Seasonal Simulaions of a coupled ice-ocean model in the Bohai Sea and North Yellow Sea Yu LIU,Qinzheng LIU,Jie Su*, Shan BAI,Maoning Tang National Marine Environmental Forecasting Center * Ocean University of Qingdao
Contents Introduction Model Description Model Results Conclusions
Introduction The Bohai Sea and its coastal regions are important economical areas in China. Ice condition affects the navigation and oil exploration in winter. The oil platforms have been destroyed and navigation stopped by sea ice during severe winters.
The ice conditions have been classified as different grades (the ice severity index) according to the ice thickness and extent based on the observed data and historical records in the Bohai Sea and the northern Yellow Sea.
Some previous works Sea ice dynamic model was developed, and numerical experiments of ice drift was taken in the Bohai Sea (Wang, et al., 984; 994, Wu, et al., 995; 998). A dynamic-thermodynamic sea ice model was developed in the Bohai Sea, and has been applied to operational ice numerical forecast in NMEFC, China (Yang et al., 99 and Bai et al., 998). An ice-ocean coupled model was developed by using POM and a viscous - plastic thermodynamic-dynamic sea ice model in the Bohai Sea, and continuous sea ice simulations for representative winters were taken(su et al., 4 5).
Motivation All the previous calculation domains of regional sea ice models or iceocean coupled models were limited to the Bohai Sea in China. However, the Bohai Sea and the Yellow Sea are closely linked, and the sea ice occurrence and disappearance in North Yellow Sea is closely related to the marine environment of the Bohai Sea, thus may in some way affect the sea ice occurrence and disappearance in the Bohai Sea. Based on the Arctic ice-ocean coupled model (Wang, 5) and some improvements including model s parameter optimization, the shallow water treatment and open boundary treatment, a coupled ice-ocean model suitable for the Bohai Sea and North Yellow Sea with a broad shallow water area was developed, which has more difficulty in open boundary treatment than previous domain. In addition, all previous simulation periods of regional ice-ocean coupled models were no more than winters in China. By using this ice-ocean coupled model and the regional atmospheric reanalysis data, a series of seasonal evolutions of sea ice cover have been simulated and analyzed from the winters of 997/998 to 8/9 in the Bohai Sea and North Yellow Sea.
Model Description POM + sea ice model (V-P) POM: in spring, summer and fall Coupled ice-ocean model: in late autumn, winter and early spring
Model Description (I) Processes: 5 tidal constituents, density current wind-driven current, sea ice in winter New Features: New treatment for Super shallow water h η (no constraints for the topography-water depth) (Liu et al. 5) Gradient force for steep topography CHU et al 997 3 point CCD scheme
Model Description (II) Boundary conditions: Elevation: tide forcing + radiation elevation U U (t ) = ± C [η η (t )] h -D velocity: 3-D velocity: Radiation (group velocity + advection) Temp. & sal.: outflow Radiation inflow climatological restoring Turbulence: =
Model Description (III) Radiation boundary conditions [φbn, j + rxφbn +, j ry (φbn, j φbn, j )] + rx While ry> = [φbn, j + rxφbn +, j ry (φbn, j + φbn, j )] + rx While ry< φbn,+j = φ n + b, j
Model Description (IV) Sea Ice model Viscous-plastic dynamic model -layer thermodynamic model Heat balance at top/bottom surface, without snow cover Multi-category thickness distribution dynamic ice deformation
Model Description (V) Ice-ocean Coupling(Mellor and Kantha,989) FT=-ρwCpCTz(Tf-T) FS=-CTz(T-T) (ocean heat flux under ice layer) (ocean salinity flux under ice layer) τ = Aτ iw + ( A)τ a τ wi = ρ wc Dw Vw Vi (Vw Vi ) CTz = C Sz u* Prt ln( z / z ) / k + BT u* = Prt ln( z / z ) / k + BS
sketch map of the air-ice-ocean interaction
Model Configuration Calculating Area 7 3 7 E 37 4 N ice model Arakawa B 3.7km.8km ocean model Arakawa C 3.7km.8km layers are σ -.36, -.7, -.43, -.86, -.49, -.57, (., -.8, -.74, -.857, -.). Table. Categories of the sea ice thickness n 3 4 Thick ness (cm) - -5 5- n 7 8 9 5 6-5 5- Thick ness -5 5-3 3-4 4-5 5-75 (cm) >75
numerical tests MM5 reanalysis Seasonal simulations of the winters of 997~9 Initial ocean conditions derived from a climatologically, objectively analyzed temperaturesalinity data atlas. Time Step MM5 h 6s Ice model 6s Ocean Model external s internal 6s Spin-Up Seasonal : Forecasts 3d 4d During the simulation, there are no ice fields restoring.
