Section 6.6: Conclusions
117
contribution that can be made by statistical analysis of climate observations.
At the outset, there is careful analysis and correction of raw observations to
build up computerised historical datasets that are as homogeneous as possible. Then statistical analysis can produce information on various aspects of
climate variability. This chapter has illustrated the use of EOF analysis to
analyse spatial patterns of worldwide SST variability and to describe the temporal variations of those patterns. Diagnostic correlation studies describe the
relationship between the SST EOFs, the marine atmosphere, and as a case
study, the relationship of marine variability to remote atmospheric variability
over a continental region, the Sahel. Some of the analysis provides confirrnation of the general applicability of existing theory. Other results provoke
the development of more speculative hypotheses for testing in physical and
dynamical models of varying complexity. One such hypothesis concerns the
existence of a tropic-wide pattern of atmospheric anomalies of which Sahel
rainfall is apart. Another hypothesis concerns the effect of tropical South
Atlantic SST anomalies on Atlantic and West African circulation anomalies. For the results presented in this study, two developments in the future
will be particularly important for furt her advances: Firstly, much improved
data describing the three-dimensional structure of the atmosphere for re cent
years are ab out to become available after are-analysis of the historical data
by numerical modelling centres including ECMWF, and secondly, improved
understanding of the coupled ocean-atmosphere system will result from the
development and application of coupled models of the ocean-atmosphere system to study interactive behaviour and the development of seasonal to multidecadal climate anomalies. To model and understand ocean-atmosphere
connections to continental areas, it mayaIso be necessary in some regions to
improve the representation of land surface - atmosphere inter action , which
mayaIso playa significant role in climate variability (Nicholson, 1988).
117
contribution that can be made by statistical analysis of climate observations.
At the outset, there is careful analysis and correction of raw observations to
build up computerised historical datasets that are as homogeneous as possible. Then statistical analysis can produce information on various aspects of
climate variability. This chapter has illustrated the use of EOF analysis to
analyse spatial patterns of worldwide SST variability and to describe the temporal variations of those patterns. Diagnostic correlation studies describe the
relationship between the SST EOFs, the marine atmosphere, and as a case
study, the relationship of marine variability to remote atmospheric variability
over a continental region, the Sahel. Some of the analysis provides confirrnation of the general applicability of existing theory. Other results provoke
the development of more speculative hypotheses for testing in physical and
dynamical models of varying complexity. One such hypothesis concerns the
existence of a tropic-wide pattern of atmospheric anomalies of which Sahel
rainfall is apart. Another hypothesis concerns the effect of tropical South
Atlantic SST anomalies on Atlantic and West African circulation anomalies. For the results presented in this study, two developments in the future
will be particularly important for furt her advances: Firstly, much improved
data describing the three-dimensional structure of the atmosphere for re cent
years are ab out to become available after are-analysis of the historical data
by numerical modelling centres including ECMWF, and secondly, improved
understanding of the coupled ocean-atmosphere system will result from the
development and application of coupled models of the ocean-atmosphere system to study interactive behaviour and the development of seasonal to multidecadal climate anomalies. To model and understand ocean-atmosphere
connections to continental areas, it mayaIso be necessary in some regions to
improve the representation of land surface - atmosphere inter action , which
mayaIso playa significant role in climate variability (Nicholson, 1988).