Improved representation of nearshore processes at the southern coast of Japan using a 4DVAR assimilation system Norihisa USUI, Yosuke Fujii, Kei Sakamoto.

Improved representation of nearshore processes at the southern coast of Japan using a 4DVAR assimilation system Norihisa USUI, Yosuke Fujii, Kei Sakamoto and Masafumi Kamachi (Japan Meteorological Agency/Meteorological Research Institute) GOV COSS-TT U of Miami, Jan. 2012 1 Outline Overview of present operational systems in JMA Introduction of a next generation coastal system High-resolution model 4DVAR assimilation scheme Development plan of the coastal system Preliminary results of 4DVAR assimilation system Summary and future issues 2 Present operational DA systems in JMA: MOVE/MRI.COM Global (MOVE/MRI.COM-G) North Pacific MOVE/MRI.COM-NP Western North Pacific MOVE/MRI.COM-WNP 75 o S 75 o N 1 o x 1 o, L o E-75 o W, 15 o S-65 o N 117 o E-160 o W, 15 o N-65 o N One-way nesting 0.1 o x0.1 o, L o x0.5 o, L54 Initialization for ENSO and seasonal forecast Parent model for WNP model Monitoring and forecasting of ocean state around Japan (SST, Kuroshio, mesoscale eddy, etc) 3 Present operational systems in JMA: Ocean model MRI.COM-GMRI.COM-NPMRI.COM-WNP Domain Quasi-global (75 S - 75 N) North PacificWestern North Pacific Resolution 1.0 x 1.0 , 50L0.5 x0.5 , 54L0.1 x0.1 , 54L DiffusionIsopycnal+ Gent & McWilliams(1990)Biharmonic Viscosity Harmonic Smagorinsky Biharmonic Smagorinsky Mixed layer modelNoh and Kim(1999) Sea ice model Not applied Thermodynamics with thickness category EVP sea ice model MRI Community Ocean Model (MRI.COM; Tsujino et al. 2010) Z-coordinate, free-surface Quasi-enstrophy conservation scheme Topography diagonally upward advection scheme (Ishizaki and Motoi 1999) 4 Present operational systems in JMA: Assimilation scheme MRI Multivariate Ocean Variational Estimation System (MOVE; Usui et al. 2006) 3DVAR with vertical coupled temperature and salinity EOF modes (Fujii & Kamachi 2003) Incremental Analysis Updates (Bloom et al. 1996) Observation: Along track SSH anomaly Gridded SST (MGDSST: Japan-GHRSST) T-S profiles Sea ice concentration (for MOVE-WNP) 5 Present operational systems in JMA: Assimilation scheme 3DVAR with vertical coupled T-S EOF modes Cost function: Transformation from mode amplitude to T-S fields ( z x ) Deviation from first guess Deviation from T S observation Deviation from altimetry data z : amplitudes of T-S EOF modes (control variables) x : temperature and salinity Singular values of T-S EOF T-S EOF modes Standard deviation of the background field Next generation coastal system (MOVE/MRI.COM-Jpn): Purpose 6 Upgrade the present ocean information Sea surface temperature Kuroshio and Oyashio Sea ice in the sea of Okhotsk etc New targets of the coastal system Kyucho : a sudden and swift current caused by Small scale disturbances propagating along the Kuroshio (frontal waves) Coastal waves induced by along-shore winds Abiki : large-amplitude seiches generated by atmospheric pressure disturbances Next generation coastal system (MOVE/MRI.COM-Jpn): Purpose 7 Formation of Abiki Kyucho event Hibiya and Kajiura (1982) Isobe et al. (2010) Next generation coastal system: High-resolution ocean model Resolution: 1/33 o x 1/50 o (~ 2km) Domain: 117 o -167 o E, 25 o -50 o N Incorporate tidal forcing WNP model domain MRI.COM-Jpn: Jpn model domain 8 Next generation coastal system: Data assimilation Present operational system employs 3DVAR scheme as an assimilation method Computational cost is relatively low. Open ocean mesoscale phenomena such as Kuroshio meanders and mesoscale eddies can be reproduced. Need for a more sophisticated assimilation method But 3DVAR scheme is not appropriate for high-frequency phenomena in coastal regions 9 Next generation coastal system: Data assimilation MOVE-4DVAR: 10 Take advantage of experience of the present 3DVAR system Statistics of the background errors (T-S EOF) Observation operators Nonlinear constraints, etc A prototype 4DVAR system has been developed using the WNP (0.1x0.1) model The prototype 4DVAR system will be extended to an incremental 4DVAR system as an assimilation part of the next generation coastal system, MOVE/MRI.COM-Jpn z : amplitudes of T-S EOF modes x : temperature and salinity Next generation coastal system: Time schedule 11 2008 The present system (MOVE/MRI.COM-WNP) was used in operation. 