Mesoscale variability in time series data: Satellite-based estimates for the U.S. JGOFS Bermuda Atlantic Time-Series Study (BATS) site

• Published in: Journal of Geophysical Research. C. Oceans 2002-08, Vol.107 (C8), p.7-1-7-21
• Main Authors: Glover, David M., Doney, Scott C., Mariano, Arthur J., Evans, Robert H., McCue, Scott J.
• Format: Article
• Language: English
• Subjects: Marine; mesoscale; remote sensing; SSH; SST; time series; variability
• ISSN: 0148-0227; 2169-9275; 2156-2202; 2169-9291
• DOI: 10.1029/2000JC000589

Objectively analyzed fields of satellite sea surface temperature (SST, advanced very high resolution radiometer (AVHRR) Pathfinder) and sea surface height anomaly (SSHA, combined TOPEX/Poseidon–ERS‐1/2) are used to characterize, statistically, the mesoscale variability about the U.S. Joint Global Ocean Flux Study (JGOFS) Bermuda Atlantic Time‐Series Study (BATS) site. These results are applied to the in situ BATS time series data and a local one‐dimensional (1‐D) physical upper ocean model to better understand the contribution of mesoscale eddies to the time series record and the model‐data mismatch. Using a low‐pass spatial filter, we decompose the anomalies from the seasonal cycle into two components: the large‐scale, regional climate variability and a mesoscale signal. The mesoscale SST and SSHA fields are positively cross‐correlated at a statistically significant level, consistent with near‐surface isotherm displacements for cyclonic and anticyclonic eddies. The results from time‐lagged cross‐correlation analysis show that detectable eddy signatures exist in the in situ SST data and that eddies are a noticeable (∼10%) but not dominant error source for the 1‐D model solution. Several factors may be at work: the 1‐D model captures a more regional signal, whereas the BATS in situ data include small‐scale spatial heterogeneity; the satellite data and 1‐D model are indirectly coupled via the National Centers for Environmental Prediction (NCEP) reanalysis forcing data; and the satellite‐based mesoscale variability estimates are also missing specific events because of the sparse space‐time sampling of a polar orbiting, visible/infrared wavelength sensor. The mesoscale eddy cross‐correlation signature did not show up clearly in a similar analysis conducted on the original anomaly fields, highlighting the fact that climate scale variability needs to be carefully removed to isolate the eddy signature.