Title: Evaluating Uncertainties in GPM Oceanic Precipitation Retrievals
1Phantom Precipitation and Other Problems in TRMM
Products
Matthew Miller and Sandra Yuter Department of
Marine, Earth, and Atmospheric Sciences North
Carolina State University Raleigh, NC USA
2TRMM 3 hour Global Rainfall
The 23-year (1979-2001) annual mean precipitation
(mm day-1) based on GPCP Version 2 includes
satellite, gauge, and radar (Adler et al. 2003)
Is data set adequate for purpose? What are
strengths and weaknesses?
3Motivations for Diagnosing TRMM Error
Characteristics
- Identify geographic regions and seasons where
existing algorithm physics may be incorrect or
incomplete - Region by region uncertainties aid application of
TRMM data sets into numerical models,
forecasting, and climate diagnostics - TRMM TMI algorithm similar to that used with
other passive microwave satellites
4Relative Comparison Between Independent
ProductsPrecipitation Radar (PR) vs. TRMM
Microwave Imager (TMI)
- Important for regions without surface-based rain
estimates - Agreement ? Truth
- But, lack of agreement implies problem with one
or both algorithms - Check physical consistency with empirical data
5Comparison of TRMM Instantaneous Rain Rate
Products
W Pacific Near Kwajalein
47 days 16 June 1 Aug 2001
Version 6
Version 5
PR TMI
PR TMI
Number of Pixels
Number of Pixels
Log10(R)
Log10(R)
1500 km x 1500 km ocean region
Both Unimodal
TMI Bimodal
6Version 6
16 June 1 Aug 2001
SW Pacific off Australia
Bay of Bengal
PR TMI
PR TMI
Number of Pixels
Number of Pixels
Log10(R)
Log10(R)
716 June 1 August 2001
PinkTMI bimodal BlueTMI unimodal GreenTMI
strongly skewed
81 January 16 February 2002
PinkTMI bimodal BlueTMI unimodal GreenTMI
strongly skewed
9V6 Global PDF, PR rescaled to 85 GHz scale
LAND
PDF ()
Rain rate (mm/hr)
Stout and Kwiatkowski
10Scattered Shallow Precip Near Strait of
Gibraltar
PR Rain Rates
TMI surface type
LAND-Spain
OCEAN
LAND Morocco
COAST
11Scattered Shallow Precip Near Strait of
Gibraltar
PR Rain Rates
TMI surface type
TMI rain rates
12Widespread Deep Convection Hurricane Ophelia
TMI surface type
TMI surface type
PR Rain Rates
USA North Carolina
Atlantic Ocean
13Widespread Deep Convection Hurricane Ophelia
TMI surface type
TMI rain rates
PR Rain Rates
14Widespread Deep Convection Hurricane Ophelia
TMI surface type
TMI rain rates
PR Rain Rates
10 mm/hr
20 mm/hr
15Widespread Deep Convection Hurricane Ophelia
TMI rain rates
Coastal S-band Z
16Unphysical Ice in Rain Layer Over Land
TMI Precip Ice Profile 1.5 2 km altitude
TMI Precip Ice Profile 0 0.5 km altitude
TMI rain rates
0?C height 4.5 km
17Atlantic Phantom Precip
TMI rain rates
PR Rain Rates
18Atlantic Phantom Precip
TMI rain rates
Coastal S-band Z
19Histogram of Phantom Precip Off East Coast
Phantom Precip Rain Rate Histogram
Bounded area
20Atlantic Phantom Precip
- IR indicates warm top low-level clouds
- Sounding derived cloud top height 3100m
- 0 height at 3300m
GOES 12 IR
21Erroneous Cloud Ice
14km TMI Cloud Ice
- Upper Air Sounding Observed Cloud Top at 3.1km
- TMI cloud ice from 3.5-14km
- Indicative of non-physical hydrometeor profile
22North Taiwan Coast1 Feb 2000 Case from Berg et
al. 2006
- Studied TMI/PR differences in rainfall detection
- Found large differences in frequency of detected
precipitation between TMI and PR over East China
