combining microwave and radar data) against all other volunteered results, for two days in the ARM 2000 Cloud IOP at the ARM SGP site. The cloud was very uniform and stratiform on both days.
20:44
courtesy of Dave Turner, PNNL