Model Verification
Tide & SST Distribution of Co-amplitude and Co-phase lag Lines of M Tidal Constituent Elevation 4 6 4.5 4 7 8 4 Latitude(N) 6 38 3 39.5 3 3 SST (April) 4 6 7 4 8 6 38.5 3 4 6 33 39 Latitude 4 33 3 39.5 9 6 3 9 8 4 Tidal Chart (M) 4 4.5 8 39 9 38.5 8 38 7 37.5 6 37.5 7.5 8 8.5 9 9.5.5 Longitude.5.5 37 7.5 8 8.5 9 9.5.5.5.5 Longitude(OE) 5
Simulated SST at (-) 3 3 O C ) ( 5 5 5 p o n i t 3 o T e m p e C r a ) t u r e ( 5 5 5 3 4 5 6 J u l i a n D a y s 7 8 9
Structure of Temp. and Sal. 3 3-3 - Depth(m) Salinity 9-3 4 8-4 7 - -5 6 - Depth(m) -6 7.5 8 8.5 9 9.5.5.5.5 8 Longitude(OE) -3 6-4 Temperature -5 4-6 7.5 8 8.5 9 9.5.5.5.5 Longitude( O E) 5
.3 -. -. Depth(m) Structure of velocity.4-3 -. -4 -. -5.5 U-component -.3-6 7.5 8 8.5 9 9.5.5.5.5 Longitude(OE) -.4 9.5-6 Dep th (m ).3. -3. 8-4. 3 4 5 6 7 8 9 5 3.5-5 V-component -6 7.5 8 8.5 9 9.5.5.5.5 L on gitud e( O E) 33 -. -. 4 Current Vector 3 7 -.4
Ice Covered Area and Ice Thickness ( m ) (-). 7. 6. 5. 4. 3.. ( k m 4 x 3. 5 ) 3. 5. 5. 5
4.9 Ice Concentration 4.5.8 4.7 3 39.5 x 4.6 Ice-Covered.5 Area 39.5 38.5.4.3 (k m ) 38 37.5..5 37 7.5 8 8.5 9 9.5.5.5..5 Sea Ice in the Bohai Sea.5 5 5 ( ~ 5 ) 3 3 5
Seasonal Simulation The ice-ocean coupled model was used to simulate sea ice evolution in the Bohai Sea and North Yellow Sea during the winters of 997-9. Simulation begins on October 5th every year. The spin-up time of coupled simulation is generally about 3 days. Results of sensitivity experiments in the Bohai Sea and North Yellow Sea show that 3 days are needed for the model to achieve stability in the case of no wind and no tide condition. However, it can achieve stability in several days with the influence of wind and tide. In actual simulation, spin-up time depends mainly on whether the simulated or forecasted surface temperature is consistent with the air and sea ice conditions at that time.
