Stochastic Modeling and Global Warming Trend Extraction For Ocean Acoustic Travel Times

A possible indication of the existence of global climate warming is a negative trend for the travel time of an acoustic pulse along a fixed long path, or paths, in the ocean over a period of many years. The goal of this report is the development of methods specifically for determining the presence of a long term trend for climate change from a temporal sequence of measurements of acoustic propagation times. Robust statistical methods for determining whether a significant trend is present in a given set of time series data have been developed and, for illustration, applied to some specific traveltime time series generated by the MASIG and GFDL ocean models. In this report we consider line + noise and ARIMA statistical models. We show that if the time series are long enough, somewhat over 20 years, then series such as those simulated by the MASIG and GFDL models can be classified reliably as line + noise when this is the case. However, it is shown that the results are considerably different for the two ocean models under consideration and that these models can not currently be relied upon by themselves to predict global warming. Experimental data is most certainly needed, not only to measure global warming itself, but to help improve the ocean model themselves. (AN) Prepared in cooperation with Southern Methodist Univ., Dallas, TX.