36
1 FWC 2006-10-24 IPWG MIT Lincoln Laboratory * This work was sponsored by the National Aeronautics and Space Administration under Contract NNG 04HZ53C, Grant NNG 04HZ51C, and Grant NAG5-13652, and the National Oceanic and Atmospheric Administration under Air Force Contract FA8721-05-C-0002. Opinions, interpretations, conclusions, and recommendations are those of the author and are not necessarily endorsed by the United States Government. Satellite-based Estimation of Precipitation Using Passive Opaque Microwave Radiometry* Frederick W. Chen, Laura J. Bickmeier, William J. Blackwell, R. Vincent Leslie MIT Lincoln Laboratory (Lexington, MA, USA) David H. Staelin, Chinnawat “Pop” Surussavadee MIT Research Laboratory of Electronics (Cambridge, MA, USA) 3 rd Workshop of the International Precipitation Working Group Melbourne, VIC, Australia 24 October 2006

Satellite-based Estimation of Precipitation Using Passive Opaque Microwave Radiometry*

  • Upload
    nicola

  • View
    34

  • Download
    0

Embed Size (px)

DESCRIPTION

Satellite-based Estimation of Precipitation Using Passive Opaque Microwave Radiometry*. Frederick W. Chen, Laura J. Bickmeier, William J. Blackwell, R. Vincent Leslie MIT Lincoln Laboratory (Lexington, MA, USA) David H. Staelin, Chinnawat “Pop” Surussavadee - PowerPoint PPT Presentation

Citation preview

Page 1: Satellite-based Estimation of Precipitation Using Passive Opaque Microwave Radiometry*

1FWC 2006-10-24

IPWG

MIT Lincoln Laboratory

* This work was sponsored by the National Aeronautics and Space Administration under Contract NNG 04HZ53C, Grant NNG 04HZ51C, and Grant NAG5-13652, and the National Oceanic and Atmospheric Administration under Air Force Contract FA8721-05-C-0002. Opinions, interpretations, conclusions, and recommendations are those of the author and are not necessarily endorsed by the United States Government.

Satellite-based Estimation of Precipitation

Using Passive Opaque Microwave Radiometry*

Frederick W. Chen, Laura J. Bickmeier, William J. Blackwell, R. Vincent LeslieMIT Lincoln Laboratory (Lexington, MA, USA)

David H. Staelin, Chinnawat “Pop” SurussavadeeMIT Research Laboratory of Electronics (Cambridge, MA, USA)

3rd Workshop of the International Precipitation Working GroupMelbourne, VIC, Australia

24 October 2006

Page 2: Satellite-based Estimation of Precipitation Using Passive Opaque Microwave Radiometry*

MIT Lincoln Laboratory2FWC 2006-10-24

IPWG

Outline

• Physical basis

• Algorithm development– AMSU (Advanced Microwave Sounding Unit)– ATMS (Advanced Technology Microwave Sounder)

• Future work

• Summary

Page 3: Satellite-based Estimation of Precipitation Using Passive Opaque Microwave Radiometry*

MIT Lincoln Laboratory3FWC 2006-10-24

IPWG

Physical Basis

• Transparent channels (or window channels)– Warm water vapor signatures over cold ocean– Scattering signatures due to ice particles over land

• Opaque channels– Varying atmospheric opacity– Sensitive primarily to specific layers of atmosphere

OPAQUEBANDS

TRANSPARENTBANDS

Page 4: Satellite-based Estimation of Precipitation Using Passive Opaque Microwave Radiometry*

MIT Lincoln Laboratory4FWC 2006-10-24

IPWG

54-GHz and 183-GHz Weighting Functions

54-GHz

183-GHz

Page 5: Satellite-based Estimation of Precipitation Using Passive Opaque Microwave Radiometry*

MIT Lincoln Laboratory5FWC 2006-10-24

IPWG

Estimation of Precipitation Rate with Opaque W Channels

(54-GHz and 183-GHz)

• Precipitation rate ~ humidity × vertical wind velocity

• Absolute humidity– 54-GHz band reveal temperature profile– 54-GHz and 183-GHz bands reveal water vapor

profile

• Vertical wind velocity

– Stronger vertical wind →

– Stronger vertical winds results in increased backscattering of cold space radiation

– Perturbations (cold spots) in 54-GHz data reveal cloud-top altitude

– Absolute albedos reveal hydrometeor abundance– Relative albedos (54 vs. 183-GHz) reveal

hydrometeor size

Greater hydrometeors size

Greater hydrometeor abundance

Higher cloud-top altitude

Page 6: Satellite-based Estimation of Precipitation Using Passive Opaque Microwave Radiometry*

