Guimei LIUNatinal Marine Environmental Forecasting Center,
Beijing, China
Halifax, Canada
May 2019
Past, present, and future
of marine environmental
forecasting services in China
Outline
3 Future(2019-): innovation
2 Present(2013-2019):development
1 Past(1965-2013): history
• Past: 1965-2013
⚫ 1965: established as the Marine Hydro-meteorological Forecasting General Observatory
⚫ 1966: marine weather and wave forecast(China seas)⚫ 1969: sea ice forecast(Bohai Sea)⚫ 1970: tropical cyclone storm surge forecast⚫ 1972: extra-tropical storm surge forecast⚫ 1975: sea temperature forecast (East China Sea)⚫ 1983: renamed as the National Marine Environmental
Forecasting Center⚫ 1986: release forecasts to public by radio and TV⚫ 1996: El Niño forecast
Past: 1965-2013 First group photo Release forecast by fax
Marine
environmental
forecast for
global coastal
cities
LoMF,
Established Key
Laboratory of
Research on
Marine Hazards
Forecasting
1966年
Released sea wind,
wave forecast
covering China seas
1970年 1974年 1977年 1980年
Released sea ice forecast
covering Bohai Sea and
north of Yellow Sea
Released storm surge
forecast in south of
China
Released SST forecast
in China seas
Marine forecast
for special
services
1984年 1985年 1986年 1996年 2001年
Polar Route
Forecast
Released forecast
products on
CCTV and radio
El Nino
prediction
Red tides
warning and
forecast
Public forecast
by NMEFC
Marine
Environmental
Forecasts of
Coastal Tourist
Resorts
2002年 2005年 2006年 2008年
Released forecast
products of
swimming beach
environmental
conditions
Tsunami warning
and forecastMarine
environmental
forecast in the
Gulf of Aden
& Green Tide
CGOFS v1.0
2013年
Sea Wind &
WaveSea Ice Storm Surge Sea Temperature
Polar Route
2010年 2012年
Marine environmental forecasting history in China
1989年
Numerical
forecasting
system for SST
Numerical
forecast for SSTEl Nino Red Tide
Green Tide
Numerical forecast for
wave, storm surge, wind
Numerical forecast for
temperature-salinity-
currents and wave
Tsunami CGOFS v1.0Global,regional,coastal numerical
forecast systems
86-90:To develop numerical forecast for SST
of Z-coordinate system in China seas and
the northwestern PAC; To develop
numerical forecast for wave.
91-95:To develop forecast system from fix to variable depth of mixed
layer; To introduce objective correction (data assimilation) to forecast
system; Numerical forecast for storm surge
01-05:3D temperature-salinity-current forecast system of Bohai Sea, East
China Sea, China Sea and Northwestern PAC; wave forecast system
(WAM/SWAN)of China seas and Northwestern PAC
06-10:POM for China Seas, HYCOM for northern PAC, MOM4
for global ocean; numerical forecast products
temperature, salinity, current, wave and other elements
11-15:Chinese Global operational Oceanography Forecasting System (CGOFS v1.0)
FVCOM
Empirical Forecast
WAM
4
SWAN
WW3
Autocontrol
Objective
correction
1960s 1970s 1980s 1990s 2000s 2010s
POM
HYCO
M
ROMS
Marine environmental forecasting history in China
1900s … … 1980s 1990s 2000s 2010s
GODAE 1997-2008
GOV 2009-
Argo 1998- Real-time observation
Objective correction (data assimilation) was applied to forecast
From 1970s,computer and marine numerical simulation began to develop
Observational data is more and more
From GEOS-3 in 1975,application of remote sensing data was more widely
WOCE 1990-2002
TOGA 1985-1994
TAO/TRION (1984) Real-time observation network
Initiation of Operational Oceanography
Adjustment and demonstration of
Operational Oceanography
Continuous improvements of Operational
Oceanography
Development of OperationalOceanography
Initiation of Modern
Oceanography
Operational
Oceanography History
MOM4, HYCOM, ROMS, POM and FVCOM
Bohai Sea
LYG PortChina Seas
North Pacific
Northwest Pacific
