Abstract
At the sea surface, the atmosphere and the ocean exchange momentum, heat and freshwater. Mechanisms for
the exchange are wind stress, turbulent mixing, radiation, evaporation and precipitation. These surface fluxes are
characterized by a large spatial and temporal variability and play an important role in not
... read more
only the mean
atmospheric and oceanic circulation, but also in the generation and sustainment of coupled climate fluctuations
such as the El Niño/La Niña phenomenon. Therefore, a good knowledge of air-sea fluxes is required for the
understanding and prediction of climate changes.
As part of long-term comprehensive atmospheric reanalyses with `Numerical Weather Prediction/Data
assimilation' systems, data sets of global air-sea fluxes are generated. A good example is the 15-year
atmospheric reanalysis of the European Centre for Medium--Range Weather Forecasts (ECMWF). Air-sea flux
data sets from these reanalyses are very beneficial for climate research, because they combine a good spatial and
temporal coverage with a homogeneous and consistent method of calculation. However, atmospheric reanalyses
are still imperfect sources of flux information due to shortcomings in model variables, model parameterizations,
assimilation methods, sampling of observations, and quality of observations. Therefore, assessments of the errors
and the usefulness of air-sea flux data sets from atmospheric (re-)analyses are relevant contributions to the
quantitative study of climate variability.
Currently, much research is aimed at assessing the quality and usefulness of the reanalysed air-sea fluxes. Work
in this thesis intends to contribute to this assessment. In particular, it attempts to answer three relevant questions.
The first question is: What is the best parameterization of the momentum flux? A comparison is made of the wind
stress parameterization of the ERA15 reanalysis, the currently generated ERA40 reanalysis and the wind stress
measurements over the open ocean. The comparison reveals some clear differences in the mean drag coefficient.
In addition, this study has indicated that progress has been made from the ERA15 to the ERA40 reanalyses by
replacing the model parameterization with a constant Charnock parameter with one which depends on the sea
state.
The second research question is whether comparison of the response of an ocean model with ocean
observations can be exploited to assess the quality of air-sea fluxes of the ERA15 reanalysis. To answer this
question in a systematic way an inverse modeling approach is adopted using a four-dimensional variational data
assimilation (4DVAR) scheme. Firstly, the functioning of the 4DVAR system is demonstrated from identical twin
experiments. These experiments reveal that in the equatorial Pacific, a large reduction in wind-stress and
upper-ocean temperature misfits can be achieved using an assimilation time window of eight weeks. It is
concluded that the usefulness of inverse ocean modeling technique for global surface flux assessment is limited.
The main merit of the developed ocean 4DVAR scheme will be to diagnose errors in the ocean analyses of the
ocean model.
The last research question is: are the ERA15 fluxes useful for the study of regional patterns of climate variability?
The climate mode of consideration is the Antarctic Circumpolar Wave. This study stresses the importance to
have the right climatological forcing conditions to assess time scales of climate variability and it confirms the
usefulness of ERA15 air-sea fluxes as ocean model forcing fields to study climate variability on the interannual
time scale.
show less