-
IORAS activities for DRAKKAR in 2006General topic: Development
of long-term flux data set for interdecadal simulations with
DRAKKAR models
Task: Using 2ePDF to derive homogeneous time series of turbulent
fluxes at least in the North Atlantic.
What means homogeneous?
Sampling in 1960s+ should be as bad as before WW2 Impact of
parameterizations should be minimized to the extent possible Impact
of changes in observational practices should be minimized as
well
Result expected: The fluxes will not be correct, but their
variability might be reliable
Data: all VOS data from ICOADS (1880-2005)
-
RandomVOS-like sub-samplingReal-timeVOS-like sub-samplingRandom
sampling errorTotal sampling errorObjectiveanalysis error6-hourly
NWP individual variables Re-computation of surface fluxesusing bulk
formulae Bulkparameteri-zationsHomogenization of sampling in
space:
-
Sampling errors in fluxes:Double
exponentialdistribution(2ePDF)10-3
-
Homogenization of sampling in time: MC-sub-sampling for
n=7,15,25,50 per 5x5, 2x2 boxper 5x5 box
-
Parameterizations: COARE-3.0 (no skin, no free convection,
instead - 4/3 convection scheme for calm winds, no mature
turbulence scheme)Parameters: Wind Beaufort only, WMO1100 => to
Lindau (1995) scale SST buckets only, no engine intakes Air
temperature all
Production iterative run for every month: 1st guess - 2eWPDF
derivation of monthly means for 10-degree boxes south of 40N and
for gerrymander network north of 40N 2ePDF computation for 5x5
degree boxes, if the PDF does not fit at 95% level, use the 1st
guess for the thresholds on the parameters then repeat the
procedure again
-
123-yr (1880-2002) climatology
-
Climatological differences between 2ePDF-derived and
traditionally averaged fluxes
-
Linear tends, sensible + latent, winter
-
Regional time series of winter sensible + latent heat
flux1234
-
Winter (JFM) EOFs
-
Link of the leading mode in turbulent fluxes with NAO
index:30-yr running correlation changes over time considerably
-
NAO is a dominant signal in driving NA circulation?NAO-based
reconstructions of forcing are possible
Eden and Willebrand 2001, Eden and Jung 2001Eden and Willebrand
2001, J. ClimateEden and Jung 2001, J. ClimateHEAT FORCING IS
IMPORTANTReconstruction for 1865-1997:MHT, 48N
-
Canonical correlation of the heat fluxes with HADSLP for
different periods
-
Conclusions:5-degree 123-yr homogeneous time series of
turbulentfluxes (1880-2002) were derived using 2ePDF integration
multi-decadal variability is visible
Product is available at 2-degree resolution for 1910-1940 and
1948-2005
Individual variables are also available, so that use offorcing
based on bulk formulae is possible
During the period 1915/20 1950/55 winter surface turbulent
fluxes are just loosely connected with NAO, being closely related
to NAO during the decades before and after this periodForcing
long-term runs of ocean GCMs in [still] coarse resolutionPotential
application
-
Other activities: validation of NOAA blendedsatellite winds 11
satellites, 0.25x0.25 resolution, global, 1987-onwards,
6-hourly effectively from 1994
Zhang et al. 2006
-
Validation of NOAA blended satellite winds 06:0018:00Much more
detailed structure than ERA40
More reliable tropical winds than in both NCEP and ERA40
Good representationof coastal patterns
-
Computation of humidity:Strategy:
Statistical multiregression approach (not a neural network yet)
for the decade of 1960s Deviations from seasonal climatology are
considered Exponential functions + polynomials for Ta, V, SLP Done
locally, for every month and every box
-
Effect in latent heat flux estimates:RMS = 7 W/m2
Slope = 0.996
Intercept = + 1.4 W/m2
-
Precipitation from NWP:Threshold on small precipitation,
associated with the spin-up effect: 0.02 0.1 mm/day up to 10-12
mm/month in subpolar latitudes CMORPH 0.25x0.25, daily, satellites
+ model, 2001-onwards
-
WindsCardone (Oceanweather Inc.) kinematically corrected winds,
up to 0.2 degree, 3-hourly (wind workstation)