Hydrometeorological Testbed (HMT) Ensemble QPF
Transcript of Hydrometeorological Testbed (HMT) Ensemble QPF
Hydrometeorological Testbed(HMT) Ensemble QPF
Isidora Jankov1, Steve Albers1, Huiling Yuan3, Zoltan Toth2, Tim Schneider4, Allen White4 and Marty Ralph4
1Cooperative Institute for Research in the Atmosphere (CIRA),Colorado State University, Fort Collins, CO
Affiliated with NOAA/ESRL/ Global Systems Division2NOAA/ESRL/Global Systems Division
3Cooperative Institute for Research in Environmental Sciences (CIRES)University of Colorado, Boulder, CO
Affiliated with NOAA/ESRL/Global Systems Division4 NOAA/ESRL/Physical Sciences Division
12 April, 2011 Belgrade
Outline
• HMT overview• Role of ESRL/GSD in HMT• Some 2009-2010 season results• Potential future changes in the ensemble
design
Goal is to improve forecasts of rain and snow and associated hydrologyUses local‐state‐federal, and private‐public‐academic partnerships
HMT WEST ‐ Cool Season
HMT EAST – All Season, including Hurricane Landfall
HMT CENTRAL –Warm Season
Benefits: Accelerates improvements in QPF and flood forecasting, with impacts on transportation, ecosystems, emergency management, flood control and water spply. Science and field tests will advise on how best to fill gaps in observational and modeling systems.
HMT Overview
o During the winter season significant precipitation events in California are often caused by land‐falling “atmospheric rivers” associated with extra tropical cyclones in the Pacific.
o Atmospheric rivers are elongated regions of high values of vertically integrated water vapor over the Pacific and Atlantic oceans that extend from the tropics and subtropics into the extratropics and are readily identifiable using SSM/I.
o Due to the terrain steepness and soil characteristics in the area, a high risk of flooding and landslides is often associated with these events.
ATMOSPHERIC RIVERS
ESRL/GSD group roles:• Design of the LA ensemble• Provide real time ensemble precipitation forecasts•Collaborate with colleagues in addressing various scientific questions
ESRL/GSD HMT Role
EXPERIMENT DESIGN 2009‐2010
Nested domain:• Outer/inner nest grid spacing 9 and 3 km, respectively.• 6‐h cycles, 120hr forecasts for the outer nest and 12hr forecasts for the inner nest • 9 members (WRF‐NMM and ARW)• Mixed models, physics & perturbed boundary conditions from NCEP Global Ensemble• 2010‐2011 season adding initial condition perturbations and NMMB to the ensemble will be explored?
QPF
7
Example of 24-h QPFFor 9 ensemble members for the9-km resolution domain.
HMT QPF and PQPF24‐hr PQPF
0.1 in.
1 in.
2 in.
48‐hr forecast starting at 12 UTC, 18 January 2010
Probabilistic QuantitativePrecipitation ForecastsFour rerun HMT-West-2006 cases:IOP1, 4, 10, 12
Cross-validation over the ARB
Reliability curves are improved (red line along the diagonal) for the thresholds 1-25 mm/6-h
Internal histograms:Frequencies changed
0 10 20 30 40 50 60 70 80 90 1000
10
20
30
40
50
60
70
80
90
100
forecast probability (%)
obse
rved
freq
uenc
y (%
)
a) 1 mm/6 h
ENSNewPerfectENSNew
0 10 20 30 40 50 60 70 80 90 1000
10
20
30
40
50
60
70
80
90
100
forecast probability (%)
obse
rved
freq
uenc
y (%
)
b) 5 mm/6 h
ENSNewPerfectENSNew
0 10 20 30 40 50 60 70 80 90 1000
10
20
30
40
50
60
70
80
90
100
forecast probability (%)
obse
rved
freq
uenc
y (%
)
c) 10 mm/6 h
ENSNewPerfectENSNew
0 10 20 30 40 50 60 70 80 90 1000
10
20
30
40
50
60
70
80
90
100
forecast probability (%)
obse
rved
freq
uenc
y (%
)
d) 15 mm/6 h
ENSNewPerfectENSNew
0 10 20 30 40 50 60 70 80 90 1000
10
20
30
40
50
60
70
80
90
100
forecast probability (%)
obse
rved
freq
uenc
y (%
)e) 20 mm/6 h
ENSNewPerfectENSNew
0 10 20 30 40 50 60 70 80 90 1000
10
20
30
40
50
60
70
80
90
100
forecast probability (%)
obse
rved
freq
uenc
y (%
)
f) 25 mm/6 h
ENSNewPerfectENSNew
Calibration of PQPF
Yuan et al. 2008, JHM
Real-time QPF verification for HMT-West
OAR/ESRL/GSD/Forecast Applications Branch
11
Runoff experiments
Validation events: 5 IOPs
The distributed hydrologic model:
Two‐Dimensional Runoff Erosion and Export (TREX) model
100 m2 pixel
Ensemble created from 12 combinations of hydro model parameter perturbations using inputs from:
1) 0‐6 h ensemble mean QPF, 3‐km
2) Stage IV QPE, ~ 4 km
3) CNRFC QPF day 1‐ day 3 forecasts, ~ 4 km
By Yuan, H.,J. J. Gourley, P. J. Schultz, J. A. McGinley, Z. Flamig, C.J. Anderson
Average skill scores for streamflow simulations from 14 IOPs
Ranked Probability Score (RPS) is computed for using the 0‐6 h ensemble mean QPF, 6‐h Stage IV, CNRFC day1 to day 3 forecasts with 14 IOPs during three winter seasoons.
