3rd Argo Science Workshop @ Zhejiang Hotel, Hangzhou, China on 2009/03/25 (Wed) 27 (Fri) Impact evaluation of Argo and sea surface altimeter data for the reproducibility of FRA-JCOPE: An ocean prediction system for the seas around Japan H. Yoshinari 1, A. Okuno 1, T. Setou 1, D. Ambe 1, Y. Miyazawa 2, FRA-JCOPE Group 1,3,4 1: National Research Institute of Fisheries Science, Fisheries Research Agency 2: Frontier Research Center for Global Change, JAMSTEC 3: Hokkaido National Fisheries Research Institute, Fisheries Research Agency 4: Tohoku National Fisheries Research Institute, Fisheries Research Agency
JCOPE Data JCOPE1 transfer (JAMSTEC) system : easy to transfer in situ data to GTS Argo by Web-GUI visualization FRDATA Increase of realtime availability of in situ data Increase of accuracy of nowcast and forecast Feedback system : return profits to observer by Web-GUI Increase of motivation and value of observation Increase of sustainability of observation (Courtesy of Dr. K. Komatsu)
FRA-JCOPE Ocean Forecast System Using various observational data for assimilation e.g., 1) Argo T & S Data 2) Satellite Sea Surface Altimeter Data (SSH) Purpose: To evaluate each data (Argo & SSH) assimilation impact for reproducibility of FRA-JCOPE Useful & Important Information for Improvement of #) Reproduction Accuracy and #) Prediction Accuracy
Data (Outputs): FRA-JCOPE Reproductions (Reanalyses): #) Duration: 2003/01/01 2006/12/31, Daily Mean #) Specifications: 1) Argo & SSH Assimilated (Standard) 2) Argo Non-assimilated 3) SSH Non-assimilated Observational data for comparison: #) Argo T & S #) WOA05 Monthly Climatology T & S #) Ambe07 (Ambe et al., 2009) Weekly Mean Sea Surface Velocity Data [ #) Japan Public Office s R/V T & S: Independent Data ]
How to evaluate: Vertical FRA-JCOPE Water Mass (Analysis) Structures Regions Distributions of profile (T & S) Compare T & S @ SubTropical Region (STR) #)θ- S Diagrams #) Time Series of T & S @ 25.2σ θ : Representative Density + : Argo of North Pacific Subtropical Mode Water data assimilated into FRA-JCOPE in 2006/12 : Fisheries Research Institute : Others Sea Surface Velocity KER Calculate KSR Correlation Coefficients for Vectors (Fofonoff and Hendry, 1985; Yamamoto et al., 2002) @... #) KuroShio Region STR (KSR) #) Kuroshio Extension Region (KER) (Yoshinari et al., 2008)
Monthly Meanθ- S Diagrams in STR SSH Non-assimilated Argo Non-assimilated Standard Argo WOA05
Time series of monthly mean T & S on 25.2σ θ in STR Rank of Reproducibility: 1. SSH Non-assimilated or Standard (Statistically Insignificant Differences) 2. Argo Non-assimilated Temp. Sal. Argo Standard Argo Non-assimilated SSH Non-assimilated
How to calculate correlation coefficients of vector values Consider 2 different vectors: VL1: (u 1,v 1 ), VL2: (u 2,v 2 ) Correlation coefficient (C) between VL1 and VL2 is calculated as follows; Rotation angle from VL1 to VL2, that makes correlation coefficient the maximum, called Veering Angle (VA), is calculated as follows; Note: Superscript bar denotes spatial (or temporal) average.
Evaluation of Sea Surface Velocity in KSR Rank of Reproducibility: 1. Standard or Argo Non-assimilated 2. SSH Non-assimilated VA 0 Standard Argo Non-assimilated SSH Non-assimilated Reproduced Well
Evaluation of Sea Surface Velocity in KER Rank of Reproducibility: 1. Standard or Argo Non-assimilated 2. SSH Non-assimilated VA 0 Standard Argo Non-assimilated SSH Non-assimilated Reproduced Well
Time series of monthly mean T on 25.2σ θ in STR : Comparison with Independent (Non-assimilated) Data Temp. Independent Data Standard Argo Non-assimilated SSH Non-assimilated
Time series of monthly mean S on 25.2σ θ in STR : Comparison with Independent Rank of Reproducibility: (Non-assimilated) Data 1. SSH Non-assimilated or Standard (Statistically Insignificant Differences) Sal. 2. Argo Non-assimilated Independent Data Standard Argo Non-assimilated SSH Non-assimilated
Summary and Discussion Rank of Reproducibility Water Mass Structures in STR: SSH Non-assimilated =~ Standard >>> Argo Non-assimilated Assimilating Argo data is essential to reproduce T & S Sea Surface Velocities in Both KSR & KER: Standard =~ Argo Non-assimilated >>> SSH Non-assimilated Assimilating SSH data is essential to reproduce Sea Surface Velocity
T & S s Insignificant Difference between Standard and SSH Non-assimilated One of data assimilation procedures might not work well inappropriate Vertical Projection Coefficients are adopted?
(Following slides are for questions & back up)
The velocity field by combination of altimeter & buoy (Ambe, 2007) Estimate of temporal mean velocity field This method is based on Uchida and Imawaki (2003). Absolute velocity V d (x,t) from sea-surface drifting buoy Temporal anomaly V a (x grid,t grid ) from satellite altimeter Temporal mean velocity Buoy-derived velocity was calculated every sampling position. Mean components at a (x grid ) Improved points from UI03 Reducing variation of s temporal and spatial scale Increasing referred samples of
The velocity field by combination of altimeter & buoy (Ambe, 2007) Data The formula to estimate Ekman current (Niiler et al., J.G.R., 2003)
The velocity field by combination of altimeter & buoy (Ambe, 2007) The mean (geostrophic) velocity field around the Kuroshio (1993 1999, reference period of AVISO dataset)