COAPEC Thematic Programme Proposal February 2000

 

How well do Coupled Models portray the Observed Thermohaline Circulation?

 

Dr Karen J. Heywood, Dr David P. Stevens and Dr Elaine L. McDonagh

University of East Anglia, Norwich

 

Introduction

 

The northward heat transport in the Atlantic Ocean has a well known impact on European climate. Much of the heat transport is associated with the thermohaline circulation (THC). The THC is often presented in modelling studies in terms of the overturning stream function. Determinations of this overturning in the ocean, both its mean state and variability, are sparse to date. We shall analyse the THC in the coupled ocean-atmosphere model HadCM3. We shall use observed hydrographic section data and inverse methods to determine where, when and why models and the real world diverge or converge.

 

A large part of the climate variability of Europe is associated with the North Atlantic Oscillation (NAO) pattern, measured through an atmospheric pressure difference between Iceland and the Azores. The NAO has shown major low frequency variations during the last few decades. Some of the low frequency variability of the North Atlantic Ocean is believed to be caused by fluctuations in the NAO. The NAO influences deep water formation in the Nordic Seas and thus it is suggested that it will affect the THC (Dickson et al., 1997). Whether the NAO is a coupled phenomenon (or simply atmosphere driving ocean variability) is a topic of great debate, and one which we may be able to shed some light upon through our combined use of both coupled climate model output and independent ocean observations. The THC is of global importance for climate stability and/or sudden switches in state. Its variability may be caused by formation and spreading of deep waters. The THC is known to be variable on palaeoclimatic time scales; this project will address the observed variability of the THC in hydrographic sections in the North Atlantic during the last 3 decades.

 

The World Climate Research Programme’s project CLIVAR (Climate Variability and Predictability) emphasises the importance of this topic, as does the Euroclivar group, set up by the EU Fourth Framework to establish European plans. CLIVAR has identified Principal Research Areas in the topics of the North Atlantic Oscillation, and Variability of the Atlantic Thermohaline Circulation. The THC is of particular relevance to European climate because it is suggested that variations in the amounts of heat released to the atmosphere during deep water formation in the Northeast Atlantic and Nordic Seas may cause fluctuations in the weather patterns (storms, precipitation) in western Europe. Globally, there is concern that our climate may be sensitive to small perturbations; further understanding of these processes is necessary to predict them.

 

Until recently coupled ocean-atmosphere climate models have usually employed an air-sea flux correction to prevent unacceptable drifts in properties such as sea surface temperature during long model runs. The use of such flux corrections meant that the output had to be considered with caution, particularly in terms of assessing changes in the thermohaline circulation and deep water formation. A new generation of coupled models, with better resolution and better parameterizations, have now been developed which do not require flux correction. In this project we shall analyse output from one of these new models, HadCM3, developed by the Hadley Centre for Climate Prediction and Research. Details of the model formulation and a discussion of ocean heat transport in HadCM3 are given by Gordon et al. (2000). HadCM3 has a resolution of 1.25° x 1.25° in its model ocean with 20 vertical levels, and includes a simple sea ice model but no sea ice dynamics. It is a great improvement on HadCM2, particularly for ocean heat transport (Wood et al., 1999). We shall analyse the control run of this model (other runs with increased concentrations of carbon dioxide are being studied in-house at the Hadley Centre and are not available to COAPEC researchers). The control run is 1000 years in length. Preliminary analysis of the HadCM3 control run shows that it can exhibit deep convection in the Labrador and Greenland Seas.

 

The Thematic Programme COAPEC (Coupled Ocean-Atmosphere Processes and European Climate) is one of the UK contributions to CLIVAR. The major goal of COAPEC is to determine the impact on climate of the coupling between the atmosphere and the Atlantic Ocean on seasonal to decadal timescales. COAPEC offers the opportunity to analyse in detail the output from a coupled climate model (HadCM3). Although the Hadley Centre are undertaking their own investigation into this rich data set, they have particularly encouraged the involvement of observational oceanographers who can bring experience of analysing in situ hydrographic sections. This proposal has been developed through joint discussion with Richard Wood’s group at the Hadley Centre, and we expect close and productive collaboration between UEA and our named collaborators.

