SPACEBORNE_MICROWAVE_OBSERVATIONS_OF_RAIN 1972-1997.ppt

SPACEBORNE MICROWAVE OBSERVATIONS OF RAIN 1972-1997

Thomas T. WilheitTexas A&M Univ.Alfred T. C. Chang

Formerly of NASA/GSFC

Period Runs from Launch of ESMR on Nimbus 5 to Launch of TRMM

Al Chang (no longer with us) was a key player in all of this.

Graphics a messy problemPowerPoint didn’t exist in 1972Color was rare.

Organization of TalkIR-Based Rainfall RetrievalsPassive Microwave over OceanPassive Microwave over LandProfiling AlgorithmsAlgorithm WorkshopsRadar

IR Measurements Estimates of Rainfall

As far back as the ‘60s it was noted that cold clouds (IR) and bright clouds (VIS) were correlated with rainfall.

Bright clouds seemed slightly better correlated but nighttime problemThus, IR measurements attracted more interest.

Geosynchronous satellites provided enough observations to be very useful.

IR rain algorithms became a cottage industry.

Precipitation Measurements from Space Workshop (1981) had presentations on 6 different techniques. I don’t think any are in current use.

Phil Arkin’s Method

One simple method seems to have survived.

Phil Arkin used the ship radar data from the 1974 GARP Atlantic Tropical Experiment (GATE) and Geosync data to examine correlations between cloud top temperature and rainfall.

6 hour accumulations had ~80% correlations with cloud top temperatures colder than 235K. (slope ~ 3mm/hr)

i.e. If the cloud top is colder than 235, it’s raining 3 mm/hr.

Shows a useful degree of skill.

Assorted refinements over the years

The point is the desperation of the meteorological community for rainfall data.

Microwave Radiometry Comes to the Rescue

Launch of Nimbus 5 December 1972

Electrically Scanned Microwave Radiometer: 19.35 GHz, 25km @nadir±50° cross-track scan (45 x 165 km resolution @ edges), H-pol

ESMR

Calibration Accuracy ca. 5K (@Nadir) NEΔT ca. 2K Single Frequency so Geophysical Noise was Worse.

Initial Images Had Terrible Streaks. Traced to Cross Polarized Grating Lobes. Transformed so that All Beam Positions Statistically Like Nadir BP

Coastlines and Ice Edge Blurred in MosaicsPredictive Ephemerides were Lousy.Deployment Mechanism Made the Antenna Rock 6° p-p

After all this was cleaned up we could do some science.

Typical Quick Look Image from ESMR

Land Features Obvious

Features over the Ocean with TBs too high to be explained by SST, Wind, Non-Raining Cloud

Quantitative Theory

OK, we can see rain (of some unspecified intensity) over the ocean.Can we be more quantitative?

Ed Rodgers and Merle Rao compared ESMR data with WSR-57 radar data from Miami. It looked reasonably good, but we needed a theory.

Equations for radiative transfer in rain are messy but well-known.

But how to solve them??? To get the radiance in any one direction, we need the radiance scattered in from all other directions.

Bob Curran had brought a program originally written by Ben Herman (U. Arizona) to GSFC. He gave Al Chang a copy and Al converted it for microwave.

Now what do we put into the equation of radiative transfer?

19.35 GHzGround-Based

37 GHzGround-Based

ESMR vs Miami WSR-57

Early Applications

Merle Rao, Bill Abbott and John Theon collaborated to generate an atlas of oceanic rainfall from ESMR

Quality control problems—mostly from ephemerides.

Freezing level problem

First observation of the South Atlantic Convergence Zone?

Bob Adler and Ed Rodgers looked at the energy balance of a hurricane

Results were reasonable

Beam Filling was ignored in all these applications

SSM/I

First SSM/I was launched on DMSP F-8 in 1987 19.35 22.235, 37 & 85.5 GHzDual Pol except @ 22. (85V failed early on F-8)6 Subsequent Copies

Additional Channels and Better Calibration Better Rainfall Retrievals.

Rain Algorithm Developed for Global Precipitation Climatology Project5° x 5° x 1 Month BoxesFreezing Level from 19V /22V combinationRain from Histograms of 2*TB19V – TB22V Linear Combination Mitigated Water Vapor VariabilityFit Parameters of Mixed Log-Normal Rain PDF to TB histogramChiu’s Beam Filling Correction

SSM/I Derived Rainfall Amount August 1987

What about rain over land?

High & Variable emissivity makes it difficultThis is the late-’70s view. Basis of 85.5 GHz channel on SSM/I

Nimbus-6 ESMR 37 GHz, Conical Scan, Dual Pol

Ed Rodgers and Honnappah Siddalingaiah looked at ESMR-6 over land

1978 Tropical Storm Cora Flight

Don’t try this kind of logic at home; I’m a paid professional

SSM/I land rain capability based on liquid hydrometeor scattering. Observed at 37 GHz/ Should be better at higher frequencies

Ga. Tech had a 91.65 GHz radiometer suitable for flight on the NASA CV 990

CV-990 cannot fly over/through interesting land rain (too rough)

It can fly through most oceanic precipitation

At these frequencies interesting rain (10s of mm/h) are opaque.Land surface emissivity doesn’t matter.

So we flew over ocean to test a land rain capability

Expected to see Tbs of 240 to 250K with little polarization

Rain over land can be seen via scattering by ice.

Bergeron Rain Drop Formation Process

Variability in size distribution/ layer thickness makes a quantitative relationship difficult.

Profiling Algorithms

After Launch of SSM/I Two Groups: Kummerow and Smith

Interesting Problem Attracted Many New Researchers into Rain

Two Obvious Pieces of Information in Oceanic RadiancesAttenuation of Liquid Layer Scattering by Frozen Layer

Additional Degrees of Freedom More Subtle

Kummerow Moved to Bayesian Approaches with Additional Information from Database

Algorithm Intercomparison Projects

NASA/WETNET PIPPIP-1 Aug-Nov 1987 GlobalPIP-2 1987-1993 17S to 60N (27 cases)PIP-3 1992 Global

+ Jan. & Jul. 1991 &1993

Global Precipitation Climatology Project Algorithm Intercomparison Project

AIP-1 Summer 1987 JapanAIP-2 Winter/Spring 1991 EuropeAIP-3 Austral Summer ‘92-’93 TOGA/COARE

IR and Microwave Algorithms, Physical and Empirical

Ground Truth Difficult to Impossible

IR algorithms No Physics but Lots of SamplesMicrowave Scattering Weak Physics and Very Poor SamplingMicrowave Absorption Good Physics and Very Poor Sampling (Ocean Only)Performance Depends on How a Given Scenario Relates to Strengths & Weaknesses above

Why not fly a Radar?

Suggested as early as the ‘50s (Harry Wexler)

If you think of a Radar in isolationOne you can afford is pretty much uselessA useful one costs the gross national product.

Any reasonable Radar will have a very narrow swath. No Sampling

Think of a Radar as part of a rain measurement system.

Radar is a physics probe a calibrator.

Then TRMM was launched and everything changed.