Sea - Hypothesized TMI observed unusually high LWC
cloud with relatively high emission, but low
reflectivity (18dBZ) due to high aerosol content
23North Taiwan Coast1 Feb 2000 Case from Berg et
al. 2006
TMI rain rates
PR Rain Rates
24Histogram of Phantom Precip Off North Taiwan
Coast1 Feb 2000 Case from Berg et al. 2006
Coastal S-band Z
Phantom Precip Rain Rate Histogram
Bounded area
Radar data courtesy of T.-C. Chen
25Erroneous Cloud Ice
14km TMI Cloud Ice
- Upper Air Sounding Observed Cloud Top at 3.9km
- TMI cloud ice from 4-14km
- Indicative of non-physical hydrometeor profile
26Conclusions I
- About half of regional ocean PDFs degraded in TMI
V6 compared to V5 - Implausible bimodal PDFs of rainrate
- Most commonly occur during local summer and heavy
rainfall areas - Serious problems with TMI rainfall estimation in
coastal regions - Shallow precipitation missing over coast/land
- (no ice scattering or scattering too weak)
- Widespread deep precipitation
- Rain rate discontinuity at coast/ocean boundary
(more obvious with heavier precipitation)
27Conclusions II
- TMI phantom precipitation over ocean
- Does not appear in either PR or more sensitive
coastal radar - Occurs in stratus clouds under stable layer
- Phantom rain rates up to 2.3 mm/hr, modes vary
with case (0.6 to 1.2 mm/hr) - TMI hydrometeor profile does not physically
represent the actual situation - Unusually high LWC cloud (Berg et al. 2006) would
have Z values of 18 dBZ - Database issues more likely
28Interpretation
- Non-physical TMI hydrometeor profiles appear to
be symptom of overreaching in database - Occurs in different meteorological situations
- Closest profile in GPROF database does not
pertain to actual situation - Could use to determine locations and of highly
uncertain rain rates - Adequacy for purpose of precipitation retrieval
29(No Transcript)
30Gulf Phantom Precip
TMI rain rates
PR Rain Rates
31Gulf Phantom Precip
TMI rain rates
Coastal S-band Z
32Histogram of phantom precip off Gulf coast
Phantom Precip Rain Rate Histogram
Bounded area
33Gulf Phantom Precip
- IR indicates warm top low-level clouds
- Sounding derived cloud top height 4100m
- 0 height at 4200m
GOES 8 IR
34Erroneous Cloud Ice
8km TMI Cloud Ice
- Upper Air Sounding Observed Cloud Top at 4.1km
- TMI cloud ice from 6-18km
- Indicative of non-physical hydrometeor profile
35Passive Microwave measurements are volumetric
Emission channels sensitive to rain layer
Darker color more brightness temperature
signature
Houze et al. 1989
36Passive Microwave measurements are volumetric
Scattering channels sensitive to ice layer
Darker color more brightness temperature
signature
37Florida Case Minimum Surface Temp 11 ?C
TMI Precip Ice Profile 0 0.5 km altitude
TMI rain rates
38TRMM Satellite Sensors
- Precipitation Radar (PR) 13.8 GHz
- Swath width 245 km
- Spatial resolution 4.9 km
- Minimum sensitivity 18 dBZ
- TRMM Microwave Imager (TMI)
- five passive microwave channels
- Swath width 872 km
- Rain Layer Emission 19 GHz channel, 35 km pixels
- Ice Layer Scattering 85 GHz channel, 7.7 km
pixels
Over Ocean
Over Land and Coast
39TRMM Product Releases
- TRMM satellite launchNovember 1997, Version 1
- Version 5 (V5)November 1999
- Version 6 (V6)April 2004
40NEXRAD Scan Strategy
41ROIG Sounding - 1 Feb 2000 12Z