1. Juanzhen Sun (RAL/MMM)
ICE-POP 2018: PyeongChang Winter Olympics R&D to Enhance Winter Weather Research
Juanzhen Sun (RAL/MMM)
Zhuming Ying (MMM)

2. OUTLINE Overview of ICE-POP 2018
NCAR’s participation: wind and precipitation
nowcast using FINECAST
Some preliminary results

3. ICE-POP 2018
International Collaborative Experiments for Pyeongchang 2018 Olympic & Paralympic winter games)
Co-sponsored by KMA and WMO
RDP/FDP kick-off meeting Oct 2015; 3 workshops since then
Participants: 27 agencies from 12 countries
Australia, Austria, Canada, China, Finland, Russia, Korea,
Spain, Swiss, Taiwan, UK, USA
Pyeong
Chang
150 km
Seoul
South Korea

4. Three Scientific Issues
Main Themes of ICE-POP 2018
Three working groups
Observation
NWP
Evaluation
observation network
observation campaign
Nowcasting & NWP model
To operate model for FDP
Model verification & evaluation
Three Scientific Issues
Seamless prediction (0-12h)
Predictability over Mountainous, Complex Terrain of snow storms
Toward Advanced Physics in NWP

5. ICE-POP 2018 Fields Campaign
December 2017 – March 2018 S

6. CHALLENGES FOR NWP Mechanism for East Snow Storm(ESS) 2014. 1. 21 case
1) Updraft by terrain
2) Convergence between land(Mountain) and sea
Advection of Cold air
4) Complex flows and phase changes
3) Heat/moisture flux by sea
East Snow Storm caused by Cold outbreak over East Sea
- Low predictability in NWP, few observation in Ocean
Scale Interaction, Scale Transition
- Air-Sea interaction, Synoptic to Mesco scale transition
Microphysical Phase change in Ocean – Coastal Area – Steep Topography – Complex

7. NCAR’s involvement in ICE-POP 2018
Wind and precipitation analysis and nowcast over complex
terrain using FINECAST (update every 10 min)
Scientific questions:
Do FINECAST work for winter weather?
How to estimate precipitation from radar observations
in wintertime?
What is the quality of KMA’s radar network?
How shall we deal with the complex terrain?
Will the X-band radars provide additional value?
How to produce reliable surface wind and temp nowcast with
advanced post-processing

8. Overview of FINECAST (VDRAS)
A rapid update (10min) analysis and
nowcast system
4DVar Radar and surface data assimilation
WRF model data as first guess
A cloud-resolving model used for
analysis and nowcast
Operational capability
MPI parallel processing for real time rapid computation
Installed over many regional domains
Applications
Nowcasting severe weather
Process study of convective system
Initializing NWP models
Wind energy prediction
Winter weather

9. Technical Development for ICE-POP
FINECAST for ICE-POP2018
16 Site Profile
format: ascii
6 KMA operational
Radars X-band:
Radial vel & reflectivity
KMA
Verification
FINECAST
Analysis &
Nowcast
2D product:
lev: Surface & 2km
format: GRIB
AWS surface:
Wind, Temp, Pres,
Humidity, Rainfall
Standard FINECAST
output and
diagnosis
WRF 3DVar RUC
forecasts
In-house
Verification
Outputs: prep, type, Temp, RH, hori/vert wind, visibility, qc, qr, qs
Lead time: 0--2hr
Frequency: 10 min cycle with 10 min forecast output

10. Domain & Configuration
Domain and Configuration
Domain & Configuration
PyeongChang area terrain
Size : 280km X 210km
Resolution: hori 2km, vert 200m
Vert Levels: 27, model top 5km
Center: 37.46/128.53
Radars: GNG, GDK, KWK, IIA, MYN, KSN
Case study: UTC, Jan 29, 2016 ⛷

11. The snow storm
The case Reflectivity mosaic from 01UTC to 07UTC every 10 Min ⛷

12. Comparison with WRF (Qr & DIV)
Comparison with WRF for Qr & DIV
WRF (5h FCST)
FINECAST analysis Qr Qr ⛷
DIV
DIV
Convergence
Divergence

13. Comparison with WRF (Qr & DIV)
Comparison with WRF and OBS
WRF (5h FCST)
FINECAST analysis Qr Qr ⛷
1h rainfall
Gauge + radar
Radar REFf

14. 1H accumulated Rainfall Nowcast 1H precipitation forecast
_05:00UTC
OBS (Radar + gauge)
FINECAST nowcast
OBS

15. 1H accumulated Rainfall Nowcast 1H precipitation forecast
_06:00UTC
OBS (Radar + gauge)
FINECAST nowcast
OBS
FINECAST 1H FCST
OBS
FINECAST 1H FCST
OBS

16. 1H accumulated Rainfall Nowcast 1H precipitation forecast
_07:00UTC
Full model 1H FCST
OBS (Radar + gauge)
FINECAST nowcast
OBS

17. RMSE of Radial Velocity (B-O vs. A-O)
Wind RMSE against radial velocity
(BKG-OBS)
The errors are larger than the typical error ~ 2m/s in previous studies
The largest RMSE is on the 1st elevation angle of KWK
The largest error occurs in areas far from the radar (horizontal distance >100km)
Poor radar coverage might have
caused the poor fit to
observations
B-O
WRF - OBS
A-O
High elevation angle, 资料太少，没有统计意义， need to remove
FINECAST - OBS

18. Surface wind and T analysis/nowcast
Gridded
Surface obs
FINECAST
analysis
05 UTC
Gridded
Surface obs
FINECAST
1h nowcast
06 UTC

19. RMSE against surface OBSV T Qv

20. Verification of rainfall FCST
Verification of precipitation nowcast
FSS with 10km ROI

21. Lessons learned so far Summary
Nowcasting at the scale of Olympics venues presents great challenges for NWP models
KMA’s operational radar network have a poor coverage in the PyeongChang
area and therefore the QPE uncertainty is an issue; the additional X-bands are expected to
provide some help
FINECAST analyzed and predicted some mesoscale features that were reasonable
in terms of general precipitation patterns; but uncertainties for high precipitation are
quite obvious
FINECAST improved over WRF on surface wind and temperature forecasts; a terrain
adjustment scheme can further improve surface temperature forecast
The dense observations from ICE-POP 2018 field campaign create a valuable dataset for
more in-depth study on terrain-induced winter storms; we will seek opportunities for
continued collaboration with KMA.