@PHDTHESIS\{IMM2004-03370,
    author       = "S. Lophaven",
    title        = "Design and analysis of environmental monitoring programs",
    year         = "2004",
    school       = "Informatics and Mathematical Modelling, Technical University of Denmark, {DTU}",
    address      = "Richard Petersens Plads, Building 321, {DK-}2800 Kgs. Lyngby",
    type         = "",
    note         = "Supervised by Assoc. Prof. Helle Rootzen",
    url          = "http://www2.compute.dtu.dk/pubdb/pubs/3370-full.html",
    abstract     = "This thesis describes statistical methods for modelling space-time phenomena. The methods were applied to data from the Danish marine monitoring program in the Kattegat, measured in the five-year period 1993-1997. The proposed model approaches are characterised as relatively simple methods, which can handle missing data values and utilize the spatial and temporal correlation in data. Modelling results can be used to improve reporting on the state of the marine environment in the Kattegat. 

The thesis also focus on design of monitoring networks, from which geostatistics can be successfully applied. Existing design methods are reviewed, and based on these a new Bayesian geostatistical design approach is suggested. This focus on constructing monitoring networks which are efficient for computing spatial predictions, while taking the uncertainties of the parameters in the geostatistical model into account. Thus, it serves as a compromise between existing methods. 

The space-time model approaches and geostatistical design methods used in this thesis are generally applicable, i.e. with minor modifications they could equally well be applied within areas such as soil and air pollution.

In Danish:

Denne PhD afhandling beskriver statistiskemetoder til modellering af f{\ae}nomener i tid og rum. Metoderne er anvendt p{\aa} data fra det danske marine overv{\aa}gningsprogram i Kattegat, der er m{\aa}lt i perioden 1993-1997. De foresl{\aa}ede modeller er karakteriseret ved at v{\ae}re forholdsvis simple metoder, der kan h{\aa}ndtere manglende observationer, samt udnytte den spatielle og tidslige korrelation i data. Model resultaterne kan anvendes til at forbedre viden og afrapportering af milj{\o}tilstanden i Kattegat.

PhD afhandlingen omhandler ligeledes design af m{\aa}leprogrammer, s{\aa}ledes at effektiv modellering ved brug af geostatistik muligg{\o}res. Der gives en oversigt over eksisterende design metoder, og p{\aa} baggrund af disse er foresl{\aa}et en ny Bayesiansk design metode. Denne fokuserer p{\aa} at konstruere m{\aa}leprogrammer, der kan anvendes til effektiv beregning af spatielle prediktioner ved brug af en geostatistisk model, og under hensyntagen til usikkerhederne i modellens parametre. Den Bayesiansk design metode kombinerer s{\aa}ledes eksisterende design metoder.

De statistiske metoder til modellering af f nomener i tid og rum, samt de geostatistiske design metoder, der anvendes i PhD afhandlingen, er generelle metoder, og kan med sm{\aa}{\ae}ndringer anvendes indenfor andre milj{\o}mr{\aa}der, som f.eks. jord- og luftforurening."
}