Numerical Forecasting Models
Outline
1 Past(1965-2013): history
2 Present(2013-2019):development
3 Future(2019-): innovation
1. Establish Chinese Global operational Oceanography Forecasting System and the extended forecasting system
2. Data assimilation technology
3. Application of GPU based parallel detection technology
4. Quality of products and applications• collaboration
• development
• service
In recent 5 years : development of the forecasting systems
Chinese Global operational Oceanography Forecasting System
(CGOFS v1.0) and extended forecast system
Global Oceanography Forecasts:
Level 1: Global Ocean
Level 2:
Northwest Pacific and
Indian Ocean
Level 3:
Bo-Yellow-East China Sea and South
China Sea
Level 4: Polar Region
Refined forecasts:
China Coastal Zone
Ecological Forecasts:
Level 1: Northwest Pacific
Level 2: East and South China sea
Climate Prediction:
Level 1: Global
Level 2: Asia & Northwest Pacific
Chinese Global operational Oceanography Forecasting System and the extended forecasting system
GlobalIndian Ocean
North-western Pacific
SouthChinaSea
East China Sea
Bohai Sea
OGCM NEMO ROMSv3.5 ROMSv3.5 ROMSv3.5 ROMSv3.5 ROMSv3.5
Horizontal Res. 1/12° 1/12° 1/20° 1/30° 1/30° 1/60°
Vertical Res. 75L 20L 22L 30L 30L 36L
Surface Forcing GFS/CGOGS-wind/NMEFC-WRF
Tidal boundary None OTIS East China Sea
Assimilation scheme PDAF QcorretionEnOI
QcorretionEnOI
QcorretionEnOI
QcorretionQcorretion
Data assimilatedRTGSST,SSH,
Argo T/SMGDSST
MGDSST,SSH, Argo T/S
MGDSST,SSH
MGDSST, SSH MGDSST
Product Temperature, salinity, current (surface, 20m, 50m, 100m, 200m, 500m, 1000m, profiles)
Forecast range 120hr
RMSE of SLA RMSE & Bias of SST
Temp. profiles
Sali. profiles
regions
Oct.-Nov. 2018
National Model System SST SLA(m) T-profile S-profile
UK NEMO 1/4 FOAM 0.3066 0.0783 0.5535 0.1383
AustraliaMOM4 10kmoffshore
BLK omap 0.4445 0.0786 0.6800 0.1610
Canada NEMO 1/4 GIOPS 0.4404 0.0757 0.6150 0.1775
USA HYCOM 1/12 RTOFS 0.5608 NaN 0.6345 0.1440
France NEMO 1/4 PSY3 0.5328 0.0866 0.6287 0.1477
France NEMO 1/12 PSY4 NaN 0.0702 NaN NaN
China NEMO 1/12 NMEFC-NEMO 0.2941 0.0670 0.7115 0.1726
IVTT class4SST
1. Global High resolution Forecasting products
Temp
Salinity
Current
1. Global High resolution Forecasting products
EnOI for SLA
Observation data for assimilation: SLA
(Jason-2, Cryosat, SARAL and HY2)
Assimilation cycle: 7 days (at 00:00)
Parameter value: alpha=0.3
localization radius: R=250km
Before DA
After DA
SLA Bias
BDA (cm) ADA (cm)
ME 8.4 6.3
AE 10.1 7.1
RMSE 12.6 8.7
Validation 2016/9/24
2.Data assimilation technology
3DVAR for SST and T/S Argo data
T: RMSE improves 50% at the level of ~10m
S: RMSE decreased by ~0.040 for the upper 1000 m
2.Data assimilation technology
0.988
0.620
0.098
0.071
37.21%
27.98%
SST: model vs. MGDSST
The RMS of AF temperature is ~0.83 oC in 2006, which is improved by
~ 23.2% with compared to ~1.08 oC of SF.
The RMS of AF salinity performs better than the RMS of SF, with
~0.112 of AF compared to ~0.138 of SF.
⚫ GPU, 86s response to forecast period of 72hr
⚫ Coastal cities of China and the Southeast Asia
⚫ Observational data and forecast products for
the public
3. Application of GPU based parallel detection technology
• 2017 Fastest response:6min,average response:11min
• 2018 Fastest response:3min,average response:9.8min
IOC/UNESCO SCSTAC
2 区
1 区
3 区
High resolution storm surge numerical forecasting system
1 2
3 4
Marine hazard & disaster forecast
Response to global climate change
Marine fishery and aquafarm Coastal natural resources
0
2
4
6
8
10
Aug., 1959Aug., 1981Aug., 1988Aug., 1998Aug., 1999Aug., 2002Jun., 2003Aug. 2003Sep., 2003Aug., 2005Aug., 2006Aug., 2011
0
5
10
15
20
201108200308
?
15.4
2.6
5<5
0.5
10.9
0.60.55?
