DECADAL SCALE SEA-LEVEL VALIDATION OF THE ENSEMBLES ENSEMBLE OF
OCEAN REANALYSES
P. Rogel (1)
(1)
CERFACS/URA1875, 42 Av Coriolis, Toulouse, France, Email: rogel@cerfacs.fr
A new set of global, low-resolution ocean reanalyses
over the ERA40 period (1960-2006) has been produced
for sake of seasonal to decadal climate prediction within
the EC-FP6 ENSEMBLES project. These reanalyses all
use ERA40 winds and fluxes as forcing, the EN3
quality-controlled temperature and salinity profiles data
base built at the UK Met-Office (Ingleby and
Huddelston, 2007), and are strongly constrained by the
observed Sea Surface Temperature. From those
reanalyses produced by several groups (including
ECMWF, INGV, UK Metoffice, and CERFACS), a
common set of variables (temperature, salinity, velocity,
upper ocean heat content, sea level height, mixed layer
depth and depth of the 20°C isotherm) have been stored
with a common NetCDF format, interpolated on a
common 1°x1° resolution, 33 vertical levels grid, and
made
publicly
available
at
http://ensembles.ecmwf.int/thredds/catalog/ocean/catalo
g.html.
This paper presents a comparative validation of those
reanalyses against a set of quality-controled multidecadal tide gauge sea level measurements. This is of
importance for assessing the quality and signification of
climate indicators such as the upper ocean heat content
(see Figure 1).
controlled tide gauge gauge, in about 100 sites, used
elsewhere to reconstruct sea level variations (Llovel et
al., 2009), has been used as an independent validation
database. Although it samples imperfectly the ocean,
especially because most gauges are located near the
coasts, it is sufficiently diverse and variable on the
decadal time scale to be used for comparison with the
reanalyses. A systematic quantitative comparison has
been made using classical statistic tools.
Figure 2: Time evolution of the spatial correlation between reanalysed sea level and
observed sea level at 99 tide gauges locations.
Results show that the quality of the reanalyses is better
at low latitudes, but that a significant amount of
variance of the low frequency signals at mid-latitudes is
well reproduced. The north Atlantic is shown as the
poorest constrained region, though the observational
coverage is relatively important. The reanalyses data set
can also be seen as an ensemble of independent
estimations of the ocean state, and its distribution as an
estimate of the true error distribution of the latter.
Results show that such a “multi-model” ocean
reanalysis has statistically significant better skill than
any of the single reanalyses (Figure 2).
References:
1. Llovel W., A. Cazenave, P. Rogel A. Lombard and M.
Bergé Nguyen, 2009: 2-D reconstruction of past sea
level (1950-2003) using tide gauges records and spatial
patterns from a general ocean circulation model, Climate
of the Past, Vol. 5, No 2, pp 217-227.
Figure 1: Global (80°S-80°N) annual mean upper (0-300 m) ocean averaged temperature
anomalies (wrt the 46-year average) in 20 ENSEMBLES ocean reanalyses.
This can only been done by comparing reanalyses with
a truly independent data set, distributed over the globe,
representative of decadal variations over the multidecadal period of interest (1960-2005). A set of quality
2. Ingleby B. and M. Huddleston (2007): Quality control of
ocean temperature and salinity profiles - historical and
real-time data. J. Mar. Sys., 65,148–175.