uncertainty quantification for gpm-era precipitation measurements yudong tian
DESCRIPTION
Uncertainty Quantification for GPM-era Precipitation Measurements Yudong Tian University of Maryland & NASA Goddard Space Flight Center Collaborators: Ling Tang, Rebekah Esmaili Christa Peters-Lidard, Bob Adler, George Huffman, Joe Turk, Bob Joyce, - PowerPoint PPT PresentationTRANSCRIPT
Uncertainty Quantification for GPM-era Precipitation Measurements
Yudong Tian University of Maryland & NASA Goddard Space Flight Center
Collaborators:Ling Tang, Rebekah Esmaili
Christa Peters-Lidard, Bob Adler, George Huffman, Joe Turk, Bob Joyce,
Xin Lin, Ali Behrangi, Kuo-lin Hsu, Takuji Kubota, John Eylander, Matt Sapiano, Viviana Maggioni, Emad Habib, Huan Wu
EGU, 30 April, 2014
2
Outline
What we learned from TRMM-era measurements --
“The Error Structure”
1. concepts and procedures
2. error composition
3. systematic & random errors
Applications of the Error Structure
4. scaling of errors
5. sources of errors
What error structure to expect with GPM-era measurements
3
Procedure to quantify uncertainty
Step 1: Get a reference dataset
It is all uncertain if we know no truth
If truth is available …
Truth
Measurements can be validated
Truth (mm/day)
3B42
(m
m/d
ay)
4
Truth (mm/day)
3B42
(m
m/d
ay)
missed precip (-D)
hits (H)
false precip (F)
Procedure to quantify uncertainty
Step 2: Error decomposition (Tian et al., 2009)
(total error) (hit error) (missed) (false)E = H – D + F
5
Truth (mm/day)
3B42
(m
m/d
ay)
systematic error (θ)
random error (σ)
Procedure to quantify uncertainty
Step 3: Separate systematic and random error
E= H – D + F
6
The Error Structure unifies uncertainty definition and
quantification
Uncertainty quantification = ( -D, F, θ, σ )
Total Error EE = R – Rref
Error Decomposition Scheme
(total error) (hit error) (missed) (false)E = H – D + F
(Tian et al., 2009)
Error Decomposition, Winter (DJF)
(total error) (hit error) (missed) (false)E = H –D + F
3B42
3B42RT
CMORH
PERS’N
NRL
10
Determining the Error Structure – next step, hits error
Truth (mm/day)
3B42
(m
m/d
ay)
systematic error (θ)
random error (σ)
11Ti
Xi
Ti
Xi
Hit error (H) is multiplicative (Tian et al., 2013)
ii bTaX eTX ii
Additive error model Multiplicative error model
What do α, β, and σ mean?
β
12
α: scale error (ideal: 1)β: shape error (ideal: 1) systematic error
Two parameters quantify systematic error
13
Random error σ: (ideal: 0)
TMPA 3B42 TMPA 3B42RT NOAA Radar
One parameter quantifies random error
14
Procedure of uncertainty quantification – 3 steps to the Error
Structure
Uncertainty quantification = ( -D, F, α, β, σ )
15
Scaling of errors: how Error Structure changes with
space/time scales
16
How systematic and random errors vary with space/time scales
Systematic errors(α, β)
Random errors
σ
scale error ln(α)
shape error (β)
random error (σ)
)(
stdev
eTX ii
scale error ln(α)
shape error (β)
random error (σ)
spatial scale time scale
17
18
Errors can be traced back to Level-2 retrievals (Tang, Tian & Lin, 2013; Poster #Z42 today)
AMSR-E/TMI/SSMI
SSMIS
AMSU-B
MHS
19
Summary
What we learned from TRMM-era measurements:
The Error Structure and procedures to determine it
1. concepts and procedures
2. error composition
3. separation of systematic
& random errors
4. scaling of errors
5. sources of errors
20
21
Extra slides
Error can be further decomposed:
1) Decomposition of errors:
The 3 components can be proved to be independent to
one another.
