Title: Geosynchronous Microwave Sounding of Precipitation Parameters at Convective Scales
1Geosynchronous Microwave Sounding of
Precipitation Parameters at Convective Scales
David H. Staelin and Chinnawat Surussavadee Prese
nted at the Third Workshop of the International
Precipitation Working Group Melbourne,
Australia, 23-27 October, 2006
Staelin and Surussavadee October 2006
2MM5 vs. AMSU TBs (K)
SurussavadeeStaelin
3(No Transcript)
4F(l) Ice Scattering Model
F(l) Model Sphere ice density F(l)
Fluffy ice, Mie
Question Match F(l) to DDSCAT ss or to
back-scatter?
Answer Total ss works better because coldest
pixels scatter many times, losing sense of
direction
SurussavadeeStaelin
5AMSU Retrievals vs. AMSR-E over Ocean
SurussavadeeStaelin Chadarong McLaughlin
6AMSU Retrievals vs. AMSR-E over Land
Africa
Canada
0017 UTC
0007 UTC
844 UTC
726 UTC
Great Lakes
Great Lakes
AMSU (N16)
AMSR-E (Goddard)
AMSR-E (Goddard)
AMSU (N16)
SurussavadeeStaelin Chadarong McLaughlin
7AMSU Retrievals over U.S. Midwest
SurussavadeeStaelin Chadarong McLaughlin
8Instrument Precipitation-Rate Comparisons
9Geo-Microwave Sensor Concepts
Even a 2-m dish Can be integrated on GOES DTrms
? 0.5K (400 GHz, t 0.04s) Weight 50 kg, 130
watts
GeoSTAR
GEM
Nodding Subreflector
Sketch by MIT Lincoln Laboratory
Sketch by Ball
Sketch by JPL
Staelin and Surussavadee October 2006
10Image Sharpening
Original 5-km image
Sharpened pattern
Original pattern
260 oK 220 180
30 km
22.1 km
Requires Nyquist sampling G(q) FFT W(fq) To
minimize MSE Noise increases with
sharpening
Blurred 30-km resolution with noise
Sharpened 30-km res.
SurussavadeeStaelin
11GEM Channel Selection Issues
118 GHz (O2)
425 GHz (O2)
Clear-air Incremental weighting functions (IWF)
for temperature and humidity vs. offset (MHz)
from line center (from Klein and Gasiewski,
2000). Peaks of GEM/GeoSTAR weighting
functions analyzed here. Note that high humidity
can preclude penetration below 2 km at 118, 166
GHz.
183 GHz (H2O)
380/340 GHz (H2O)
Staelin and Surussavadee October 2006
12Geo-Microwave Precipitation Retrievals
Training (NCEP/MM5, 122 global storms)
Simulations or observations
1
8
118 GHz (4 O2 Channels) 166-183 GHz (4 H2O
Channels) 380 GHz (4 H2O
Channels) 425 GHz (4 O2
Channels) Land/sea, elevation
Yes
Neural Net 2
8
Neural Net 1 (for categorization)
Rain rate estimate R (mm/h)
R gt 8?
8
R (mm/h)
Neural Net 3
No
34
8
Nets 2 and 3 were trained with different R values
(NN?1.3 better matches MM5, retrieval colors)
2
Staelin and Surussavadee October 2006
13Rain Rate and Water Path Retrievals
1.2-m GEM antenna 118/183/380/425-GHz bands
Chinese Front, June 22, 2003
SurussavadeeStaelin
14Instrument Options Evaluated
Spatial Resolution (km) at Nadir
Frequency A B C
D E F G H I J
K Band (GHz) 1.2 1.2S 2 2S 1.2 1200 600 90
0 600 300 300 52.8 25 50 50 25 50 118.75 97
62 58 38 97 25 50 50 166 69 44 42 28 69 25
183.31 63 42 38 26 63 25 380.2 30 22 18 14 4
24.76 27 20 17 12
Dmax 2.8 m for U array Dmax 5.6 m for U
array
Aperture synthesis systems must average longer
(x10 for bandwidth, x4 for
4-bands, x10 for area coverage, ?10 for receiver
noise)
Staelin and Surussavadee October 2006
15Representative Instrument Options
Surface precipitation rate images for four
instrument options (mm/h)
SurussavadeeStaelin
16Effects of Beam Sharpening
25km
Front
Warm rain
SurussavadeeStaelin
17MM5 vs. 1.2-m GEM Rain Rate (mm/h)
MM5 with 15-km resolution
SurussavadeeStaelin
182-m vs. 1.2-m Precipitation Rate (mm/h)
SurussavadeeStaelin
19MM5 vs. 1.2-m GEM Rain Rate (mm/h)
MM5 with 15-km resolution
SurussavadeeStaelin
202-m vs. 1.2-m Precipitation Rate (mm/h)
SurussavadeeStaelin
21Summary and Conclusions
F(l) radiative transfer matches MM5 to AMSU
observations well GEM and AMSU retrievals are
feasible for
- rain and snow
(mm/h) over land and sea (not yet over ice, snow)
- water paths for rain
water, snow, and graupel (mm) Both 1.2- and 2-m
micro-scanned antennas can be integrated on GOES
Image sharpening (optional processing) yields
antenna resolution ?1.3 The simplest aperture
synthesis systems comparable to a 1.2-meter GEM
antenna use at least 900 receivers in two bands,
are ? 5.6m across, and cannot repeat so
frequently as GEM GEM should be part of PMM --
unique for convective-scale evolution
Staelin and Surussavadee October 2006