Impacts of aircraft and radiosonde descent data on global model … · 2020. 11. 30. · Impacts of...
Transcript of Impacts of aircraft and radiosonde descent data on global model … · 2020. 11. 30. · Impacts of...
7th WMO Impact workshop
Impacts of aircraft and radiosondedescent data on global model at the KMA
Ji-Hyun Ha, Yoon-Jeong Hwang, Dayoung Choi and Yong Hee Lee
Numerical Modeling Center, KMA
Introduction
• In situ observations are very important as a complement to satellite-based observations.
• When assimilated into the numerical models, in situ observations serve as a reference data and correct the model. Therefore, the impacts of the in situ observations need to be investigated.
• This presentation is investigated the effects of the aircraft and radiosonde descent data in the global forecasts by conducting the denial and the sensitivity experiments.
Science questions: ① Impact assessment of AMDAR②Regional upper-air network studies with
the use of sonde descent data
1. Impact of aircraft observations
• Model and DA system- Global Model: Korean Integrated Model (KIM), 25 km- Data assimilation: Hybrid-4DEnVar (3DVAR + LETKF), 50 km- 2018.7.1 ~ 7.19 (spin-up: 2018.6.23 ~ 6.30)
• Experiment design1. CTL : data assimilation of aircraft data
(12 level vertical thinning) 2. EXP1: No data assimilation of aircraft data3. EXP2: vertically dense aircraft data
(12 à 21 level vertical thinning)
Coverage of AMDAR and AIREP
Impact of the aircraft observations on the analysis (1)
• Comparison between CTL and EXP1Negative means the improvements of the CTL
U V
T Q
Aimproved
- The aircraft data improves the mid-latitude upper-troposphere, which corresponds to the aircraft trajectory.
Impact of the aircraft observations on the analysis (2)
• Region A averaged: improvedT Analysis (K) T 6h forecast (K) 6h forecast T RMSE
IFS analysisCTLEXP1
• Background error and analysis increment (2018.7.10.00 UTC) T Background error @ 250 hPa T Analysis increment @ 250 hPa
CTLCTL
EXP1 EXP1
à cold bias corrected by increasing increment in CTL compared to EXP1
A
A
cold bias decreased
warmer increment
Impact of vertically dense aircraft observations on the analysis (1)
U
T
V
Q
• Comparison between CTL and EXP2Negative means the improvements of the EXP2
Bimproved
- Vertically dense aircraft data improves the tropical and northern hemisphere upper-troposphere.
• Region B averaged: improvedT Analysis (K) T 6h forecast (K) 6h forecast T RMSE
IFS analysisCTL EXP2
EXP1
à Small cold bias corrected by increasing increment in EXP2 compared to CTL
• Background error and analysis increment (2018.7.10.00 UTC) T Background error @ 250 hPa T Analysis increment @ 250 hPa
CTLCTL
EXP2 EXP2cold bias decreased in tropical region
warm increment suppressed in tropical region
Impact of vertically dense aircraft observations on the analysis (2)
Impact of the aircraft observations on the forecast
• Improvement of GPH forecast error by the aircraft observations
- The decline in the aircraft due to the COVID-19 would have affected the global model performance.
ü RMSE
NH (500hPa)
EXP1CTL
NH (100hPa)
ü ACC
%
ü ACCEXP1CTL
• Improvement of GPH forecast error by the vertical resolutionü RMSE
500 hPa850 hPa
100 hPaNH
NH (100hPa) EXP2CTL
% %
NH (850hPa) EXP2CTL - The dense aircraft data
improves the upper-troposphere. optimal vertical resolution and high-quality data needed.
2. Impact of regional radiosonde descent observations
• Sonde trajectory
JUL (2019.7) DEC (2019.12)
Use of descent data from 6 stations (red: descent data)
• Sonde descent vs AMDAR
Gangneung Pohang
removal• Experiment design1. CTL : No data assimilation of descent data2. EXP : data assimilation of descent data
(drift + removal above of 100 hPa) - 2019.8.1~15. (spin-up: 2019.7.25~7.31)
longitude, latitude: ±0.5, pressure: ±5hPa, time: ±30min
à Similar quality level compared to AMDAR
Impact of the descent data on the analysis
CTL EXPCTL
• innovation • Error against IFS analysis
à Reduction of innovation
àImprove the tropical and northern mid-latitude wind and temperature
Impact of the descent data on the forecast
• Improvement of 3-day forecast errors
• Improvement of 5-day GPH forecast errors (NH) (anal) 1.6%, (sonde) 1.2% (East Asia) (anal) 2.1%, (sonde) 0.3%
T GPH
Negative means the improvements by the use of descent data
- The descent data improves the temperature and geo-potential height forecasts in the mid/high-latitude regions.
• In winter, it did not show as much improvement as in summer. à(future) time thinning, observation error tuning planned.
Remarks
• The aircraft data made the temperature warmer in the northern mid-latitude upper-troposphere, which helped to reduce the cold bias and errors.
• The vertically dense aircraft observations reduced the cold bias, and these observations worked by suppressing the warm increments.
• The regional sonde descent data from 6 stations showed the similar level of quality compared to the AMDAR data, and it improved the environment in the tropical and mid-latitude.
• The features need to be re-investigated with the optimal vertical aircraft resolution and for sondedescent data in the northern hemisphere winter season due to the strong winds.
Thank you for attention.