MIT Lincoln Laboratory6FWC 2006-10-24

IPWG

Particle Sizes Revealed in NAST-M Data

54 GHz

118 GHz

183 GHz

425 GHz

Visible

Leslie & Staelin, IEEE TGRS, 10/2004

TB

Page 7: Satellite-based Estimation of Precipitation Using Passive Opaque Microwave Radiometry*

MIT Lincoln Laboratory7FWC 2006-10-24

IPWG

AMSU Radiometry

• Passive W sounder

• AMSU-A– 12 channels in opaque 54-

GHz O2 band– Window channels near 23.8,

31.4, and 89.0 GHz

• AMSU-B– 3 channels in opaque

183.31-GHz H2O band– Window channels near 89.0

and 150.0 GHz

AMSU-A Channel Frequencies (GHz)

23.8

31.4

50.3

52.8

53.596 ± 0.115

54.4

54.94

55.5

57.290344

57.290344 ± 0.217

57.290344 ± 0.3222 ± 0.048

57.290344 ± 0.3222 ± 0.022

57.290344 ± 0.3222 ± 0.010

57.290344 ± 0.3222 ± 0.0045

89.0

AMSU-B Channel Frequencies (GHz)

89.0

150.0

183.31 ± 1

183.31 ± 3

183.31 ± 7

Page 8: Satellite-based Estimation of Precipitation Using Passive Opaque Microwave Radiometry*

MIT Lincoln Laboratory8FWC 2006-10-24

IPWG

General Structure of AMSU Algorithm(Chen and Staelin, IEEE TGRS, 2/2003)

• Signal processing– Regional Laplacian interpolation– Image sharpening– Principal component analysis

• Neural net– 2-layer feedforward neural net– 1st layer: tanh transfer function– 2nd layer: linear transfer function

Page 9: Satellite-based Estimation of Precipitation Using Passive Opaque Microwave Radiometry*

MIT Lincoln Laboratory9FWC 2006-10-24

IPWG

Signal Processing Components

• Neural-net correction of angle-dependent variations in TB’s

• Cloud-clearing via regional Laplacian interpolation– Temperature profile characterization– Cloud-top altitude characterization

• Principal component analysis for dimensionality reduction– Temperature profile PC’s

– Window channel / H2O profile PC’s

• Image sharpening– AMSU-A data sharpened to AMSU-B resolution

Page 10: Satellite-based Estimation of Precipitation Using Passive Opaque Microwave Radiometry*

MIT Lincoln Laboratory10FWC 2006-10-24

IPWG

The Algorithm: Precipitation Masks &Precipitation-Induced Perturbations

PR

EC

IPIT

AT

ION

DE

TE

CT

ION IMAGE

SHARPENING

CO

RR

UP

T D

AT

AD

ET

EC

TIO

N

LIM

B-&

-SU

RF

AC

EC

OR

RE

CT

ION

REGIONALLAPLACIAN INTERPOLATION

Page 11: Satellite-based Estimation of Precipitation Using Passive Opaque Microwave Radiometry*

MIT Lincoln Laboratory11FWC 2006-10-24

IPWG

The Algorithm: Neural Net

Trained to NEXRAD

Page 12: Satellite-based Estimation of Precipitation Using Passive Opaque Microwave Radiometry*

MIT Lincoln Laboratory12FWC 2006-10-24

IPWG

Final Output

Page 13: Satellite-based Estimation of Precipitation Using Passive Opaque Microwave Radiometry*

MIT Lincoln Laboratory13FWC 2006-10-24

IPWG

Example of Global Retrieval

Page 14: Satellite-based Estimation of Precipitation Using Passive Opaque Microwave Radiometry*

MIT Lincoln Laboratory14FWC 2006-10-24

IPWG

ATMS

• Similar to AMSU

• To be launched on NPP (2009) & NPOESS satellites– NPP = NPOESS Preparatory Project

• Improvements over AMSU– Additional channels in 54-GHz and 183-GHz bands– Better resolution in 54-GHz band– Better sampling

Nyquist sampling of 54-GHz data Identical sampling of all channels

Page 15: Satellite-based Estimation of Precipitation Using Passive Opaque Microwave Radiometry*

MIT Lincoln Laboratory15FWC 2006-10-24

IPWG

Simulating ATMS TB’s

• MM5 Atmospheric Circulation Model– Provides temperature profile, water vapor profile, hydrometeor profile, …– Used Goddard hydrometeor model (Tao & Simpson, 1993)

• Radiative Transfer– TBSCAT due to Rosenkranz (IEEE TGRS, 8/2002)

Multi-stream, initial-value– Improved hydrometeor modeling due to Surussavadee & Staelin (IEEE TGRS,

10/2006)

• Filtering– Accurate matching of TB’s on MM5 grid to ATMS resolution and geolocation

using “satellite geometry” toolbox for MATLAB Computing angular offset of surface locations from boresight Computing satellite zenith angles from scan angle Computing geolocation from scan angle