Smaller RPS is better. The high‐reso ensemble QPF is the best in terms of peak and volume, and is worse than Stage IV input in the peak timing.
ENSEMBLE OUTPUT DISPLAYThe ensemble output has been available
To WFOs through ALPS http://laps.noaa.gov/forecasts/
TPW (cm) PW Flux (q x v)
Precipitation Mean
PrecipitationSpread
Cloud ice and liquid Xsect
Xsect Reflectivity 06 Oct. 2010 18UTC
LAPS GEP1 GEP2 GEP3 GEP4 GEP5 GEP6
00 hr
03 hr
06 hr
The difference between the local model and regional ensemble mean is our ‘local
perturbation’ 15
00Z 06Z
LAM Forecast
Regional Ensemble Mean
Global Model Analysis interpolated on LAM
grid
Perturbation
Legend
Forecast Time
Cycling Initial Perturbations
P1
P2
P3
00hr
03hr
06hr
Wind SpeedJuly 30 2010 00UTC
LAPS CYC NOCYC
17
Cloud CoverageJuly 30 2010 00UTC
00hr
03hr
06hr
LAPS CYC NOCYC
0
0.5
1
1.5
2
2.5
Lin WSM6 Thompson SchultzM
ixin
g ra
tio x
10^-
3 (k
g/kg
)
cloud water cloud ice rain snow graupel
MICROPHYSICAL ASSPECTS: WATER SUBSTANCE PARTITION
SIMULATIONS• 5 “atmospheric river” events• 3km grid spacing• 4 different Microphysics
(Lin, WSM6, Thompson & Schultz)
Comparison of synthetic reflectivity to S‐band Radar Data (Obs. vs. Thompson)
Integration domain
Evaluation and Comparison of Microphysical Algorithms in WRF-ARW Model Simulations of Atmospheric River Events Affecting the California Coast 2009:JHM, 10,847-870
The model showed a tendency to overestimate the upslope wind component duration and intensity as well as moisture content.
Before
After
Observations
~50% decrease in precip. overestimation from model run using Lin scheme after adjusting some parameters (e.g. accretion of snow by graupel and lowering thethreshold for conversion of snow tograupel)
Possible use of synthetic satellite imagery, as an additional way to indirectly evaluate the performance of various microphysical schemes, was evaluated.
Observations‐GOES‐10 10.7µm
WSM6
Schultz
24-hr forecast valid at 31 Dec. 2005 at 12 UTC
0
0.2
0.4
0.6
0.8
1
220 230 240 250 260 270 280
Brightness Temperature (K)
POD
0
0.2
0.4
0.6
0.8
290
Brightness Temperature (K)
TS
0
0.2
0.4
0.6
0.8
1
220 230 240 250 260 270 280
Brightness Temperature (K)
FAR
0123456
220 230 240 250 260 270 280
Brightness Temperature (K)
Bias
Lin WSM6 Thompson Schultz Morrison
An Evaluation of Five WRF-ARW Microphysics Schemes Using Synthetic GOES Imagery for an Atmospheric River Event Affecting the California Coast. (In press ) JHM.
Questions?