Seasonal Simulation Examples from 3..5 to 4.3.3 a b d a c e 3..7 f b4..5 4..7
Winter of 3/4 MEFS-si Seasonal Simulation Ice Concentration Ice Ice Ice Freezing Ending Period Cal 7 / / 76d Real 6 / 4 / 8d Ice Thickness(cm)
Years The date of maximum sea ice coverage Actual Calculation Error 997-998 998..3.8-3 998-999 999... 999-..8.6 8 -..6.6 -..7.7-3 3..5. 6 3-4 4..5.3 5 4-5 5..5.6 5-6 6..8.8 6-7 7..4.8 4 7-8 8..6.5 9 8-9 9..3. - Average error.33 Max error -3/9 mean square deviation 3.77 Comparison of the date of maximum sea ice coverage between simulated results and observed analysis
Ice-freezing date Years Ice period(d) Ice-ending date Actual Calcul ation Erro r Actual Calcul ation Error Actual Calcul ation Error 997-998 997.. 3. - 998..7.8 57 7 3 998-999 998..5. -3 999..8.6-76 77 999-999.. 8.9 -..4 3. 6 69 74 5 -.. 3. -.3. 3. 9 9 -..7.5 8... - 78 69-9 -3..8.8 3..5.3-6 58-3-4 3..6. 6 4..4.9-5 8 69-4-5 4... 5.3.4 3.4 8 8 5-6 5..6. 4 6.3.3 3.4-9 98 85-3 6-7 6..7.8 7..6. -5 6 47-5 7-8 7..9.3 8..9.9 63 6-8-9 8..3. 9 9.3.6 3.3-3 84 7 - Average error.8 -.67-3.67 Max error -/9-9/6-5/3 mean square deviation 5.58 3.7 8.44 Comparison of sea ice period between simulated results and observed analysis
Maximum sea ice extent of Liaodong Bay (nm) Years Actual Calculati on Error 997-998 7 56-6 998-999 58 38-999- 78 8-5 99-6 - 48 36 - -3 58 7 4 3-4 6 53-8 4-5 76 9 4 5-6 77 74-3 6-7 48 35-3 7-8 66 55-8-9 65 5-5 Average error -7 Max error -/4 mean square deviation Comparison of maximum sea ice extent of Liaodong Bay between simulated results and observed analysis
Simulated and observed daily ice area in winter of 5/6 Simulated and observed daily ice extent in winter of 5/6 The simulation of sea ice extent and coverage reflects the characteristics of seasonal sea ice evolution, however, during the melting period of sea ice, simulated sea ice melts much faster than real conditions.
Simulated sea ice thickness and concentration in the Bohai Sea and North Yellow Sea can represent the basic characteristics of sea ice distribution.
Conclusions. In general, the development of simulated sea ice conditions are consistent with the actual evolution processes, and the simulations of the ice-freezing date, ice-ending date and ice periods agree reasonably well with observations, and some are even identical to field data.. The simulation of sea ice extent and coverage reflects the characteristics of seasonal evolution of sea ice well, however, during ice-melting period, simulated sea ice melts much faster than reality.
Conclusions 3 Normally, the sea ice thickness of the west part in the Liaodong bay is less than that of the east part, which can be reproduced well by the ice-ocean coupled model. But the defect is that the ice thickness difference between the east area and the west area near the coast is significantly larger than that of satellite remote sensing data analysis. 4 Simulated sea ice concentration in the Bohai Bay, Laizhou Bay and North Yellow Sea can represent the basic characteristics of sea ice distribution, while the values of simulated sea ice concentration is generally lower and ice thickness is thinner.
further work a) To verify the water depth of the Bohai Bay, Laizhou Bay and North Yellow Sea and to get the real 3-d oceanic fields of oceanic temperature and salinity and take data assimilation for model initial fields. b) To check the meteorological field errors or feedback mechanisms of melting process in the model. c) To verify if the value of sea ice compressive strength and the ice thickness distribution are suitable.
recent work (preliminary results of ice assimilation) data assimilation scheme Nudging: da k = ( Aobs Amod ) dt τ where k is the weighting coefficient of optimal interpolation for combining two estimates of the same quantity (Deutsch, 965, Keguang Wang, ) Optimal interpolation: σ m k= σm +σo where σm and σo are model and observation standard deviation. 34
An example of nudging 45 no assimilation 45 nudging Need more numerical test
Thanks for your attention