1.9
200608200508
200309200306200208
199908199808
198808198108
Are
a o
f h
yp
ox
ia z
on
e (
×10
3 k
m2)
(DO
<2
mg
/L)
Time
area
195908
DO min.
DO
min
imu
m (m
g L
)-1
4. Products & Applications
Sanchi
Daily forecast trajectory error
0
5
10
15
20
25
26
74
83
11
7
15
5
Drifting time/h
dis
tan
ce e
rro
r/k
m
Observed trajectory
Trajectory forecasting service for tanker Sanchi
Range prediction error in 24h
• The tanker carrying 136,000 metric tons of light crude
oil has been adrift and on fire following a collision
with another vessel in the East China Sea
(124°56′42″E 、 30°42′42″N) on Jan 6, 2018.
Tanker sinked on Jan 14 and caused oil leakage.
• NMEFC activated emergency forecast for drifting
trajectory of victims and the tanker on Jan 7. The
model was upgraded based on semi-analytic method,
and proved to be with better generality than previous
models.
• After comparison with observed trajectory, the model-
predicted result showed acceptable accuracy.
Behavior and fate scenario simulation of
submarine oil spill in shipwreck
North-western Pacific operational forecasting system
(1/20°)
✓ The emergency forecast of the future 48-hour of the oil distribution was
performed and published on daily basis from 14 January to 2 February.
✓ A long term fate and behavior for condensate and bunker oil during January and
February was performed by coupling SINTEF OSCAR oil spill model, satellite
observations, with operational wind and current forecasts of NMEFC.
✓ The leakage from the submerged tanker was also investigated.
✓ A validation study was carried out for the wind, current, oil distribution and
shoreline hits.
The forecasting conclusions have successfully supported the decision
making for the response of the Sanchi oil spill.
Weathering process comparison between fuel oil
and condensate oilPrediction of Fuel oil drift and diffusion for 30 days
Oil spill forecasting service for tanker Sanchi — drift,
diffusion and weathering
Intelligent grid forecasting
Global ocean 25km——Coastal sea 10km
(including wave, temperature, sea surface wind)
Green ≥ 0.9
0.9 > Yellow ≥ 0.85
0.85 > Orange ≥ 0.8
Red < 0.8
24h 48h 72h
Development of marine forecast
Observation
channelfoundation
trench
Forecasting
weather wave current
before one month
select time window
before 5 days
decide time window
before 1 day
confirm time window
operation
In-situ support
Hong Kong-Zhuhai-Macao Bridge Construction
NMEFC has been providing timely and accurate forecasting service to the bridge construction for
over 6 years, and solved multiple technical challenges including the critical requirements of time-
window, accuracy, and emergency response
Region:
North of SCS
inner wave, wind, surface
wave, currents, temperature
and density
Task:
Feb to Jul, 2017
First mining by Marine
Engineering Limited
Company of CNPC
Requirements:
Safty of “No.1 Blue
Whale”Semi Submersible
Platform
Safety of conveying pipe
NMEFC provided forecast products and in-situ support for the safety of the mining operation
South China Sea Natural Gas Hydrate Test mining
In recent 10 years: development of super-computing ability
Serial computing
Parallelcomputing
OpenMP
MPI
CPU GPU
APP and automation
platform
TODAY HEADLINES NETEASY NEWS EXPRESS NEWS
TENCENT NEWS Yidian zixun
APP Weibo
Release of forecasting products
Outline
2 Present(2013-2019):development
3 Future(2019-): innovation
1 Past(1965-2013): history
Forecasting coverage advances from major ports and observation stations to
important maritime routes, deep seas and global waters. A combination strategy of
points, lines and interfaces would expand the coverage of environmental forecasting.
• Broadening of protecting scope
• Expansion of forecasting coverage
Exploring and Developing
Major sea ports
Important maritime
routeDeep seas
Global
Region
Physical Oceanography
Ecological Oceanography
Environmental Oceanography
Military Oceanography
F-60 F-59 F-2F-58 F-1 O-60 O-59 O-2O-58 O-1
History Forecasting Data
Correlation Downscaling
History Observational Data
深度学习方法
Focus region
Neural Networks
Deep learning
New Challenge
Super-data Super-computing Intelligent forecast
⚫ To develop a new large-scale super-computing platform with multi-disciplinary integration of ocean,
atmosphere and earth.
⚫ To apply advanced technology such as supper data analysis, scientific engineering computing and
artificial intelligence.
New Challenge
Thank you for your listening!