The 19th International TOVS Study Conference, 26 March — April 1, 2014
Satellite Radiance Assimilation in HWRF
Xiaolei Zou
Department of Earth, Ocean and Atmospheric Science
Florida State University
Collaborators:
Fuzhong Weng, Banglin Zhang, Lin Lin,
Zhengkun Qin and Vijay Tallapragada
1
Outline
• A Comparison of Data assimilation and Forecast Results
with Two Different Model Top Altitudes
 Assimilation of upper-level channels
 Differences in storm Debbie’s track forecast
• Impact of ATMS radiance assimilation on hurricane
track and intensity forecast
 A unique feature of ATMS
 A consistent positive impacts
• Impact of NOAA-15 AMSU-A Data on QPFs and Its
Implications for Three-Orbit Constellation
 11p.02
• Current and Future Plan
HWRF Domain Sizes for Tropical Storm Debby
ghost D2
Three changes
made to HWRF:
inner nest
middle nest
3X domain
Cold start -> warm start
Model top 50 -> 0.5 hPa
Vortex initialization
parent
domain
The Importance of Upper Atmosphere on TCs
Environmental factors involve atmospheric conditions in the upper
troposphere and the stratosphere:
 Steering flow (Carr and Elsberry, 1990)
 Vertical wind shear
(Davis and Bosart, 2006; DeMaria, 1996)
 Approaching upper-level trough
(Leroux et al., 2013)
 Eddy angular momentum flux convergence
(Pfeffer and Challa, 1981; Bosart et al. 2000)
 Stratospheric cooling (Ramsay, 2013)
 Quasi-biennial oscillation in the stratosphere
(Chan, 1995)
Modeling these environmental factors affecting track and TC
intensification requires a sufficiently high model top.
HIRS Channels
Pressure (hPa)
Pressure (hPa)
AMSU-A Channels
Weighting Function
Weighting Function
upper-level
channels
AIRS Channels (281)
middle-level channels
low-level channels
Weighting Function
These 281 channels
are selected for data
assimilation in the
GSI/HWRF system.
Weighting Function
The pressure at which
WF reaches a maximum is indicated in color.
Weighting Function
The Best Track of Four 2012 Atlantic Landfall
Hurricanes Selected for This Study
Sandy
Beryl
Debby
Isaac
7
Track Predictions of the 2012 Operational HWRF
• The operational HWRF model produced an eastward propagating tracks
while Debby moved northeastward when model forecasts were initialized
before June 25, 2012
• The operational HWRF model produced reasonably good track forecasts
after June 25 and afterward.
The track prediction of Debby before June 25, 2012 was a major challenge.
500-hPa Geopotential and Wind Vector Distributions
1200 UTC
June 23 ,2012
1800 UTC
June 23, 2012
0000 UTC
June 24 ,2012
0600 UTC
June 24, 2012
1200 UTC
June 24, 2012
1800 UTC
June 24, 2012
1200 UTC
Low
1800 UTC
Low
Channel Number
0600 UTC
Low
High
High
Channel Number
High
Channel
Dependence
and Daily
Variations of
ATMS Data
Counts
Assimilated
for Modeling
Tropical
Storm
Debby
June 2012
Data Count (x103)
Data Count (x103)
Data Count (x103)
O-B and O-A Distributions of ATMS Upper-Level in L61
O-B
Ch12
Ch11
Ch10
Ch9
O-A
Ch12
Ch11
Ch10
Ch9
Debby at 1800 UTC June 24, 2012
AIRS Channel Dependence of Data Count Assimilated
During Tropical Storm Debby
peak WF
Pressure (hPa)
peak WF
Pressure (hPa)
Data Count (x103)
lower model top
Data Count (x103)
Higher model top
Channel Number
More upperlevel channel
data are
assimilated in
L61 with a
higher model
top (0.5 hPa)
than L43 whose
model top is
located around
50 hPa.
Standard deviation (K)
Before
After
wavelength
Wavelength (um-1)
Higher
Model top
Before
After
wavelength
Wavelength (um-1)
Lower
Model top
Standard deviation (K)
Model Fit to AIRS Observations before and after DA
Channel number
The std. of O-A is greater than that of O-B for upper-level channels in L43.
Comparison of Track Forecasts between L61 and L43
Track Error (km)
mean of L43
mean of L61
std. of L43
std. of L61
L61
Track Error (km)
L43
Track Error (km)
?



