SLIDE 1 Direct Assimilation of Radar Reflectivity Data using a Convective scale EnKF System
Jingyao Luo1, Hong Li1, Ming Xue2,3 and Youngsun Jung2
1Shanghai Typhoon Institute, China 2Center for Analysis and Prediction of Storms and 3School of
Meteorology University of Oklahoma,USA January 24, 2019
SLIDE 2 Motivation
n Tropical cyclone(TC) intensity forecast is still a challenging problem in research and operational practice. n Radar is one of few observation platform capable of
TC internal structure and circulation
precipitation at high spatial and temporal resolution. n EnKF is able to deal with complex ice and multi-moment microphysics and more suitable to assimilate radar data(Vr and Z) directly.
SLIDE 3
n The reflectivity operator for double-moment is developed by CAPS:
GSI-based EnKF System
Milbrandt and Yau (2005)
SLIDE 4 n Forecast model: WRF-ARW v3.9.1,3km n DA systems: GSI-EnKF, 40members n OBS: composite radar reflectivity
GFS analysis IC+perturbations 06Z28
40-mem
11Z28
Deterministic forecast with ensemble mean(noDA)
06Z29
EnKF every
15minutes With Z
12Z28
Deterministic forecast with ensemble mean(Thompson)
Domain and Obs
First Cycle Starts 5h forecast 1h DA
Experimental design
SLIDE 5 𝑣" 𝑤" 𝑥" 𝜄" 𝑞" 𝑟(
"
𝑟)
"
𝑟*
"
𝑟+
"
𝑟,
"
𝑟-
"
𝑂/0
"
= 𝑣1 𝑤1 𝑥1 𝜄1 𝑞1 𝑟(
1
𝑟)
1
𝑟*
1
𝑟+
1
𝑟,
1
𝑟-
1
𝑂/0
1
+𝐿 𝑎 − 𝐼 𝑞1 𝑨1 𝑟(
1
𝑟*
1
𝑟-
1
𝑟,
1
𝑂/0
1
Radar data: Z=Z(qr+qs+qg) K is the Kalman Gain, a function of background and obs error covariances H is the observation operator
EnKF for Radar DA with a 2-Moment Microphysics Scheme
SLIDE 6 Z Obs bkg =NoDA Background is stronger
GSI EnKF Radar DA
NoDA 1 2 3 4 5 Radar- DA
11Z 12Z
Radar-DA(ana)
SLIDE 7 Z Obs NoDA
GSI EnKF Radar DA
Radar-DA(ana) Radar-DA(bkg)
NoDA 1 2 3 4 5 Radar- DA
11Z 12Z
SLIDE 8 Z Obs NoDA Radar-DA(ana)
GSI EnKF Radar DA
NoDA 1 2 3 4 5 Radar- DA
11Z 12Z
SLIDE 9
201805281200
RMS Innovation
SLIDE 10
First Cycle
Hydrometers analysis Increment
qr qs qi
Why Radar DA is better?
All hydrometers decreased after radar DA.
SLIDE 11
T W Dynamic consistent First Cycle Wind-850hpa Large-scale variables were updated correctly.
Other variables
SLIDE 12
Qi Qs Qr Qnr
Radar-DA
Qi Qs Qr Qnr
NoDA
NoDA vs. Radar-DA at final cycle
SLIDE 13
Track Forecast
SLIDE 14 Intensity Forecast
Positive Impact
NoDA Radar-DA Best-Track
SLIDE 15
1-hour forecast Z Observation NoDA Radar-DA
precipitation forecasts
SLIDE 16
2-hour forecast Z Observation NoDA Radar-DA
precipitation forecasts
SLIDE 17
Name Radar Z DA Mp-scheme Thompson-NoDA No Thompson Thomson Yes Thompson Lin_NoDA No Lin Lin Yes Lin
Additional Experiments
SLIDE 18
n A convective-scale EnKF DA system was successfully applied to directly assimilate radar Z data for a TC case. n Radar Z assimilation has positive impact on both intensity and precipitation forecast, the impact on precipitation maintained at least 1 hour while the impact on intensity can last for 12 hours. n EnKF well updated not only hydrometers but also other cross-variables (u,v,t,w), so as to produce well-balanced analysis state.
Summary
SLIDE 19
Thanks for your attentions!