Page 16: Satellite-based Estimation of Precipitation Using Passive Opaque Microwave Radiometry*

MIT Lincoln Laboratory16FWC 2006-10-24

IPWG

MM5 Rain Rate: Typhoon Pongsona, 2002/12/8

Page 17: Satellite-based Estimation of Precipitation Using Passive Opaque Microwave Radiometry*

MIT Lincoln Laboratory17FWC 2006-10-24

IPWG

AMSU vs. ATMS, 183±7 GHz

Observed AMSU Simulated ATMS

• Simulated ATMS 183±7 GHz data shows reasonable agreement with AMSU-B

• Morphology difference between AMSU observations and MM5 predicted radiances is due to the inaccuracy of the NCEP analyses used to initialize the MM5 model

Page 18: Satellite-based Estimation of Precipitation Using Passive Opaque Microwave Radiometry*

MIT Lincoln Laboratory18FWC 2006-10-24

IPWG

AMSU vs. ATMS, 50.3 GHz

Observed AMSU Simulated ATMS

• Simulated ATMS 50.3-GHz data with finer resolution and sampling shows finer features than AMSU-A

• Intense eyewall signature in simulated ATMS 50.3-GHz data due to NCEP initialization & limited 5-hr MM5 spinup time producing excess of large ice particles

Page 19: Satellite-based Estimation of Precipitation Using Passive Opaque Microwave Radiometry*

MIT Lincoln Laboratory19FWC 2006-10-24

IPWG

Future Developments

• Adapting Chen-Staelin algorithm (IEEE TGRS, 2/2003) for ATMS

• Exploiting Nyquist sampling in the 54-GHz band

• Using methods from window-channel-based algorithms

• Improving the signal processing of Chen-Staelin algorithm

• Improving neural net training– Representations of circular data

Page 20: Satellite-based Estimation of Precipitation Using Passive Opaque Microwave Radiometry*

MIT Lincoln Laboratory20FWC 2006-10-24

IPWG

Recently Launched & Future Instruments

• Similar to AMSU-A/B– AMSU/MHS on NOAA-18 (2005)– AMSU/MHS on NOAA-N’, METOP-1, METOP-2, METOP-3

• ATMS– NPP (2009)– NPOESS

W instruments on geostationary satellites?– < 1 hr revisit times

Page 21: Satellite-based Estimation of Precipitation Using Passive Opaque Microwave Radiometry*

MIT Lincoln Laboratory21FWC 2006-10-24

IPWG

Summary

• Physical basis of precipitation estimation using opaque W channels

– Atmospheric sounding capabilities of opaque W channels– Cloud shape and particle size distribution from NAST-M 54-,

118-, 183-, and 425-GHz data

• AMSU precipitation algorithm– Relies primarily on 54-GHz and 183-GHz opaque bands– Signal processing: regional Laplacian interpolation, principal

component analysis, image sharpening

• ATMS precipitation algorithm development– Simulation system using MM5/TBSCAT

Page 22: Satellite-based Estimation of Precipitation Using Passive Opaque Microwave Radiometry*

22FWC 2006-10-24

IPWG

MIT Lincoln Laboratory

Backup Slides

Page 23: Satellite-based Estimation of Precipitation Using Passive Opaque Microwave Radiometry*

MIT Lincoln Laboratory23FWC 2006-10-24

IPWG

NAST-M

• NAST = NPOESS Aircraft Sounder Testbed– Risk-reduction effort by NPOESS Integrated Program Office– Cooperative effort of NASA, NOAA, & DoD

• Equipped with 54-, 118-, 183-, and 425-GHz radiometers

• Flown on high-altitude aircraft– ER-2 (NASA)– Proteus (Scaled Composites)

• ~2.5-km resolution near nadir

Page 24: Satellite-based Estimation of Precipitation Using Passive Opaque Microwave Radiometry*

MIT Lincoln Laboratory24FWC 2006-10-24

IPWG

Scattering in the 54-GHz and 183-GHz Bands

0.7 mm 2.4 mm

Page 25: Satellite-based Estimation of Precipitation Using Passive Opaque Microwave Radiometry*

MIT Lincoln Laboratory25FWC 2006-10-24

IPWG

AMSU Geographical Coverage

• Aboard NOAA-15, NOAA-16, & NOAA-17

• Nearly identical AMSU/HSB on Aqua

Page 26: Satellite-based Estimation of Precipitation Using Passive Opaque Microwave Radiometry*

MIT Lincoln Laboratory26FWC 2006-10-24

IPWG

AMSU-A/B Sampling & Resolution

• AMSU-A– 3 1/3° sampling (~50 km near nadir)– 3.3° resolution (~50 km near nadir)

