temporal interpolation using geostationary data: it’s ... · 25rd ceres science team meeting...

NASA Langley Research Center / Atmospheric Sciences

Temporal Interpolation UsingGeostationary Data: It’s About Time

D. Young, T. Wong, P. Minnis, and K. CostulisNASA Langley Research Center

J. Stassi, C. Nguyen, R. Raju, S. Sun-Mack,J. Boghosian, and E. Kizer

SAIC

P. Heck, J. KenyonAS&M

25rd CERES Science Team MeetingBrussels, Belgium

January 21-23, 2002


Time Interpolation and Spatial Averaging (TISA)Overview of Talk

• What is TISA ?

• TISA Status

• Monthly mean comparisons with ERBE-like

• Using geostationary data– Calibration of imager data

– Cloud property retrievals

• Results– Comparisons of monthly means w/ and w/o GGEO data

– Calibration sensitivity

– Exploring options

• Validation


Where TISA Fits Into CERES Processing

CERESInstrument

Data

Geolocate&

Calibrate1

ERBE-LikeInversion

2

ERBE-LikeAveraging

3

Clouds & TOA Flux

4

Grid TOA & SRB

9

MonthlyTOA & SRB

10

SARB5

GridSARB

6

TimeInterpolate

SARB7

MonthlyMeanSARB

10

GridGeoData11

ERBE-LikeProducts

TOA and Surface

Products

Synoptic & MonthlyTOA, Surface, and

Atmospheric Products


TISA Products

• Subsystem 10 (Data Product: SRBAVG)– Two Interpolation Methods (GGEO & non-GGEO)

– Uses Cloud Properties from GGEO Data

– Clear Sky Interpolation Using GGEO

– Total-sky Fluxes Derived Using ERBE-like ADMs

– Surface Fluxes

• SFC– Gridded SSF

– Used as input to producing SRBAVG

• GGEO– Gridded Geostationary imager data

– Used for temporal interpolation


Status of TOA and Surface Products

• QC reports and plots developed• Initial Validation of Flux and cloud algorithms

– Possible minor changes to clear-sky

• Transitional Format– Accommodate both ERBE and new DRMs– DRM construction and validation (# ADMS changed from 12

to 592)

• GGEO data– Calibration now automated

• Testing GGEO-radiance vs GGEO cloud albedomethods

• Validation in progress


ERBE-like Directional Models(Albedo as a Function of Solar Zenith Angle)

BB

BB

BB

BB

BBB

J

J

J

JJ

JJJJ

JJ

H

H

H

HH

HHHHHH

F

F

FF

FFFFFFF

0

10

20

30

40

50

60

70

0 30 60 90

Alb

edo

(per

cent

)

Solar Zenith Angle (degrees)

B Overcast

J Mostly Cloudy

H Partly Cloudy

F Clear

OCEAN

BB

BB

BB

BB

BBB

JJ

JJ

JJ

JJJJJ

H

HH

HH

HHHHHH

F

FF

FFFFFFFF

0

10

20

30

40

50

60

70

0 30 60 90

Alb

edo

(per

cent

)

Solar Zenith Angle (degrees)

B Overcast

J Mostly Cloudy

H Partly Cloudy

F Clear

LAND


Using Imager Data from GeostationarySatellites for Interpolation (GGEO)

• Define diurnal variations• Define cloud properties• Need calibration• LW

– Radiance - to - flux conversion– Narrowband-to-Broadband– Normalize to CERES

• SW– Radiance to flux conversion - need cloud properties– Narrowband-to-Broadband– Normalization– Second method by Haeffelin


CERES Time Interpolation IncorporatesGeostationary Data to Reduce Instantaneous Errors

Mean Instantaneous Interpolation Rms Errors Are Reduced By 50%For Both LW And SW TOA Flux Using Geostationary Data

B

B

B

B

BB

B

B

BB

B

B

B

B

BB

B

BBB

B

B

B

B

B

B

B

B

B

B

BB

BB

B

B

B

B

B

B

B

B

B

B

B

J

J

JJ

J

J

JJ

J

JJ

JJ

JJ

JJ

J

J

J

JJ

J

J

J

JJJJ

J

JJ

JJ

J

JJ

150

200

250

300

350

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

LW F

lux

(W/m

2)

Local Time (Day of Month)

B ERBS data

J NOAA-9 data

ERBE TSA

GOES 3-hourlyCERES TSA


Calibration Technique

• Calibration is tied to VIRS for TRMM– VIRS calibration appears to be stable for visible & IR

channels (Minnis et al. 2002)

– Using VIRS as standard provides consistency with CERESinstantaneous cloud properties

• Gridded geostationary imager data matched with

VIRS data in time and angle

• Separate correlations for each imager

• Separate correlations for land and desert

• Calibration trends compared with Minnis et al.