ionprecipitat False:
ionprecipitat Missed:
errorHit :
error Total:
F
D
H
E
FDHE
Over US, reliable reference data are available
CPC-UNINOAA CPC Unified Daily Gauge
Analysis 0.25○ 24 h gauges Chen et al. 2008
CPCNOAA CPC near-real-time daily
precipitation analysis 0.25○ 24 h gaugesHiggins et al.,
2000
STIV NOAA NCEP Stage IV data 4-km 1 h NEXRAD and gaugesLin and Mitchell,
2005
PRISMParameter-elevation Regression
on Idependent Slopes Model4-km monthly gauges
Daly et al. 1994; Daly et al. 2002
GSMaPGlobal Satellite Mapping of
Precipitation (GSMaP MVK+ Version 4.8.4)
0.1○ 1hIR, SSM/I, TRMM, AMSU-B, AMSR-E
Okamoto et al., 2005; Kubota et
al., 2007
3B42TRMM Multi-satellite Precipitation Analysis research product 3B42
Version 60.25○ 3 h
IR, SSM/I, TRMM, AMSU-B, AMSR-E,
gauges
Huffman et al., 2007
3B42RTTRMM Multi-satellite Precipitation Analysis Real-time experimental
product 3B42RT 0.25○ 3 h
IR, SSM/I, TRMM, AMSU-B, AMSR-E
Huffman et al., 2007
CMORPHNOAA Climate Prediction Center
(CPC) MORPHing technique 0.25○ 3 hIR, SSM/I, TRMM, AMSU-B, AMSR-E
Joyce et al., 2004
PERSIANNPrecipitation Estimation from Remotely Sensed Information
using Artificial Neural Networks0.25○ 3 h IR, TRMM
Hsu et al, 1997; Sorooshian et
al., 2000
NRLNaval Research Laboratory's
blended technique 0.25○ 3 hIR, VIS, SSM/I,
TRMM, AMSU-B, AMSR-E
Turk and Miller, 2005
ReferencesFull nameDatasetSpatial
resolutionTemporal resolution
Sensor platforms
4 ground-based datasets
6 satellite-based datasets
24
Errors are season/region-dependent over United States
US West US East
Winter
Summer
Ground-based observations are patchy and sparse
25
2626
Methodology 1: Using measurement spread Uncertainty: range of disagreement among independent
measurements of the same physical quantity
Suitable when “ground truth” is not available
Uncertainties over the globe: Winter and Spring
27
Winter
Summer
Uncertainties are season/region-dependent
Ensemble mean of global precipitation
28
Satellite-based observations enable global study of precipitation
Winter
Summer
29
Outline
1. Overview of global precipitation measurements
2. Uncertainty in precipitation
– the story of an evolving paradigm
3. Efforts to improve global precipitation observations
Semi-independent satellite-based datasets
3B42TRMM Multi-satellite Precipitation
Analysis research product 3B42 Version 6
0.25○ 3 hIR, SSM/I, TRMM,
AMSU-B, AMSR-E, gauges
Huffman et al. [2007, 2009]
3B42RTTRMM Multi-satellite Precipitation
Analysis Real-time experimental product 3B42RT
0.25○ 3 hIR, SSM/I, TRMM, AMSU-B, AMSR-E
Huffman et al. [2007, 2009]
CMORPHNOAA Climate Prediction Center
(CPC) MORPHing technique 0.25○ 3 hIR, SSM/I, TRMM, AMSU-B, AMSR-E
Joyce et al. [2004] and Janowiak et al.
[2005]
GSMaPGlobal Satellite Mapping of
Precipitation (GSMaP MVK+ Version 4.8.4)
0.1○ 1hIR, SSM/I, TRMM, AMSU-B, AMSR-E
Okamoto et al. [2005] and Kubota et al.
[2007]
PERSIANNPrecipitation Estimation from
Remotely Sensed Information using Artificial Neural Networks
0.25○ 3 h IR, TRMMHsu et al. [1997] and
Sorooshian et al. [2000]
NRLNaval Research Laboratory's
blended technique 0.25○ 3 hIR, VIS, SSM/I, TRMM,
AMSU-B, AMSR-ETurk and Miller
[2005]
ReferencesFull nameDatasetSpatial
res.Temporal
res.Sensor platforms
Several satellite-based, global, high-resolution datasets are available
3B42TRMM Multi-satellite Precipitation
Analysis research product 3B42 Version 6
0.25○ 3 hIR, SSM/I, TRMM,
AMSU-B, AMSR-E, gauges
Huffman et al. [2007, 2009]
3B42RTTRMM Multi-satellite Precipitation
Analysis Real-time experimental product 3B42RT
0.25○ 3 hIR, SSM/I, TRMM, AMSU-B, AMSR-E
Huffman et al. [2007, 2009]
CMORPHNOAA Climate Prediction Center
(CPC) MORPHing technique 0.25○ 3 hIR, SSM/I, TRMM, AMSU-B, AMSR-E
Joyce et al. [2004] and Janowiak et al.
[2005]
GSMaPGlobal Satellite Mapping of
Precipitation (GSMaP MVK+ Version 4.8.4)
0.1○ 1hIR, SSM/I, TRMM, AMSU-B, AMSR-E
Okamoto et al. [2005] and Kubota et al.
[2007]
PERSIANNPrecipitation Estimation from
Remotely Sensed Information using Artificial Neural Networks
0.25○ 3 h IR, TRMMHsu et al. [1997] and
Sorooshian et al. [2000]
NRLNaval Research Laboratory's
blended technique 0.25○ 3 hIR, VIS, SSM/I, TRMM,
AMSU-B, AMSR-ETurk and Miller
[2005]
ReferencesFull nameDatasetSpatial
res.Temporal
res.Sensor platforms
Uncertainties also depend on rain rate – light rain, high uncertainty
33
Basic characteristics of satellite-based precipitation uncertainty
•Depend on rain rate: ~25% heavy rain, ~100% light rain
• Depend on region: high latitude, snow-cover, > 100%
• Depend on season: winter > summer
34
Methodology 2: when ground truth is available (e.g., over U.S.)
E = R - R0
Error = Measurement - Truth
35
Methodology 1: Using measurement spread when no ground truth
Methodology 2: when ground truth is available (e.g., over U.S.)
E = R - R0
Error = Measurement - Truth
Methodology 3: Error decomposition
(total error) (hit bias) (missed) (false) E = H – D + F
36
Outline
1. Overview of global precipitation measurements
2. Uncertainty in precipitation
– the story of an evolving paradigm
3. Efforts to improve global precipitation observations
Precipitation over snow-covered surfaces is highly uncertain
37
Time series of error components. They sometimes enhance, sometimes cancel each other
West East
ionprecipitat False:
ionprecipitat Missed:
biasHit :
bias Total:
F
D
H
E
FDHE
Uncertainty determines reliability of information
What we do not know affects what we know
Information
KnownsKnowledge
Signal Deterministic
Systematic errors
yUncertaint
1~yReliabilit
Uncertainty
UnknownsIgnorance Noise StochasticRandom errors
What about NWP forecasts?
41
The multiplicative error model predicts better
Additive Model Multiplicative Model
Model-predicted measurements
Actual measurementsComparison of data distributions