2014 The prototype 4DVAR system with the WNP model ( 1/10 o x 1/10 o ) will be in demonstration phase. 2017 MOVE/MRI.COM-Jpn using the incremental 4DVAR scheme will be in operation. Forward model: MRI.COM-Jpn (1/33 o x 1/50 o ~ 2km) Analysis model: MRI.COM-WNP (1/10 o x 1/10 o ~ 10km) 12 Preliminary results of the prototype 4DVAR assimilation system Backward propagation of adjoint sensitivities Hovmellor diagrams of the adjoint variables along the Kuroshio axis in Jan Assimilation window: 1-month (Jan 1 st Jan 31 st, 2000) Warm water intrusion from the Kuroshio into a coastal area 10-Jan Jan Feb-2000 Obs (SST) 3DVAR 4DVAR 14 (5-day window) (10-day window) Sea level anomaly at the southern coast of Japan Tosashimizu 2. Uragami 3. Toba 4. Irozaki Black: Observation Blue : 3DVAR (5-day window) Red : 4DVAR (10-day window) Location of tide gauge stations 15 Sea level anomaly (cm) Enhancement of temperature variability in the coastal area Difference in subsurface (100m) temperature variability between 3DVAR and 4DVAR 16 Summary (1): 4DVAR results 17 MRI has been developing a 4DVAR scheme as an assimilation part of the next generation coastal system in JMA. Results of a prototype 4DVAR system were evaluated focusing on nearshore processes. A warm water intrusion event caused by a disturbance propagating along the Kuroshio was successfully reproduced by the 4DVAR system. The 4DVAR system improve high-frequency variability in coastal areas compared with 3DVAR. The 4DVAR scheme is suitable for the next-generation coastal system. Summary (2): MOVE/MRI.COM-Jpn System name: MOVE/MRI.COM-Jpn High-resolution (~2km) coastal model (MRI.COM-Jpn) Incremental 4DVAR assimilation using the 10km analysis model Objectives: Upgrade the present ocean information in the open ocean Nowcasting and forecasting coastal phenomena Assessment strategies: Further evaluation using coastal obs such as tide gauge, buoy and HF radar Applications: Kyucho and Abiki (potential) forecast Oil spill forecast and rescue activity in case of ship disaster Offline coupling with a biogeochemical model (for research use) Future issues: How to initialize the high-resolution model Tide assimilation method Computational cost for the 4DVAR scheme 18 19 Extra Slides 20 Computational cost for the prototype 4DVAR system OGCM: MRI.COM-WNP (~10km) Number of grid: 673 x 442 x 54 Time step: 10 min. Assimilation window: 10 days Observation: Satellite altimeter Gridded SST In situ T-S profiles Elapsed time by Earth Simulator2 (SX-9): 5hours (30 iterations, 64MPI) Elapsed time by MRI super computer (SR-16000): 4.5hours (10 iterations, 112MPI) 21 JMA web: (figure and commentary) To general users (nowcast/forecast fields: SST, T, S, U, V, SSH, figure commentary) JMA leased line: (Gridded point value: GPV; Grib/Grib2 format) To Japan Coast Guard (JCG) : (nowcast/forecast fields: U,V) To Japan Ministry of Defense (JMSDF) : (nowcast U,V,T,SSH) To Japan Meteorological Business Support Center (JMBSC) (nowcast/forecast fields: SST) JMA radio facsimile broadcast (JMH): (figure) To Japan Meteorological Business Support Center (JMBSC) (nowcast/forecast SST) To ships in the seas: (nowcast: U, V, T) NEAR-GOOS data server: (figure, GPV) To registered users (nowcast fields, of text file with the NEAR-GOOS format) Products and Users 22 Fisheries Temp. front and Current at m depth Safe and economical shipping SST, Strong Current and Sea Ice Recreation Temperature and current (near coast) Rescue activity in case of ship disaster SST and Current (high resolution) at surface Prediction of oil drift SST and Current (high resolution) Coastal flood Current variation and warm eddy approaching Purposes of Ocean Information Service in JMA