The relationships between Argo Steric Height and AVISO Sea Surface Height Phil Sutton 1 Dean Roemmich 2 1 National Institute of Water and Atmospheric Research, New Zealand 2 Scripps Institution of Oceanography, UCSD Ocean Surface Topography Science Team: 4 th Argo Science Workshop Venice September 2012
Motivation Want to compare and contrast statistics of SH and SSH. Want to understand the relationship of SH and SSH Investigate the steric component of SSH and its depth distribution. Where possible, use co-located data so as to remove any effects of mappings of different samplings. Take advantage of some sub-topex hydrographic data to give near-perfect spatial co-location and full depth data. Investigate regional patterns in the SSH/SH relationship. Argo's Objectives The data will enhance the value of the Jason altimeter through measurement of subsurface temperature, salinity, and velocity, with sufficient coverage and resolution to permit interpretation of altimetric sea surface height variability. Standard deviation (cm) of binned monthly time-series of raw SSH, SH and SSH-SH.
Data: AVISO: Along track data unfiltered, unsubsampled (6km) Gridded product (for comparison with XBT) Co-located Argo: Good coverage Not truly co-located Not truly repeats (mean field issues?) Not full depth Co-located Hydrographic Not many samples Well co-located in space (sub topex) Full depth XBT Not co-located Good time extent 0-800m
Approach: Analysing scatter plots of ΔSSH and ΔSH. Δs are calculated either: i) relative to a mean ii) between measurements for the repeat hydrography Use co-located Argo/SSH for areal and decadal analyses Use co-located hydrographic/ssh for specific tracks Use XBT time series to investigate temporal stability.
The relationship between SH and SSH SSH = = SH SH 0 0 2000 SH + Mass + SH 2000 = α SH + Mass Assume: 2000. 0 2000 Then: SSH So: = SH + ) + y = mx + c 0 2000.(1 α Slope = ( 1+ α) Intercept = Mass Mass More precisely, α is a measure of deep, correlated changes in steric height and mass, while the intercept indicates the uncorrelated deep steric and mass changes. Slope 2000 = 1.21 Intercept = -0.02
Principal component analysis vs linear regression Looking at the same data in two ways: Green = PCA: slope of ΔSSH vs ΔSH is the reciprocal of the slope of ΔSH vs ΔSSH. Red = least squares: both slopes less than 1. Principal Component Analysis minimises the sum of the squared orthogonal distances between the points and the best fit line. In practice, this is achieved by rotating the plots until Σy 2 is minimised.
Repeat hydrography and AVISO Four repeat sections around NZ Sub -Topex Full depth Can look at differences between surveys: no mean in analysis Calculate slope and intercept for ΔSSH vs ΔSH 0_1000, ΔSH 0_2000, and ΔSH 0_bottom
Line 1: northeast of NZ Calculate slope and intercept for ΔSH 0_1000, ΔSH 0_2000, ΔSH 0_bottom Error bars show ±(range in SSH/2), where the range in SSH is the difference between the SSH values prior to and after the hydrographic sampling. Most of the scatter from near the slope
Repeat hydrography and AVISO Line W1: 0/1000: Slope = 1.45 0/2000: Slope = 1.12 0/bottom: Slope = 1.10 Line CP: 0/1000: Slope = 2.32 0/2000: Slope = 1.42 0/bottom: Slope = 1.10 Line 1: 0/1000: Slope = 1.16 0/2000: Slope = 0.92 0/bottom: Slope = 0.91 Line 9: 0/1000: Slope = 1.12 0/2000: Slope = 0.92 0/bottom: Slope = 0.92
Time dependence: 1) Argo era Monthly analysis of subtropical and subantarctic waters in SW Pacific Subtropical Water Subantarctic Water Slope and intercept pretty stable in time. Annual cycle in intercept. Intercept largely constant across area.
Time dependence: 2) XBT 77 transects overlap the AVISO timeseries from 1991 to present Slope and intercept pretty stable. Some structure but Intercept again stable over latitude band.
What about that slope < 1 NE of NZ? AVISO vs Hydrographic data: Line 1 slope (0/2000)= 0.92 AVISO vs Argo 0/2000m co-located Slope = 1.08 (204 data pairs) Latitude dependence from XBT line:
What about that slope < 1 NE of NZ? Co-located Argo. 9 x5 bin-averages of co-located points (l=96 months); Apply 5 (lon) x 3 (lat) smoothing Number of data pairs in 9X5 bin Does the correlated baroclinic and barotropic variability near North Cape suggest conversion of barotropic to baroclinic energy by scattering off topography? (e.g. Hill 2009)
Conclusions SSH and SH 0-2000 do not always close the system. Barotropic/mass changes can be significant. The slope < 1 off the NE coast of NZ would indicate mass/barotropic signals anticorrelated with the SH changes. -flow interactions with ridges, internal tides? The boundary current regions introduce the most scatterunfiltered, unsubsampled (6 km) SSH data showed less scatter than filtered, subsampled (25 km) data For much of the subtropical water in the SW Pacific, the Steric Height to 2000m captures almost all of the SSH changes. The Southern Ocean indicates an extra 40% contribution in SH below 2000m.