Recalibrated GOES-9 Visible Data

Slope = 1.51rms = 15.5

VIRS IR

Radiance (W/m2/sr/µm)

GOES-9 IR Radiance (W/m2/sr/µm)


METEOSAT Ocean IR

0

2

4

6

8

10

12

0 2 4 6 8 10 12

GGEO Radiance

VIR

S R

adia

nce

Feb-98Mar-98Apr-98May-98Jun-98

GGEO Calibration Stability


GGEO IR Calibration Stability

-5

0

5

10

15

20

Feb-98

Mar-98

Apr-98

May-98

Jun-98

Jul-98

Aug-98

Month

% D

iffer

ence

GMS-5GOES-9GOES-8METEOSAT

% Difference at 238 K

-8

-7

-6

-5

-4

-3

-2

-1

0

Feb-98

Mar-98

Apr-98

May-98

Jun-98

Jul-98 Aug-98

Month

% D

iffer

ence

% Difference at 295 K


GGEO Visible Calibration Stability

0

10

20

30

40

50

60

Feb-98

Mar-98

Apr-98 May-98

Jun-98 Jul-98 Aug-98

% D

iffer

ence GMS-5

GOES-9GOES-8METEOSAT

-10

0

10

20

30

40

50

60

Feb-98

Mar-98

Apr-98

May-98

Jun-98

Jul-98

Aug-98

% D

iffer

ence

% Difference at Low Radiance % Difference at High Radiance


GGEO Cloud Property Retrievals

• Needed for SW interpolation & for monthly clouds

• Uses IR/Vis LBTM retrievals (run as subset ofCERES cloud algorithm)

• Uses CERES surface property maps and MOAsoundings

• Properties– Cloud Amount

– Cloud Temperature

– Cloud Height (using standard 4 CERES layers)

– Optical Depth/Emittance (Daytime Only)


GGEO Cloud Amount 18 GMT February 6, 1998


GGEO Zonal Mean Cloud AmountFebruary 1998

0%

20%

40%

60%

80%

100%

-90 -60 -30 0 30 60 90

Latitude

Tot

al C

loud

Am

ount

(%

)

GGEO DayGGEO Night


GGEO and ISCCP Zonal Mean Cloud Amount GEO: February 1998 ISCCP: February 1986-1993

0%

20%

40%

60%

80%

100%

-90 -60 -30 0 30 60 90

Latitude

Tot

al C

loud

Am

ount

(%

)

GGEO DayGGEO NightISCCP Day+Night


Zonal Mean Cloud Amount Comparison GEO & VIRS: February 1998 ISCCP: February 1986-1993

0%

20%

40%

60%

80%

100%

-90 -60 -30 0 30 60 90

Latitude

Tot

al C

loud

Am

ount

(%

)

