Optimal in situ measurement networks for ocean data

This paper investigates four methods for calculation of fixed measurement positions for acquiring knowledge about an ocean area by observation and model based analysis. The qualities and limitations of the methods, applied to a rather complicated ocean area, is studied. We investigate two methods based on covariations in ocean data; (i) an extension of the data assimilation based method by [1] with the tabu search method by [2], and (ii) an extension of the footprint method by [3] by including time-lags and calculation of the percentage of coverage for each position in the ocean area that is investigated. Further, we study variance map and maps of mean values for the same purpose. The approach of using mean value maps has not been reported in the literature for ocean data. The importance of each measurement position is evaluated in order to decide how many sensors that are necessary to monitor an ocean area. The two methods based on current covariations show considerable differences in calculated positions due to the different backgrounds for the two methods. The SINTEF ocean model SINMOD, set up for an area at the Rataren marine installation close to Frθya, Norway, is used in a case study. This is a shallow area where the current pattern is rather complex due to the bathymetry.