Dynamic models for Icelandic meteorological data sets

Abstract

Dynamic models are derived from pairs of time series of daily surface and 500 hPa heightobservations in the Icelandic region in order to interpret these data. The models describe themotion of data points in the phase plane spanned by a pair of variables. For each pair, aregression model of first order (R-model) is derived, which assumes linear motion in the phaseplane. The coefficients of the corresponding Fokker’Planck equation (FP) are also derivedfrom the data. This equation describes changes of the probability density distribution of thedata pair and takes nonlinear aspects into account. The data are interpreted in the light ofboth models. The R-model yields a mean rotation and convergence which characterize the basictype of motion inherent in a data set. It is found in all cases that the timescale of contractionis shorter than the “dynamic” timescale linked to the rotation. The FP equation yields a meanmotion which resembles quite often that in the R-model as well as a diffusive part. The R-modelis not satisfactory in some cases which involve the meridional surface winds. For example, southerlies (northerlies) and temperature rotate clockwise (anticlockwise).Meridional and zonalsurface winds rotate anticlockwise in a southerly flow, but clockwise in a northerly one.Therefore, new aspects of the Icelandic meteorology can be extracted form the data by use ofthe FP equation.