Cloud/Rain partitioning using MODIS and AMSR-E

Cloud/Rain partitioning using MODIS and AMSR-E

Matt Lebsock

AquaAMSR-E & MODIS

•Drizzle defined where 750 meter reflectivity exceeds -15 dBZ•Area and Low cloud fraction weighted oceanic mean is 19.2%

The world is a drizzly place

• The water path can be partitioned between cloud and Precipitation

• @ Visible/Near-Infrared wavelengths Qext->2

• @ Microwave frequencies– Tb are sensitive to both cloud and precipitation water

€

W total =Wcloud +W precip

€

τ =3Wcloud

2ρ lre,cloud+3W precip

2ρ lre,precip

Physical Basis For Cloud/Rain Separation

Small

Spatial Resolution• MODIS

– 1km• AMSR-E

• Use Backus-Gilbert method to resample the AMSR-E footprints to a common resolution @ 23 GHz (31km x 18km)

• Average the MODIS cloud products to the 23 Ghz resolution with the antenna gain function as a weighting parameter.

FREQUENCIES (GHz) 6.925 10.65 18.7 23.8 36.5 89.0

IFOV(km x km) 74 x 43 51 x 30 27 x 16 31 x 18 14 x 8 6 x 4

Optimal Estimation Retrieval

€

Φ= y −F[ ]TSy

−1 y −F[ ] + x − xa[ ]T Sa

−1 x − xa[ ]

Parameter A priori Source Constraint

CWV ECMWF-interim 1.5 kgm-2

SST ECMWF-interim 0.25 K

WIND ECMWF-interim 2.5 ms-1

CWP MODIS 1000 gm-2

6.9 10.7 18.7 23.8 36.5 89.0V H V H V H V H V H V H

Native 0.32 0.34 0.49 0.57 0.55 0.47 0.56 0.54 0.51 0.41 1.18 0.91Re-sampled 0.52 0.55 0.48 0.56 0.27 0.23 0.56 0.54 0.14 0.11 0.19 0.15Assumptions 0.6 0.98 0.8 1.28 1.07 1.74 1.20 1.68 1.21 2.30 1.87 3.47

Observational Uncertainty

A-Priori Uncertainty

January 2007 Results

No Cloud Fraction

Cloud Fraction

Precipitation Water

January 2007 Statistics

Radiative Transfer SimulationsBrightness Temperatures Polarization

There does appear to be a signal in the brightness temperatures

Emission Signal Scattering Signal Depolarization Signal

Water Vapor?

Land Influence?

Continued Work

• Add a realistic DSD• Add rain fraction to retrieval• Include an optical depth constraint

€

Φ= y −F[ ]TSy

−1 y −F[ ] + x − xa[ ]T Sa

−1 x − xa[ ] +τ − τ a[ ]

2

σ τ2

€

x = CWV,SST, WIND,TWP,r[ ]

Summary

• It appears possible to place bounds on the ratio of precipitation water to cloud water in liquid clouds.– Critical assumptions

• Precipitation DSD• Insensitivity of MODIS to precipitation

• Applications– Improved GPROF database– Studies on the control of precipitation production

• Aerosol indirect effects• Thermo-dynamical controls

GPROF Algorithm

Radiometer ObsTMI, AMSR-E, SSM/I

Background retrievalSST, TPW

Compare Tb with a-priori database

RadiometerAppropriate

Database

Rainfall Product

€

P(R |Tb ) = P(Tb |R)P(R)

Does the A-priori database containThe correct statistics of rain/no-rain

GPROF 2008 Database Generation

Agreement ?

Add rain below sensitivity of PR&

Recombine and Compute Tb

TRMMRadiometer Obs

Rain?Background Retrieval

SST, Wind, TPW, CLW

Radar Rain Profile

Combine & Compute Tb

TRMMRadar Obs

No

Yes

Compare Tb to Obs

YesNo

Compare Tb to Obs

Agreement ?Make Database

Modify DSD in PR pixels w/o PIA information&

Recombine and Compute Tb

Yes

Compare Tb to Obs

Make Database

No Yes

Make Database

PR-CloudSat Matchups

10%

50%

Berg et al., 2010

7 Wm-2