Chapter I. Introduction

I.1. Introduction

WOCE's principle goal, as set forth in the WOCE Scientific Plan (WCRP 6, 1986) is to improve ocean modelling for the ultimate purpose of climate prediction. This goal is supported by a number of objectives which involve description of specific aspects of the ocean circulation and dynamical understanding of ocean processes; such understanding and description are required for progress in validation and improvement of models. The WOCE Implementation Plan (WCRP Reports 11, 12, 1988) described how the scientific objectives are to be achieved, with careful attention to the field work necessary for the global description, and a framework for the process experiments which have been carried out.

The WOCE field observation phase is drawing to a close, ocean modelling has moved to a much higher level of simulation than was possible at the beginning of WOCE, and data assimilation methods are being developed by many groups. WOCE has been extended to the year 2002 to take advantage of the full data set and advances in modelling, for much fuller development of the models and particularly methods for data assimilation. This final term of WOCE is called the AIMS (Analysis, Interpretation, Modelling and Synthesis) phase. The plan laid out herein describes the goals and recommended structures for AIMS.

A major scientific thrust for the remaining years of WOCE is to assess which ocean processes are really important for climate prediction. To have any predictive value, models must accurately describe the present ocean state. In particular, which processes below the surface layer must be observed and modeled in order to increase skill? Is there a minimum resolution in space/time for observations and models which increases predictive skill?

By the end of WOCE we must have estimated heat and freshwater transports and divergences with reasonable error levels. We should be able to say something useful about the time scale and magnitude of variation. We should attempt to characterize (if not parameterize) the deficiencies in model dynamics at various resolutions. We should quantify the principle mechanisms for transporting heat, freshwater and mass - these results should be central to arguments about what mechanisms should be present in a viable ocean model. That is, what are the relative roles of the Ekman layer, subduction, convection, mixing, and the deep overturning circulation? We should improve estimates of diapycnal fluxes and upwelling. Improving such estimates will require process studies designed to reveal the roles of smaller scale phenomena in the general circulation.

In addition to improvements for climate prediction, the WOCE data set and modelling will result in improved regional descriptions of circulation, in all ocean basins as well as in the specific Core 2 (Southern Ocean) and Core 3 (North Atlantic and Brazil Basin) regions.

Synthesis of WOCE data and improved modelling calls for equally prioritized progress in three overlapping areas: data analysis, prognostic modelling and data assimilation. The WOCE goals are written in terms of improvement of ocean modelling for climate prediction. Since much of our understanding of processes in the ocean comes from observations, models cannot be improved without thorough consideration of data. Adequate resources for ocean model development are needed to carry the current piecemeal progress forward. Ocean data assimilation, even for the goal of estimating the state of the ocean in a dynamically-consistent manner as opposed to its use in climate prediction, is still in the developmental stage. The data assimilation techniques being used in oceanographic data assimilation are similar in complexity to state of the art assimilation methods used in numerical weather prediction, although the different nature of the two problems requires also different assimilation approaches. Also, advanced assimilation techniques are rapidly entering the arena of ENSO prediction. Nevertheless, ocean models, even high resolution ones that are initialized by data assimilation, still leave room for improvements, which is where the WOCE AIMS phase is expected to make a significant contribution.

The AIMS plan is organized as follows.

* Chapter 2: the WOCE scientific goals are described with brief accounts of current progress towards achieving them and a description of how WOCE can facilitate continued progress, through combination of the three approaches: data analysis, modelling and data assimilation. These methods should not be viewed as separate, but should be used in whatever combination is most fruitful for achieving the WOCE goals.

* Chapter 3: WOCE data analysis is resulting in descriptions of phenomena and hypotheses regarding ocean dynamics. Analysis based principally on the data sets themselves must continue, with increasing steps towards combined use of data and models. Complete combination of data with models is considered under data assimilation. The goals of the data analysis are described. Priorities could be assigned if necessary, but methods should not be prescribed. The WOCE data types are reviewed, including the organizational structures which are necessary for assembling, archiving, and distributing the data, in light of the expected WOCE users for each type of data. Although WOCE field observations technically will cease at the end of 1997, some of the data streams will continue as part of other climate programs, and so some portions of the organizational structures which assist in acquiring the data should be continued, although perhaps with funding separate from WOCE.

* Chapter 4: The current state of ocean modelling relevant for WOCE is reviewed and a strategy for the next 5 years is proposed. WOCE modelling is focused on the ocean, but central to its purpose is evaluation of ocean models of varying formulation and resolution for use in couple climate models. The section includes recommendations for model developments and validation and an assessment of the resources (manpower, computing, centers, workshops, publications) which are needed or would be useful.

* Chapter 5: WOCE data assimilation is described, including an assessment of the present state, goals and resource assessments for the next five years. It is emphasized that the combination of models and data for WOCE is state estimation rather than predictive assimilation. While it was expected in the original WOCE science plan that data assimilation would eventually be central to WOCE, developments at that time were minimal, and so a full discussion of assimilation issues was not possible. Major developments have taken place in ocean data assimilation in the past few years, and it is expected that development will accelerate. Three areas of development for WOCE are prescribed, ranked in order of immediate to long-term achievability: inverse methods with simplified dynamics, crude assimilation into high resolution models, statistically-accurate assimilation into coarse resolution models. and accurate assimilation into high resolution models. The ultimate goal of accurate assimilation into high resolution models will only be achieved with these developments and a large increase in computer power.

I.2.1. Ocean Model/Data Intercomparison Project

I.2.4.a. Regional workshops

I.2.4.b. Global workshop (2000 or 2001)

A modelling workshop that will address the modelling issues discussed in this report, as well as future directions in ocean modelling, will be held in 1997 or 1998 (see section IV.4.1 below). The workshop will be co-sponsored by the WOCE SMWG and the CLIVAR NEG-2 committees. The workshop will be used to report the current status of WOCE modelling efforts. It will also discuss future directions for ocean modelling. These might include technical issues such as novel numerical methods that can and should be incorporated into ocean models, the preparation of the next generation ocean models to be used within coupled models, development of state-of-the-art one-minute resolution ocean models, novel ocean-atmosphere-ice coupling schemes, etc.