 

The bringing together of coupled model output and hydrographic section data is particularly timely in view of the availability of the recently completed WOCE (World Ocean Circulation Experiment) hydrographic survey. These WOCE sections, in which the PIs and their groups were involved, were of unprecedented accuracy and station density. The Euroclivar report identifies analysis of the variability patterns in repeated WOCE sections as a high priority, to provide an index of the THC. It suggests 48°N as a priority candidate, together with 24°N. The report encourages the comparison of such indices with model simulations. Bryden and Imawaki (2000) explain that the uncertainties in the ocean surface flux climatologies calculated from meteorological observations and the bulk formulae (e.g. Josey et al., 1998; 1999) are still so large that the heat transported by the ocean cannot be reliably deduced from these alone. They review the existing direct heat flux calculations from recent WOCE sections, and stress the importance of the THC for providing much of the northward heat transport in the North Atlantic. They emphasise the effort that must be expended to decide on the particular constraints in calculating ocean heat transport. It is not adequate simply to multiply velocities by temperatures without carefully determining constraints such as western boundary current transports (in effect, these constraints minimise the errors involved in choosing reference level velocities for geostrophic calculations).

 

Freshwater fluxes facilitated by the ocean are a thornier issue than heat fluxes. Wunsch (1996) points out that the ocean carries such enormous quantities of freshwater that the net salt flux across a section is the tiny difference between two large numbers. Wijffels et al. (1992) calculated global climatological freshwater fluxes and their divergences from hydrographic sections. They discuss the uncertainties in these estimates and admit that they are tentative. Indeed Saunders and King (1995), from their calculation of freshwater flux across 45°S in the South Atlantic, suggest that the Wijfells et al. global freshwater scheme is incorrect. Helene Banks at the Hadley Centre (one of our named Collaborators on this proposal) has been analysing the freshwater budget in HadCM3, and finds that it takes about 400 years for the model to come into equilibrium, much slower than the heat budget which has more efficient feedback mechanisms.

 

Banks (2000) compares ocean heat transport in the South Atlantic from HadCM3 and some hydrographic sections, focusing particularly on 30°S and the WOCE A11 section at 45°S (Saunders and King, 1995). Globally, the heat transport in the model is similar to that deduced from ERBE (Earth Radiation Budget Experiment) and agrees reasonably well with observations (Gordon et al., 2000). Banks (2000) discusses the differences between the model and observed temperature, salinity and heat transport values on A11. One of the primary differences is the small volume of northward flowing Antarctic Bottom Water (AABW) in HadCM3 at 45°S. She finds that an increase in the northward flow of AABW leads to an increase in the outflow of North Atlantic Deep Water (NADW). The two overturning cells appear to be decoupled. She concludes that a detailed investigation of the NADW cell and formation of NADW in HadCM3 is necessary. The strength of the overturning associated with the North Atlantic THC at 24°N shows variability on a range of timescales (Figure 1). During this project we will be able to determine the range of variability on short (annual) time scales from the WOCE sections and on longer time scales from the 24°N and other sections.

 

 

Figure 1. The strength of the overturning associated with the North Atlantic thermohaline circulation at 24°N for a 200 year period of the HadCM3 control run. Thin line indicates annual mean overturning; bold line indicates 10 year running mean.

 

 

Wood et al. (1999) analyse the northward heat transport and overturning in HadCM3 associated with the THC in the North Atlantic. The maximum overturning in the North Atlantic has a mean value of 21.6 Sv during the control run after spin-up. Wood et al. point out that while such maximum overturning transports are used frequently by modellers, they are not easy to calculate from observational data. They therefore compare transports of NADW with observations at a few locations. They emphasise the need for well-understood models if we are to make decisions based on climate change scenarios.

 

Specific Objectives

 

These objectives address two of the main elements of the COAPEC work plan, namely, “What are the observed characteristics of seasonal-to-decadal climate variability in the Atlantic Sector?” and “How do the mean climate and climate variability in the Atlantic Sector simulated by a Coupled General Circulation Model differ from that observed?”.

