Ground Validation for GPM Microwave Imager

1
Ground Validation for GPM Microwave Imager
Algorithm Development
Bill Olson, Mircea Grecu, Wei-Kuo Tao, Chung-Lin
Shie, Steve Lang, Wes Berg, Chris Kummerow, and
Tristan lEcuyer
TB
2
TMI Algorithm Database Constructed from
PR/TMI Estimates
PR
TMI
reflectivity profiles
retrieved precipitation profile
TMI footprints
3
PR/TMI-based TMI Estimation Algorithm
Each estimate is a weighted average of database
profiles
weighting prob (Ri)
prob (R1)
prob (R3)
prob (R2)

database of precipitation profiles 106
4
PR/TMI-based TMI Estimation Algorithm
weighting prob (Ri)
prob TB(Ri) TBobserved
TB(R1)
TB(R2)
TB(R3)

database of precipitation profiles
5
PR/TMI-based TMI Estimation Algorithm
weighting prob (Ri)
statisical
prob TB(Ri) TBobserved
physical
database of precipitation profiles
6
Algorithm Validation and Improvement
investigate with GV precipitation process
sites
update algorithm
improve forward model and database consistency
7
Bias Identification - Product Validation Sites
use spaceborne radar (DPR) as reference
supplement with US National Network of radars
(product validation sites)
8
Algorithm Diagnostics
evaluate forward model physics -
precipitation size distributions -
non-precipitating cloud - EM properties
particularly ice mixed phase - radiative
transfer evaluate regime-dependent
precipitation statistical characteristics -
relative number of system types - precipitation
geometry - factors affecting precipitation
efficiency
9
Forward Model Physics- Aircraft Studies at
Precipitation Process Site
utilize multi-channel active and passive
sensor observations of narrow beam
active 10 - 89 GHz passive 10 - 300 GHz
find forward model physical parameterizations
that are consistent with all observations
10
Airborne Cloud Radar and Microwave Radiometer
Wakasa Bay Field Campaign
On NASA P3 Aircraft Airborne Cloud Radar (94
GHz) Precip Radar-2 (14, 35 GHz)
Polarimetric Scanning Radiometer (10.7,
18.7, 37.0, 89.0 GHz) Millimeter-wave Imaging
Radiometer (90 - 340 GHz) AMMR (21, 37 GHz
- upward looking)
11
Forward Model Physics- GMI Simulator at
Precipitation Process Sites
12
Flow Diagram
Simulator
Convolve Radiances with GMI Response
Functions Add Noise
13
Prototype GMI Simulator
14
Forward Model Parameterizations- DPR as Roving
Precipitation Process Site
15
Precipitation Statistical
Characteristics - Bias-Free TMI Algorithm
Applied to July 2000 Data
16
Precipitation Statistical Characteristics -
Long-Term Monitoring
use DPR plus ancillary data to identify
regime- dependent biases. direct product
validation sites and precipitation
process sites can also play a role. what
sub-populations of rain system types,
precipitation structures, etc., occur more
frequently in a given regime?
July 2000
January 1998
17
Latent Heating
Each estimate is a weighted average of database
profiles
weighting prob (LHi)
prob TB(LHi) TBobserved
LH
database of precipitation profiles
18
Validation of Cloud-Resolving Models- Measure
Large-Scale Forcing at Process Sites
drive CRMs with measured advective
tendencies
simulate GV radar and satellite
observations are observations reproduced in
a statistical sense?
19
SCSMEX Rawinsonde Network and Polygons

for NESA Q1, Q2, ltQ2gt Lv (PRawin -
E) Johnson and Ciesielski, (2002 person. comm.)
20
SCSMEX Obs vs. GCE- 2241 UTC 20 May 1998
21
GCE Model vs. TMI Observations (SCSMEX 30-Day
Period)
-histograms of observed and modeled 19 GHz rain
fraction index vs. 85.5 GHz scattering index.
22
Latent Heating Estimation- PR/TMI Database
(using PR/TMI Database)
Johnson and Ciesielski (2002)
23
Concluding Remarks
identify regime-dependent biases with - GMI
vs DPR/GMI difference maps - enhanced national
radar network validate physics of algorithm
using - multi-sensor aircraft observations -
radar-driven GMI simulator validate
statistics of rain system structure using -
DPR/GMI observed distributions - enhanced
national radar network latent heating -
validate cloud-resolving models - no global
reference currently
24
Regime-Dependent Precipitation Characteristics -
Long-Term Monitoring
removal of rain bias using column water vapor
direct product validation sites and
precipitation process sites can also play a
role
25
Latent Heating Profile Assignment to Database
Need cloud-resolving models to associate heating
profiles.