High frequency and resolution analysis by assimilating radar and surface observations

1
High frequency and resolution analysis by
assimilating radar and surface observations

• Juanzhen Sun
• Eunha Lim

2
Objectives

• Produce high frequency and resolution analysis
  over an IHOP sub-domain using VDRAS
• - to study features associated with storm
  initiation for the entire period
• - to initialize WRF to produce 0-6 hour
  forecasts for two cases

3
Experiments Setup

• For the convective feature study
• Domain 600km x 600 km x 5.625 km
• Resolution 3 km x 375 m
• Update frequency 17-18 min.
• For the WRF initialization study
• Domain 800 km x 680 km x 17.5 km
• Resolution 4 km x 500 m

4
VDRAS domain for the feature study
5
VDRAS analysis of 10 June, 2002 (Two hours before
the initiation)
Ref
Vr
W
T
6
Gust front (?T(t1)-?T(t)) evolution 4 hour
loop after the initiation
7
WRF initialization using VDRAS analysis

• Motivation
• VDRAS provides some details at the convective
  scales - Can these detailed analyses be used to
  drive a
• different model?
• Implication on future WRF 4DVAR - Is it possible
  to use a simpler version of WRF as the 4DVAR
  forward model?

8
Experimental Design
10 km
Sounding
RTFDDA
Extract profiles (?x/y 25km)
Sfc, Radar
VDRAS
VDRAS
Interpolation to 10 km RTFDDA domain
NESTDOWN to
3.3 km
MESO
PROF
CNTL
WRF 6hour forecast
Initial condition for WRF
9
Reducing boundary inconsistency between VDRAS and
WRF

• Interpolated fields of VDRAS analysis (MESO) to
  RTFDDA_d02
• U-wind at the 1st level
• Without (left) and with (right) weighting on
  RTFDDA

19 UTC 15 June 2002
10
00 UTC 16 June 2002- 1hr-accumulated rainfall
5hr forecast
OBS
CNTL
STEP_CNTL 12hr forecast
MESO_FP
PROF_FP
11
Rainfall verification 15 JuneThreshold 5.0
mm / hour
CNTL
MESO
PROF
CNTL

• For higher threshold, ETS of WRF with VDRAS is
  generally higher than CNTL
• Especially at first 2 hours
• WRF with RTFDDA (CNTL) shows higher BIAS

12
00 UTC 13 June 2002- 1hr-accumulated rainfall
1hr forecast
CNTL
OBS
MESO
PROF
STEP_CNTL 12hr forecast
13
03 UTC 13 June 2002- 1hr-accumulated rainfall
4hr forecast
CNTL
OBS
MESO
PROF
STEP_CNTL 15hr forecast
14
Rainfall verification 12 JuneThreshold 5.0
mm / hour
MESO
PROF
CNTL

• ETS of WRF with VDRAS analysis is better than
  CNTL

15
Summary

• VDRAS analysis reveals some small-scale details
  that can be used for thunderstorm nowcasting.
• The convective-scale information provided by the
  analysis has the potential to improve 0-6 hour
  precipitation forecasts