GGEO DayGGEO NightVIRS DayVIRS NightISCCP Day+Night


Instantaneous VIRS-GOES-9 ComparisonCloud Percentage

VIRS: 71.1% GOES-9: 71.7% Mean Difference: 0.6% RMS:14.1%

0

500

1000

1500

2000

2500

3000

-1 -0.6 -0.2 0.2 0.6 1

Bin

Freq

uenc

y


Instantaneous VIRS-GOES-9 ComparisonCloud Temperature and Optical Depth

VIRS: 268.9K GOES-9: 269.2K VIRS: 7.8 GOES-9: 6.0


Summary of GGEO-VIRS InstantaneousCloud Fraction Comparison

0.600.690.460.46GMS-5

0.520.520.430.53METEOSAT-6

0.670.590.470.47GOES-9

0.590.720.670.68GOES-9

VIRSGGEOVIRSGGEO

LandOcean


Summary of GGEO-VIRS InstantaneousCloud Temperature (K) Comparison

252.9256.1270.8274.6GMS-5

261.0265.5272.6273.2METEOSAT-5

259.7260.0270.2270.4GOES-8

258.1255.2268.9269.2GOES-9

VIRSGGEOVIRSGGEO

LandOcean


Summary of GGEO-VIRS InstantaneousCloud Optical Depth Comparison

8.67.18.56.7GMS-5

9.710.66.84.6METEOSAT-5

14.010.810.48.8GOES-8

10.013.77.86.0GOES-9

VIRSGGEOVIRSGGEO

LandOcean


Future GGEO Cloud Property Validation

• Look at all 8 months

• Viewing and Solar Zenith Dependence

• Deep Convective Cloud Albedos

• Finalize Calibration– Smooth trend

– Eliminate issues from poor sampling

• Intercomparisons– Overlapped GGEO Satellites

– ARM Satellite & Surface Observations


Flux Comparisons

• ERBE-like– 2.5° Grid– VZ < 70 °– Scene Identification by MLE– Old ERBE ADM

• SS10 CERES Monthly Means– 1.0° Grid– VZ < 45 °– Scene Identification from VIRS– New CERES ADM

• Comparison Method– Scatter Plots of SS10 Data Matched to ERBE-like 2.5° Data– Compare Histograms of Fluxes


Monthly Mean Total-sky LW Flux(non-GGEO Method)


Comparison of ERBE-like and SS10 non-GGEOTotal Sky LW Fluxes

Mean Difference = 0.6 W/m2

Sigma = 6.2 W/m2


ES4 ERBE-like and SRBAVG Flux Summary

9.018.220.7Sigma

-0.949.949.0MeanClear-Sky

SW Flux

5.814.016.6Sigma

1.2290.7291.9MeanClear-Sky

LW Flux

12.029.233.2Sigma

Mean

Sigma

Mean

-2.796.193.4Total-Sky

SW Flux

6.228.729.1

-0.6261.3260.7Total-Sky

LW Flux

ES4 -SRBAVG

SRBAVGnonGGEO

ERBE-like

(ES-4)

40°N - 40°S

W/m2


Monthly Mean Total-sky LW Flux(GGEO Method)


Monthly Mean Total-sky LW Flux Difference(non-GGEO Method - GGEO Method)


Histogram of ERBE-like and SS10 Total-SkyLW Fluxes

0

0.5

1

1.5

2

2.5

3

3.5

4

150 200 250 300 350

Total-sky LW Flux (W/m2)

Fre

quen

cy (

%)

ERBE-likeSS10 non-GGEOSS10 GGEO


Monthly Mean Clear-sky LW Flux Difference(non-GGEO Method - GGEO Method)


Histogram of ERBE-like and SS10 Clear-SkyLW Fluxes

0

1

2

3

4

5

6

7

8

150 200 250 300 350

Clear-sky LW Flux (W/m2)

Fre

quen

cy (

%)

ERBE-likeSS10 non-GGEOSS10 GGEO


nonGGEO - GGEO Total-Sky LW FluxDifferences (W/m2)

0

2

4

6

8

10

12

14

16

18

-25 -15 -5 5 15 25

nonGGEO - GGEO Total-Sky LW Flux Difference (W/m2)

Fre

quen

cy (

%)


Histogram of ERBE-like and SS10 Total-SkySW Fluxes

0

0.5

1

1.5

2

2.5

3

0 50 100 150 200 250

Total-sky SW Flux (W/m2)

Fre

quen

cy (

%)

ERBE-likeSS10 non-ERBESS10 GGEO


Histogram of ERBE-like and SS10 Clear-SkySW Fluxes

0

2

4

6

8

10

12

14

16

0 50 100 150 200 250

Clear-sky SW Flux (W/m2)

Fre

quen

cy (

%)

ERBE-likeSS10 non-ERBESS10 GGEO


SRBAVG ERBE and GGEO Clear-Sky SWFlux Comparison

Mean Difference = -0.9 W/m2

Sigma = 5.0 W/m2


SRBAVG nonGGEO and GGEO Flux Summary

5.020.020.7Sigma


SW Flux

5.015.716.7Sigma


LW Flux

9.431.433.3Sigma

Mean

Sigma

Mean

1.399.3100.6Total-Sky

SW Flux

4.528.829.1

0.2257.6257.8Total-Sky

LW Flux

nonGGEO- GGEO

SRBAVGGGEO

SRBAVGnonGGEO

40°N - 40°S

W/m2


Tropical Mean Comparison

Total-Sky SW

Clear-Sky SW


Calibration Sensitivity Test

• Test effect of imager calibration on monthly meanfluxes

• Test by varying imager gain by ±5%

• Affects both radiances and cloud retrievals


Effect of Calibration Change (IR - 5%)


