Use of FerryBox surface temperature and salinity measurements to improve model based state estimates for the German Bight

Abstract

The potential of FerryBox sea surface temperature (SST) and salinity (SSS) measurements for the improvement of state estimates in the German Bight is investigated. The paper quests the hypothesis that the parallel analysis of remote sensing and FerryBox data, as well as data simulated by a numerical model, could increase the efficiency of using the information contained in the FerryBox data when producing state estimates. The analysis uses output from a 3-D primitive equation numerical model, up-to-date remote sensing products, and classical in-situ observations as complementary information. A Kalman filter approach is applied to extrapolate one-dimensional FerryBox data acquired along the ferry route from Cuxhaven to Immingham to larger two-dimensional areas. The method makes use of a priori information about the background statistics provided by a numerical model. Maps of extrapolation errors are presented. The impact of the special FerryBox sampling with a revisit time of typically 36 h is investigated based on synthetic data. In particular the aliasing problem associated with the M2 tidal signal is discussed. It is demonstrated that reasonable extrapolation errors can be achieved with a linear interpolation method in combination with a filter operation. Real FerryBox measurements acquired in 2007 are used for assimilation experiments with a 3-D primitive equation model. A standard optimal interpolation (OI) technique is applied for this purpose. The required background statistics are estimated from a free run performed with the model. It is demonstrated that an assimilation of FerryBox SST data leads to a qualitative improvement of the SST state estimates over large areas. Our analysis showed that the natural variability of SSS along the FerryBox track is small compared to the measurement errors and the errors resulting from the specific FerryBox sampling. The use of FerryBox SSS data in an assimilation system is therefore more demanding than the use of the respective SST data. Comparisons with independent observations demonstrate that the improvements in the SSS state estimates are more pronounced for synoptic and short time events.