 

Methodology and Approach

 

The available high-quality quasi-zonal hydrographic sections in the North Atlantic will be combined with inverse methods to derive an absolute velocity field and allow the partition of the flow into Ekman, gyre and overturning circulations. The near zero net volume flux across zonal sections in the Atlantic allows these components to also be expressed in terms of the heat flux that they facilitate. The box inverse model uses the CSIRO DOBOX software (Morgan, 1994) which has been used by Dr Elaine McDonagh at UEA for the last two years to determine the circulation in the South Atlantic from hydrographic observations. This software takes initial geostrophic velocity fields, includes Ekman fluxes and surface air-sea fluxes, and adjusts reference level velocities to satisfy constraints imposed on the system. Basic constraints conserve mass, temperature and salinity (and potentially other tracers such as silicate) in layers (defined by the user). Other constraints can be added, and all constraints can be weighted to reflect the confidence one has in them. As the inversion minimises the adjustment to the initial velocity field to satisfy the imposed constraints, it is very important that the initial velocity field is realistic. Where direct velocity measurements are available, such as from floats or current meters, these initial velocity fields can be improved. The model allows for and solves for advection and diffusion between layers within the box, as well as a freshwater and heat flux at the air-sea interface. Formal error bars will be put on all of the output from the inversion. A study of the sensitivity of the solution to the initial conditions of the system will be undertaken. Such a model can also be used for a single section where an integral constraint can be applied, i.e. zero net volume flux across the section (c.f. Saunders and King, 1995). Ekman velocities will be derived from the ECMWF reanalysis (Era-40) which should be available when this funding commences. Era-40 covers the period from 1958 to present. The performance of box inverse models was tested using output from the Fine Resolution Antarctic Model (FRAM) (McIntosh and Rintoul, 1997). They concluded that such a model can be used to determine fluxes across the interfaces between layers and horizontal fluxes for large ocean regions. They recommend the use of such techniques for analysis of the WOCE data set. They note that “probably the largest source of error in box inverse models is … that ocean sections are not synoptic and the ocean circulation is not steady. … Good estimates of the mean ocean circulation will require averaging a number of synoptic sections”. Our study of both the coupled climate model and the time series of hydrographic sections will determine the timescale over which this is a valid approach and what combination of sections can sensibly be used as ‘synoptic’ when considering the overturning circulation.

 

We plan to complete separate inversions for each section of hydrographic data. Only sections which extend from land to land will be used because of the additional constraints that this adds to the system (Figure 2). There are approximately 30 such sections occupied since the late 1960’s when high quality measurements became possible. This provides us with a 30 year record, although there are many more sections towards the end of the period due to the WOCE programme. For example, there are at least 10 suitable repeats of the WOCE section AR7 between the UK and Greenland, of which 7 are already publicly available (Figure 2). Data from several of the more recent sections are not yet public, but either their release will occur before this funding starts or we are in discussions with the individual PIs to release the data for our analysis. Earlier inversions used pre-WOCE climatologies and did not consider temporal variability (Macdonald and Wunsch, 1996).

 

We will examine the output of these inversions to determine the structure and variability of the overturning circulation. These results will be compared with the characteristics of the thermohaline circulation in the Hadley Centre coupled model (HadCM3) control run. By studying both the coupled climate model and the time series of hydrographic sections, we shall be able to provide estimates of both the mean overturning circulation and how it varies temporally. During the period that the hydrographic sections (Figure 2) were taken, the NAO has moved from low values to an all time high of the direct observational record. We will look for and examine time frames in HadCM3 with similar characteristics. Clare Cooper at the Hadley Centre has already undertaken a study of the NAO in HadCM3 and we shall utilize her results. As well as comparing the HadCM3 diagnostics with the results from the inversions, the HadCM3 results will also be analysed to work out a decorrelation timescale for the variability of the overturning. This time scale would then be used to group hydrographic sections together to increase the constraints on the system and ascertain the effect on the solution. A time series for the variability of the THC at a selected number of locations, probably 24°N and 60°N, will be synthesised by mapping spatially close inversions onto the THC produced by the detailed high resolution model OCCAM. Links between the characteristics of the thermohaline circulation and other components of the climate system, in particular the NAO, will be explored using our inversions, climate data and the coupled model simulations.