Calibration Sensitivity Summary

Vis - 5%Vis + 5%IR - 5%IR + 5%

-0.22

(0.94)

0.21

(1.11)

0.05

(4.20)

0.01

(1.33)51.5

Clear-sky SW

-0.02

(0.26)

0.01

(0.27)

0.30

(0.92)

-0.29

(0.69)284.7

Clear-sky LW

-0.94

(1.31)

0.94

(1.31)

0.54

(3.10)

-0.04

(1.35)99.3

Total-skySW

0.00

(0.00)

0.00

(0.00)

-0.01

(0.08)

0.01

(0.08)257.6

Total-skyLW

Mean & rms Flux Difference (W/m2)MeanFlux


Using Cloud Albedo Models for Imager TimeSeries

• Directional Models (DRM) areconstructed empirically fromCERES data

• Use imager-derived cloudproperties to construct total albedoat interpolation times

• Models based on phase and opticaldepth

• Only use overcast DRM’s• Albedo time series normalized to

CERES observations

• More details provided in Co-I talkby Jeff Boghosian

Albedo

Solar Zenith Angle


Difference in Monthly Men SW Flux Using GGEO-radiance and GGEO-Cloud Albedo Methods


Calibration issues GGEO-Cloud AlbedoMethod


Imager Calibration Effect on Monthly MeanTotal-sky SW

Vis - 5%Vis + 5%IR - 5%IR + 5%

-2.47

(2.97)

2.55

(3.02)

4.13

(5.82)

-0.44

(1.14)98.6

Cloud

Albedo

Method

-0.94

(1.31)

0.94

(1.31)

0.54

(3.10)

-0.04

(1.35)99.3

Radiance

Method

Mean & rms Flux Difference (W/m2)MeanFlux


Comparison of SS10 and Surface Fluxes

• Use GGEO-interpolated 1°x1° Gridded Fluxes

• TOA Flux interpolated to all hours of Month

• Surface Fluxes Computed Using CERES TOA-Surface Algorithms– Downwelling Clear-sky SW

– Downwelling Total-Sky LW

– Downwelling Clear-Sky LW

• Match with 30-minute Averaged Data (Centered onLocal Half-Hour) From ARM Central Facility

• Compare Bias and rms of Interpolated Comparisonwith Instantaneous Results of Kratz et al.


Downwelling Total Sky LW Time SeriesARM SGP February 1998


Surface Downwelling Clear-sky SW Time SeriesARM Central Facility February 7, 1998

CERES Observation Times


Instantaneous & Interpolated DownwellingSurface Total-Sky SW Flux ARM Central Facility February 1998

CERES Footprint vs. Surface

Bias = -8.0 W/m2

RMS = 58.8 W/m2

Interpolated vs. Surface

Bias = -1.4 W/m2

RMS = 90.1 W/m2


Instantaneous & Interpolated DownwellingSurface Clear-Sky SW Flux (Model A)

ARM Central Facility February 1998


Bias = 27.4 W/m2

RMS = 46.1 W/m2


Bias = -32.5 W/m2

RMS = 55.9 W/m2


Instantaneous & Interpolated DownwellingSurface Total-Sky LW Flux (Model B)

ARM Central Facility February 1998


Bias = -0.4 W/m2

RMS = 19.5 W/m2


Bias = 1.8 W/m2

RMS = 28.2 W/m2


Interpolated Downwelling SurfaceClear-Sky LW Flux (Model B)

ARM Central Facility

July 1998

Bias = 0.5 W/m2

RMS = 12.7 W/m2

February 1998

Bias = -10.5 W/m2

RMS = 27.3 W/m2

Nighttime points


Future Plans

• Finalize Algorithms– Final DRM

– Fine-tune Clear-sky Algorithms

– Finalize Total-Sky SW Algorithm

• Continued Validation– Finalize Geostationary Calibrations

– Nine Months of TOA and Surface Flux Comparisons

– Study Cloud Normalization and Night Clouds

– Compare Monthly Mean and Aggregate Clouds

• Scheduled Archival in May 2002

temporal interpolation using geostationary data: it’s ... · 25rd ceres science team meeting...

Documents