last updated 22nd September 05
by 4ecotips.com

Important project seeks ongoing funding

The waters of Monterey Bay on America's Pacific coast, south of San Francisco, holds some vital clues to the development of global warming. Moss Landing Marine Laboratories is closely involved in important research in the area and the organisation's Larry Breaker says that preliminary results suggest "the warming process has not been gradual and continuous but, rather, non-uniform and step-like!"

Now Larry (lbreaker@mlml.calstate.edu) is looking for urgent funding to support his ongoing research, which he confidently believes, will have "important implications for how the process of global warming manifests itself in the ocean." He has completed a preliminary report on the research so far entitled "How the waters of Monterey Bay have warmed during the past century". Here are some technical extracts:

Daily sea surface temperatures (SSTs) have been acquired at the Hopkins Marine Station in Pacific Grove at the southern end of Monterey Bay since 1919. Because of its length, this record is well suited to the study of long-term changes within, and possibly well beyond, the bay.

Three models are employed to explain how the waters of Monterey Bay have warmed over the past century, linear, nonlinear, and step-wise. The linear model is based on linear least-squares regression, the nonlinear model on robust, locally weighted regression (Loess), and the step-wise model on change points identified using Cumulative Sums (CUSUMs).

Three, and possibly five, change points were detected that correspond to known regime shifts. The performance of these models was evaluated using residual sums of squares, and an eigenvalue decomposition of the models after CUSUMs were calculated. Both methods indicate that the step-wise model provides the best fit to the observations.

Because two of the changes were relatively small (< 0.2oC), the 5-step model provides only marginal improvement over a 3-step model. The step-wise model not only provides the best fit to the observations but it is also the most parsimonious. Although several of the step increases were small, all changes were positive yielding an overall increase of greater than 1.1oC since 1920. Because all of the observed changes in temperature were positive, we propose that they could reflect the influence of larger-scale thermal forcing rather than stochastic forcing.

Using CUSUMs, events were detected that correspond to change points and in some cases to regime shifts in the record. CUSUMs produce a distinct pattern that may be characteristic of regime shifts. These patterns, on closer inspection, often contain two change points separated by periods of roughly six months.

In 1939, 1976, and 1999, events were identified that coincide closely with reported regime shifts. An abrupt increase in SST was also detected in 1929 that may correspond to the phase shift in the Pacific Decadal Oscillation (PDO) that occurred circa 1925.

Similar CUSUM patterns were observed in the daily record of SST at Scripps Pier located 550 km south of Monterey Bay. Events in 1939 and 1976 were similar and essentially in phase, whereas the events in 1929 and 1999 were delayed by 1-2 months at Pacific Grove. Although the events in 1929 and 1976 were statistically significant based on the Wilcoxon rank sum test, the events in 1939 and 1999 were not.

Application of standard tests of significance for regime shifts may be problematic because the changes of interest are often relatively small and do not necessarily occur in a single step. However, our results suggest that tests of significance based on pattern recognition could be developed using the CUSUM signatures that characterize regime shifts since they are distinct and repeatable. In conclusion, the evidence for nonuniform heating in Monterey Bay over the past century is strong, and a step-wise model based on regime changes that produce the steps provides a reasonable explanation for how the warming process has occurred.

Larry Breaker writes: "Overall, we observe periods of apparent regime stability that are decadal to interdecadal in scale, separated by regime shifts that do not occur instantaneously but have time scales of six months or so."