Figure 2. Station positions for high quality hydrographic sections across the North Atlantic which will be utilized for this study. Sections are grouped by decade and were extracted from the WOCE, NEMO and WOA98 data bases. Numbers indicate the number of repeat sections where station positions were available in February 2000. The 2000 m and 4000 m depth contours are also shown.

 

In addition to heat fluxes, we shall deduce freshwater fluxes and their divergences in both the observed data set (for the boxes defined by pairs of sections) and in the model output. We recognise the need to be cautious in interpreting ocean freshwater fluxes deduced from inverting hydrographic sections, and acknowledge that we may even be unable to determine the sign of this freshwater transport with any confidence. However it is encouraging that Wunsch (1996) concludes that “in the long term, the direct oceanic calculations, whatever their present difficulties, appear to be the most promising method by which large-area estimates of atmosphere/ocean water transfers will be made”.

 

Output from OCCAM will also be studied, particularly to determine the detailed structure of the THC and whether it represents the observed fields and their heat and freshwater fluxes (if not, why not?). OCCAM is a global fine resolution (0.25° horizontally) model with a similar formulation to the ocean component of HadCM3. OCCAM is not coupled to an atmosphere, but it has been forced with realistic high frequency winds from re-analyses. It has not been run for a very long time and thus has not drifted far from the initial climatology (WOA94). Thus one might expect that the spatial structure of the THC closely resembles the mean observed state. This will be verified with our observational data set, and if it is the case, the model will be used to link observations at different locations.

 

Simon Josey and coworkers at the Southampton Oceanography Centre are submitting a complementary proposal to the COAPEC Thematic Programme. They plan to invert their long term climatology of air-sea fluxes, to determine which of the oceanic measurements of horizontal heat and freshwater fluxes make a coherent set. Whilst they are looking at the long term mean, we shall look at both the long term mean and the variability, in both HadCM3 and hydrographic sections. We shall use the existing SOC flux climatology as the initial conditions for our inverse model. However because this climatology is not globally consistent, we may consider use of numerical model fluxes if our sensitivity tests suggest that this may be critical.

 

Plan of Research

 

Year 1

·        Set up initial velocity fields for inversions

Year 2

·        Set up constraints on the system based on a priori knowledge of the circulation

·        Single section inversions to deduce values for overturning, volume, heat and freshwater fluxes

·        Analyse THC in HadCM3 and OCCAM by determining time series of overturning, volume, heat and freshwater fluxes

·        Present work at EGS and AGU Ocean Sciences

Year 3

·        Determine initial advection and diffusion between boxes, evaporation-precipitation and surface heat flux for boxes defined by multiple section inversions

·        Multiple section inversions from hydrographic data

·        Compare inversion and HadCM3 results with components of the climate system

·        Present work at EGS and IUGG, and prepare papers for publication

 

Project Management

 

Drs Heywood and Stevens already run a successful joint research project to undertake and analyse data from a hydrographic cruise to the Scotia Sea in 1999 (the ALBATROSS project). As with that project, this project will be managed by both PIs working with the PDRA. Weekly meetings will ensure good communication and exchange of ideas. The physical oceanography group at UEA is large and active, and provides a supportive and stimulating environment for research. Research seminars provide opportunities to hear invited speakers and to discuss work underway at UEA. The School encourages presentation of work at international conferences. Dr Heywood’s group comprises 5 PhD students and 2 postdocs. Dr Stevens’ group comprises 2 postdocs and 1 PhD student.

 

The career prospects of the PDRA will be enhanced by the experience of analysing the output of the coupled atmosphere-ocean model HadCM3. This will make them a valuable member of the international climate research community. UEA provides numerous training opportunities for its researchers and the PDRA will be encouraged to consider these. Courses available include various computing techniques as well as general transferable skills and careers advice. The PDRA’s career will be enhanced through giving talks or seminars at UEA and at other COAPEC institutions, and through learning about coupled models through close collaboration with the Hadley Centre researchers. Research staff at UEA are encouraged to give one or two lectures to undergraduates each year, to gain teaching experience, and also participate on the supervisory committees of PhD students. The PDRA will also be encouraged to participate in hydrographic cruises with which UEA is involved or coordinating.