?


2518-2712
June 2012
Mean Forecast Errors for Four 2012 Atlantic Hurricanes
Track Error (km)
Impact of Model Top Altitude on Track and Intensity Forecasts
Lower model top:
mean
std.
Time (hour)
SLP Error (hPa)
Higher model top:
mean
std.
Time (hour)
LWP (AMSU-A channels 1-2)
IWP (MHS channels 1-2)
1.70
1.40
1.10
0.80
0.50
0.25
0.20
0.15
0.11
0.09
0.07
0.05
0.03
0.01
Tb (GOES-12, 10.7µm channel)
(kg m-2)
LWP/IWP Collocation
clear-sky
warm
cloud
ice cloud
mixedphase
cloud
NOAA-18, 1441 UTC to 2303 UTC on May 22, 2008
16
AMSU-A and MHS FOVs
AMSU-A FOV
Scanline
14
15
16
17
3
2
1
1
40 41 42 43 44 45 46 47 48 49 50 51
MHS FOV
An inconsistent FOV distribution between AMSU-A and MHS
channels makes MHS cloud detection extremely challenging.
17
The ATMS FOV Distribution along a Scanline
Channels 1-2
1
Channels 17-22
43 44 45 46 47 48 49 50 51 52 53 54
ATMS FOV
1
Channels 3-16
Channels 17-22
43 44 45 46 47 48 49 50 51 52 53 54
ATMS FOV
18
A consistent FOV distribution between temperature and humidity
channels on ATMS makes the cloud detection easy to implement.
O-B and O-A Data Counts for Hurricane Isaac
ATMS Channel 9
O-A (K)
O-B (K)
ATMS Channel 6
Scan Position (FOV)
Scan Position (FOV)
19
Impacts on Intensity Forecast Hurricane Isaac
NSAT+ATMS
SAT
SAT+ATMS
vmax (kts)
vmax (kts)
NSAT
22
23
24
25
26
27
28
29 August 2012
20
Impacts of ATMS Data Assimilation on
Track Forecast of Hurricane Sandy
SAT
23
24
SAT+ATMS
25
26
27
28
29
October 2012
21
Hurricane Sandy
(PV at 200 hPa)
84-h Forecast
without ATMS
84-h Forecast
with ATMS
NCEP GFS analysis
0000 UTC October 30
22
Mean Forecast Errors for Four 2012 Atlantic Hurricanes
Track error (nm)
pc (hPa)
pc (hPa)
vmax error (kts)
vmax error (kts)
Data Counts
Track error (nm)
Impact of ATMS Data Assimilation
Time (hour)
CONV
CONV+ATMS
Time (hour)
SAT
SAT+ATMS
23
Current and Future Plan
• ATMS radiance assimilation (further refinement)
• Model top&vertical levels (further refinement)
• GOES imager radiance assimilation for TCs (on going)
• AMSU three orbits impact assessment (on going)
• CrIS/VIIRS radiance assimilation (on going)
• SSMIS/AMSR2 imager radiance assimilation (on going)
• Combined AMSU-A/MHS data stream (on going)
• Hurricane initialization using satellite data (on going)
Three Key Components for Satellite Data Assimilation
Bias Correction, Data Thinning, Quality Control
More details can be found in
Zou, X., F. Weng, Q. Shi, B. Zhang, C. Wu and Z. Qin, 2013a: Satellite data
assimilation in NWP models. Part III: Impacts of model top on radiance
assimilation in HWRF. J. Atmos. Sci., (submitted)
Zou, X., F. Weng, B. Zhang, L. Lin, Z. Qin and V. Tallapragada, 2013b: Impact
of ATMS radiance data assimilation on hurricane track and intensity
forecasts using HWRF. J. Geophys. Res., 118, 11,558-11,576.
Weng, F., X. Zou, X. Wang, S. Yang, and M. D. Goldberg, 2012: Introduction
to Suomi NPP ATMS for NWP and tropical cyclone applications. J.
Geophy. Res., 117, D19112, 14pp, doi:10.1029/2012JD018144.
Weng, F., X. Zou, and Z. Qin, 2014: Impact of NOAA-15 AMSU-A data on
QPFs and its implications for three-orbit constellation. Mon. Wea. Rev.,
(to be submitted)
25
Acknowledgement
This work was jointly supported by NSF, NOAA GOES-R
risk reduction program and JPSS Proving Ground Program.