• AMSU-B– 1.1° resolution (~15 km near nadir)– 1.1° sampling (~15-km near nadir)

AMSU-A

AMSU-B

Page 27: Satellite-based Estimation of Precipitation Using Passive Opaque Microwave Radiometry*

MIT Lincoln Laboratory27FWC 2006-10-24

IPWG

15-km AMSU vs. NEXRAD Comparison

Page 28: Satellite-based Estimation of Precipitation Using Passive Opaque Microwave Radiometry*

MIT Lincoln Laboratory28FWC 2006-10-24

IPWG

RMS Discrepancies (mm/h)between AMSU and NEXRAD

Range ofNEXRAD rain

rate

15-km(30-110 km from radar)

15-km(110-230 km from radar)

50-km(30-110 km from radar)

50-km(110-230 km from radar)

< 0.5 mm/h 1.0 1.4 0.5 0.5

0.5 – 1 mm/h 2.0 2.6 0.9 1.1

1 – 2 mm/h 2.3 2.7 1.1 1.5

2 – 4 mm/h 2.7 3.9 1.8 2.3

4 – 8 mm/h 3.5 7.4 3.2 5.2

8 – 16 mm/h 6.9 8.4 6.6 6.5

16 – 32 mm/h 19.0 17.2 12.9 14.6

> 32 mm/h 42.9 39.2 22.1 21.7

Page 29: Satellite-based Estimation of Precipitation Using Passive Opaque Microwave Radiometry*

MIT Lincoln Laboratory29FWC 2006-10-24

IPWG

Features of ATMS vs. AMSU

• Channel set– Similar to AMSU Additional 51.76-GHz channel Additional 183.31±4.5-GHz & 183.31±1.8-GHz– 165.5-GHz replaces 150-GHz on AMSU-B No 89.0-GHz 15-km channel (available on AMSU-B)

• Resolution 54-GHz and 89-GHz: 2.2° vs. 3.33° on AMSU 23.8- and 31.4-GHz: 5.2° vs. 3.33° on AMSU

• Sampling– 165.5-GHz, 183-GHz: Similar to AMSU-B Other channels: ~3× finer than AMSU-A cross-track & along-track All channels sampled at the same locations Nyquist sampling of 54-GHz and 89-GHz

• Similar sensitivity

Page 30: Satellite-based Estimation of Precipitation Using Passive Opaque Microwave Radiometry*

MIT Lincoln Laboratory30FWC 2006-10-24

IPWG

ATMS & AMSU Footprints

Page 31: Satellite-based Estimation of Precipitation Using Passive Opaque Microwave Radiometry*

MIT Lincoln Laboratory31FWC 2006-10-24

IPWG

ATMS & AMSU Footprints (Near Nadir)

Page 32: Satellite-based Estimation of Precipitation Using Passive Opaque Microwave Radiometry*

MIT Lincoln Laboratory32FWC 2006-10-24

IPWG

ATMS Rain Rate Retrieval Algorithm

• Completely new algorithm

• Neural net

• Inputs– All 22 channels– sec(satellite zenith angle)

• Training, validation, and testing sets– MM5 data over Typhoon Pongsona– 1 time step (1521 data points) each for training, validation,

and testing

Page 33: Satellite-based Estimation of Precipitation Using Passive Opaque Microwave Radiometry*

MIT Lincoln Laboratory33FWC 2006-10-24

IPWG

ATMS vs. MM5, 1.1°

Page 34: Satellite-based Estimation of Precipitation Using Passive Opaque Microwave Radiometry*

MIT Lincoln Laboratory34FWC 2006-10-24

IPWG

ATMS vs. MM5, 5.2°

Page 35: Satellite-based Estimation of Precipitation Using Passive Opaque Microwave Radiometry*

MIT Lincoln Laboratory35FWC 2006-10-24

IPWG

Representations of Geolocation

Rectangular (2-D)

Discontinuity across 180° E/W (Int’l Date Line)

Topological distortion around 90° N & 90° S (Geo. N & S Poles)

Cylindrical (3-D)

Continuity across 180° E/W

Topological distortion around 90° N & 90° S

Spherical (3-D)

Continuity across 180° E/W

No topological distortion around 90° N & and 90° S

Page 36: Satellite-based Estimation of Precipitation Using Passive Opaque Microwave Radiometry*

MIT Lincoln Laboratory36FWC 2006-10-24

IPWG

Geolocation:Comparing the Representations

• Spherical representation produces the lowest RMS errors

• RMS error with 10 weights & biases

• Linear: 0.16• Cylindrical: 0.16• Spherical: 0.01

• Weights & biases needed for RMS error < 1.5 × 10-2

• Rectangular: 23• Cylindrical: 18• Spherical: 6

RECTANGULAR

CYLINDRICAL

SPHERICAL