 

Data Stewardship

 

A vast quantity of data will be examined during the research programme; however, the resulting products will be somewhat smaller in scale. We envisage providing the observed vertical structure of the zonally averaged volume flux (as a function of pressure and neutral density), the heat and freshwater transports and error bars for each section. Optimised velocity fields will also be made available. These data will be provided on our web site after publication. These data sets describing the structure of the THC and its variability including error bars will be very valuable for the international climate modelling community for validation of coupled models in the future, after the completion of the COAPEC programme. On completion of the project, our results will be made available to the climate research community via the British Oceanographic Data Centre (BODC) or British Atmospheric Data Centre (BADC) as appropriate. During the project, we will gladly exchange data with other scientists while there is no conflict of interest.

 

Public Understanding of Science

 

The proposed research should lead to several scientific articles which we shall present at national and international scientific conferences and publish in the major journals. In addition, we propose several specific actions which will help us share our findings with the wider public. First, we will post examples of our findings on the UEA web site. As the wider public goes online, the Internet has increasingly become a forum for educating curious nonscientists about current research. Second, we expect to prepare aspects of our findings as examples and problems for undergraduate and MSc students at UEA. Finally, when we find suitably interesting results, we will willingly discuss them with the press.

 

References

 

Banks, H. T. (2000) Ocean heat transport in the South Atlantic in a coupled climate model, J. Geophys. Res., 105, 1071-1091.

Bryden H. L. and S. Imawaki (2000) Ocean heat Transport, in Ocean Circulation and Climate, ed. G. Siedler and J. Church, in review.

Dickson, R. R., J. Lazier, J. Meincke, P. Rhines and J. Swift (1997) Long-term co-ordinated changes in the convective activity of the North Atlantic, Prog. Oceanog., 38, 241-295.

Gordon, C., C. Cooper, C. A. Senior, H. Banks, J. M. Gregory, T. C. Johns, J. F. B. Mitchell and R. A. Wood (2000) The simulation of SST, sea ice and ocean heat transports in a version of the Hadley Centre coupled model without flux adjustments, Clim. Dyn., in press.

Josey, S. A., E. C. Kent and P. K. Taylor (1998) The Southampton Oceanography Centre (SOC) Ocean - Atmosphere Heat, Momentum and Freshwater Flux Atlas. Southampton Oceanography Centre Report No. 6, 30 pp. & figs.

Josey, S. A., E. C. Kent and P. K. Taylor (1999) New insights into the ocean heat budget closure problem from analysis of the SOC air-sea flux climatology, Journal of Climate, 12(9), 2856 - 2880.

Macdonald, A. M. and C. Wunsch (1996) An estimate of global ocean circulation and heat fluxes, Nature, 382, 436-439.

McIntosh, P. C. and S. R. Rintoul (1997) Do Box Inverse Models Work?, J. Phys. Oceanogr., 27, 291-308

Morgan, P.P. (1994) Box Inverse Modelling with DOBOX 4.2. CSIRO Marine Laboratories Rep. 225. 26pp.

Saunders, P. M. and B. A. King (1995) Oceanic fluxes on the WOCE A11 section, J. Phys. Oceanogr., 25, 1942-1958.

Wijffels, S. E., R. W. Schmidt, H. L. Bryden and A. Stigebrandt (1992) Transport of freshwater by the oceans, J. Phys. Oceanogr., 22, 155-162.

Wood, R. A., A. B. Keen, J. F. B. Mitchell and J. M. Gregory (1999) Changing spatial structure of the thermohaline circulation in response to atmospheric CO2 forcing in a climate model, Nature, 399, 572-575.

Wunsch, C. (1996) The Ocean Circulation Inverse Problem, Cambridge University Press, 442 pp