Gerß, Joachim

Prädiktion der Ozeantemperatur im räumlichen und zeitlichen Verlauf mit Hilfe Dynamischer Linearer Modelle

Atmospheric and oceanic processes representing important components of theglobal climate system display variability over both space and time. Appliedscientific analysis may be based upon two competing strategies, physicallyderived deterministic modelling versus statistical approaches, from which thelatter is utilized in the present case. Since observations typically constitutelarge data sets that often are spatially and temporally incomplete and exhibitcomplicated interactive structural relationships, traditional space-timemethods are of limited use. Direct specification of the joint space-timecovariance structure often is not possible due to the existence of spatialnon-stationarities and nonseparable space-time interaction. In this paperdynamic linear (state-space) models are developed instead, that model thetemporally dynamic structure in an autoregressive framework and additionallyfeature a spatially descriptive component. In order to handle largeobservational areas, dimension reduction of the spatial field is achieved byuse of empirical orthogonal functions. The method is applied to a data set ofmeasurements of the sea surface temperature in the Northwest European Shelfduring 1983-1992. The observed point measurements are predicted to a grid ofabout 20km grid size (1/3$^o$ in east-west direction and 1/5$^o$ in north-southdirection) by application of the Kalman filter. Unlike other similarspatiotemporal state-space formulations, the presented approach does not demandfor temporally fixed measurement locations. Moreover it allows for a dynamicincorporation of a (large-scale) trend component and an efficient underlyingstep of parameter estimation is involved.