Houtekamer and Mitchell, 1998 ... Clark Rowley1, Patrick Hogan1, Sergey Frolov2, Mozheng Wei1, Prasad Thoppil1, Ole Martin Smedstad3, Neil Barton2, Craig ...
A41G-0139
Ocean Ensemble Forecasting in the Navy Earth System Prediction Capability Clark 1
1 Rowley ,
Patrick
3
INTRODUCTION
Assimilation: next ensemble initial state Xa B(Xa)
prior ensemble forecast
Mozheng
1 Wei ,
Prasad
1 Thoppil ,
Ole Martin
GLOBAL OCEAN ENSEMBLE EXPERIMENTS
observations
X0 B(X0)
Sergey
2 Frolov ,
Forecast Perturbation
3 Smedstad ,
Neil
2 Barton ,
Craig
2 Bishop
Naval Research Laboratory, Oceanography Division Naval Research Laboratory, Marine Meteorology Division Vencore, Stennis Space Center, MS
2
Forecast Perturbation
1 Hogan ,
observations
Xf,new B(Xf,new)
Assimilation: next ensemble initial state
new ensemble forecast
ENSEMBLE METRICS
A pair of 20-member ensemble experiments with a global HYCOM configuration at 1/12° resolution and 32 layers is used to examine the contribution of the perturbed-observation analysis to the forecast ensemble spread. The ensemble members are initialized with 20 different 01 Jan states from a 20-year reanalysis. Members are cycled with the NCODA data assimilation system for 90 d then integrated over a 90 d forecast with NCEP CSFR forcing.
An extended range ensemble forecast system is being developed in the US Navy Earth System Prediction Capability (ESPC). A global ensemble generation capability to support the coupled ESPC ensemble forecast has being developed, and initial assessments are underway. The air/ocean/ice ensemble is generated with parallel perturbed-observation analyses. The ocean/ice systems use the Navy Coupled Ocean Data Assimilation system (NCODA). The resulting analysis perturbations are used to represent uncertainty in the initial conditions of a global ocean forecast ensemble using the Hybrid Coordinate Ocean Model (HYCOM).
Time series of the global ensemble basin average standard deviation from the unperturbed observation (black) and perturbed observation (red) experiments for the 90 d data assimilation period and the 90 d forecast period.
PERTURBED OBSERVATION ANALYSIS Perturbed Observation Ensemble Update Houtekamer and Mitchell, 1998 Ocean Obs Ocean QC
HYCOM
Observation perturbations
Control run forcing Space-time deformation
Sequential Incremental Update Cycle Ocean Obs Analysis-Forecast-Analysis Sequential Incremental Update Cycle Ocean Obs Analysis-Forecast-Analysis SST: Ship, Buoy, AVHRR Sequential Incremental Update Cycle (GAC/LAC), GOES, AMSR-E, Obs Ocean MSG, AATSR Ocean QC Analysis-Forecast-Analysis SST: Ship, Buoy, AVHRR InnovationsIncremental Update Cycle Sequential (GAC/LAC), GOES, AMSR-E, Obs Temp/Salt Profiles: XBT, CTD, Ocean MSG,Buoy AATSR Argo Float, Ocean QC Analysis-Forecast-Analysis SST:(Fixed/Drifting), Ship, Buoy, AVHRR InnovationsIncremental Update Cycle Gliders Sequential (GAC/LAC), GOES, AMSR-E, Obs Temp/Salt Profiles: XBT, CTD, Ocean MSG, AATSR Argo Float, Buoy (Fixed/Drifting), Ocean QC SSH: Altimeter, T/S profiles Analysis-Forecast-Analysis SST: Ship, Buoy, AVHRR InnovationsIncremental Update Cycle Gliders 3D MVOI Sequential (GAC/LAC), GOES, AMSR-E, Obs Temp/Salt Profiles: XBT, CTD, Ocean Sea Ice: SSM/I Increments or MSG, AATSR Argo Float, Buoy (Fixed/Drifting), Ocean QC SSH: Altimeter, T/S profiles Analysis-Forecast-Analysis SST: Ship, Buoy, AVHRR Gliders Analysis 3D MVOI Innovations (GAC/LAC), GOES, AMSR-E, Temp/Salt Profiles: XBT, CTD, Sea Ice: SSM/I Increments or MSG, AATSR Argo Float, Buoy (Fixed/Drifting), Ocean QC SSH: Altimeter, T/S profiles SST: Ship, Buoy, AVHRR Innovations Gliders Analysis 3D MVOI (GAC/LAC), GOES, AMSR-E, Temp/Salt Profiles: XBT, CTD, Ocean QC Sea Ice: SSM/I Increments or MSG, AATSR Argo Float, Buoy (Fixed/Drifting), SSH: Altimeter, T/S profiles HYCOM First Guess Innovations Gliders Analysis 3D MVOI Temp/Salt Profiles: XBT, CTD, Sea Ice: SSM/I Argo Float, Buoy (Fixed/Drifting), SSH: Altimeter, T/S profiles HYCOM Increments or First Guess Gliders Analysis 3D MVOI Sea Ice: SSM/I SSH: Altimeter, T/S profiles HYCOM Increments or First Guess Analysis 3D VAR Sea Ice: SSM/I HYCOM Increments or First Guess Analysis HYCOM First Guess HYCOM First Guess
Sea surface height ensemble standard deviation for the initial condition, after 30 d and 90 d of cycling with unperturbed observations, and after 90 d of a free run forecast. No obs perturbation With obs perturbations m mean
3D VAR
m s.dev.
NCODA performs a 3DVar analysis of temperature, salinity, geopotential, and vector velocity using remotely-sensed SST, SSH, and sea ice concentration, plus in situ observations of temperature, salinity, and currents from ships, buoys, XBTs, CTDs, profiling floats, and autonomous gliders. Sea surface height is assimilated through synthetic temperature and salinity profiles.
m
SSH
Ensemble of analysis/forecast cycles
Real profile
Perturbations to the surface and profile observations use random samples from a normal distribution scaled by the observation error standard deviation, which combines instrument and representation error. Perturbations to synthetic profiles are generated by supplying perturbed surface inputs to the synthetic profile system.
mean
m
°C s.dev.
a
°C
100 M T
Synthetic profile
Ensemble rank histograms at three depths and area-average ensemble bias and RMS error for profile temperature observations in the North Indian Ocean for the first month of the forecast period for the ensemble experiments with unperturbed (left) and perturbed (right) observation analyses.
North Indian Ocean sea surface height and 100 m temperature ensemble mean and standard deviation after 90 d of cycling with the NCODA data assimilation system with unperturbed (left) and perturbed observations (right).
North Indian Ocean sea surface height perturbations (deviations from ensemble mean) for members 1-4 of the perturbed-observation ensemble after 90 d of cycling with the data assimilation system. This work is funded by the Office of Naval Research under the Naval Research Laboratory project Decision Making using Global Probabilistic and Uncertainty Forecasts.
DECEMBER 2016 | WWW7320.NRLSSC.NAVY.MIL
