FOUR-DIMENSIONAL DATA QUALITY CONTROLL:VARIATIONAL METHOD

Given a set of meteorological data,using the variational data assimilation technique (also called the adjoint method) and some dynamic model we fulfil the dynamical interpolation of the data set.Then comparing the interpolation with the data we 90 on to define the error/mean error ratio for every individual data.The utility of using the error/mean error ratio to identify the gross (or rough) error is demonstrated.The numerical experiment results using the simple Lorentz three-dimensional tnedel are very encouraging.The advantages of this method comparing with traditional data quality control procedure are:(1) we do not need know the covarience of data errors,that probably means that it is more suitable for quality control of climate or satellite data:(2) we make use of prognostic equations,rather than diagnostic ones,which the data are to obey at least approximately;(3) we use both the spatial and temporal continuity check rather than only sPatial one and (4) we carry out the spatial or temporal continuity check by making the interpolation dynamically,rather than by making the linear or statistical interpolation.