Abstract
Ocean bottom pressure variability is analyzed from three monthly products available from (1) the Gravity Recovery and Climate Experiment (GRACE), (2) sterically corrected altimetry, and (3) from a forward run of the German part of the Estimating the Circulation and Climate of the Ocean (GECCO-2) model. Results lead to an approximate error estimate for each of the ocean bottom pressure (OBP) maps under the assumption of noncorrelated errors among the three products. The estimated error maps are consistent with the misfits of individual fields against OBP sensor data, with the caveat that a general underestimation of the signal strength, as a common, correlated error in all products, cannot be recovered by the method. The signal-to-noise ratio (SNR) increases in all products, when a 3 month running mean filter is applied. Using this filter, we estimate globally averaged errors of 8.6, 11.1, and 5.7 mm of equivalent water height for GRACE, nonsteric altimetry, and GECCO2, respectively. Based on resulting uncertainties, a new OBP product is being produced by merging all three data sets. When validated with bottom pressure observations this new OBP product has a 20% increased SNR compared to the best individual product (GECCO2-ref). Estimated total ocean mass variations explain a considerable part of OBP variability with a SNR above 1 in most of the ocean. In some regions the nonuniform part is weaker than the estimated error. However, most dynamic ocean models are designed to reproduce only the nonuniform, dynamic, OBP variability, but do not accurately describe total mass variability.
