PT5 methods

In order to analyze the available information, we will use a variety of statistical methods. The information can generally be divided into three categories, (1) horizontal fields, (2) crossections, and (3) vertical transects. Moreover, all information is available as functions of time, albeit for different durations.

The horizontal fields are of particular interest when it comes to detection of signal propagation. We intend to use various fairly simple statistical methods to this end. First, we need to determine the paths of anomalies. We will apply complex empirical orthogonal functions (complex EOFs), which have the advantage of capability of tracking propagating spatial patterns (compared to ordinary EOFs). If deemed necessary, these analyses will be supplemented by multi-channel singular spectrum analysis. We will also consider the pathways of the main ocean currents, and reported anomaly pathways. Another alternative is to objectively determine these pathways from boxed budgets by looking for anomalous events at the various box edges. Results will be supplemented by visual detection from computer animations.

Once their pathways have been described, anomalies will be tracked and described using techniques such as Hovmøller diagrams and computation of lagged correlations. We will study results from crossections, including extracted pathways, using EOF analysis, and possibly other related techniques such as analysis by Principal Oscillation Patterns.

Information from the fields will be extracted along the crossections where data are available. In addition to the obvious purpose of a general intercomparison, these complementary sets will be used in order to evaluate the applicability of the data that are only available as crossections with respect to investigations of seasonal and interannual variability.

Horizontal fields in the Arctic ocean will also be investigated for trends. For this purpose, time series analysis of results from spatial integration of the relevant properties will be performed. These results will predominantly be obtained for the subdomain(s) of the fields.

To a certain extent, analyses of the ocean and ice circulation features which have been described this far will be supplemented by analysis of processes occurring at the ocean-(ice-)atmosphere interface. Here, we will describe how the fluxes of momentum, heat and/or fresh water across this interface contribute to variability on the timescales in question. These analyses are best performed for horizontal fields, whereas for, e.g., crossections, such analyses are contaminated by advective processes.

All methods have been selected because they are well-suited for the task work that is described above. Moreover, all analyses are defined based on this task's objectives and the relevant overall objectives of the coordinated project.


Arne Melsom
Last modified: Fri Mar 30 13:16:07 GMT 2001