Review of the PIRATA report

Accomplishments of PIRATA: 1997-2005, Status and Perspectives

By a subgroup of the Ocean Observations Panel for Climate (OOPC): D. E. Harrison, F. Schott, R. Weller

October 2006

Summary: The PIRATA array and related observations taken during mooring exchange/refit and ancillary cruises have allowed a multitude of studies leading to a much better understanding of Tropical Atlantic Variability (TAV), and have made possible the initiation of ocean forecasting for the region. Prior to the multi-national establishment of the PIRATA network and related regular cruises, observations of the Tropical Atlantic were much sparser than for the Tropical Pacific. OOPC fully endorses the concept of sustained observations of TAV, including continuation of the network and of the southwestern and planned eastern extensions. However sustained observing of the region should be undertaken with an integrated observing system in which surface moorings are used where they are the most effective technology, and other technologies are used where they offer advantages. OOPC thus recommends that annual meetings of the TAV community be established which include PIRATA, and that development of a plan for of an integrated observing system for the tropical Atlantic be a priority for the TAV community.

1. PIRATA status 2006

OOPC compliments PIRATA for its accomplishments under challenging conditions. The existence of PIRATA and of its associated exchange cruises has resulted in a seasonal and interannual coverage of the tropical Atlantic that was impossible before, except for brief periods during international experiments like the GARP programs GATE and FGGE. An important result of PIRATA is that it has spawned significant enthusiasm and progress among adjacent nations, and in particular Brazil. PIRATA observations are important for operational oceanography and climate forecasting in the tropical Atlantic. Much has been learned in the analysis of PIRATA data. Doubts on the mean zonal circulation on the equator were resolved using PIRATA moored data, and revealed the existence of two westward intermediate currents (EICs) underneath the Equatorial Undercurrent (EUC), separated by an eastward flow in between. As was accomplished in the Pacific earlier using TAO exchange cruises (e.g. Johnson et al., 2002; Sloyan et al., 2003), PIRATA ship sections have recently allowed us to better quantify the transports of the eastward equatorial and off-equatorial undercurrents (Schott et al., 2003; Brandt et al., 2006; Fischer et al., 2006). Furthermore, the existence of PIRATA has allowed process studies (for example, a study on mixing in the eastern cold-tongue lead by M. Dengler of IFM-GEOMAR) that are essential in improving models, and has certainly helped triggering planning activities for internationally-coordinated equatorial (TACE) or near-equatorial (AMI, AMMA) studies. A multi-year study of the eastern equatorial regime has also been proposed to the German Deutsche Forschungsgemeinschaft (DFG), which also benefits from the momentum and perseverance of PIRATA.

2. Recommendations

Importance of moored or profiling subsurface observations at intraseasonal timescales

Recent observations in the central tropical Atlantic (Brandt et al., 2006) have emphasized the need for subsurface coverage of intraseasonal variability at time scales shorter than a month. Their spatial structures and dynamics are not yet sufficiently well understood, and their surface signatures are not satisfactorily resolved by satellite observations. This requires subsurface moored ADCP or rotor current measurements with a best data quality standard. Mooring survivability would be greatly increased if subsurface floatation is used, for example with a top float near the 100m level carrying an upward (high-resolution) and a downward (longranger) ADCP. Whether this task can also be carried out by gliders in the future needs to be investigated.

The need for eastern tropical Atlantic and cold tongue studies

Coupled models have significant difficulty in reproducing tropical Atlantic stratification and SST (e.g., IPCC-AR3 and CLIVAR/TACE reports), with the cold spot essentially appearing in the central basin and too-warm eastern tropics. Improved parameterization of the shallow eastern mixed layers is therefore important. Since advection is important in the tropical belt and eastern upwelling regimes (Foltz et al., 2003), moored current observations are essential in the region, including documentation of the poleward propagation of coastal-trapped waves (CTWs). The proposed southeastern extension (SEE) focuses on the Benguela Niños with moorings at 6-8° S; and the proposed northeastern extension (NEE) on the ITCZ migration, i.e. on issues of interest for AMMA/AMI, with planned moorings along 23° W in a region important for climate anomalies and hurricane generation. Therefore the OOPC strongly endorses both extensions and the associated research foci.

Southwest Extension

After international review, Brazil has started a program with three moored stations off its northeastern coast, between 8-19° S . In this region, surface buoys are much less prone to vandalism and meteorological observations are useful for forecasts in Brazil. Recent observations have shown that there is large intraseasonal variability in the region, and from altimetry and high-resolution model analyses (Ph.D. study K. v. Schuckmann, IFM-GEOMAR), a relation to the interior basin has been suggested. Therefore, moored observations in the zonal inflow regime will be helpful for understanding the ocean response and boundary interaction.

The PIRATA mooring concept

At the start of PIRATA, there was a general interest in and preoccupation with the “SST Dipole Mode”, leading to the two NW and SE off-equatorial mooring lines that run through the presumed dipole maxima. Good coverage of the equatorial belt and boundary interaction regions (PIRATA extensions) by moored stations is of continued highest importance, as observations and model studies have shown a large contribution of advection and diffusion to the heat and salt balance on the equator, and for the propagating boundary signals. At the NW and SE off-equatorial stations, advection plays a less important role. Given the availability of other types of observations of stratification and water mass variability (Argo float profiles in particular) a reassessment of how much of an off-equatorial moored station coverage will be needed in the future should take place. Furthermore there were high hopes in obtaining much more realistic model simulations through data assimilation. However, it is fair to say that at the present stage of evolution of ocean data assimilation systems, it is unclear how well sparse data points from a dozen or so widely distributed moored stations actually influence model simulations.

Use of surface moorings

The biggest problem PIRATA has faced is vandalism of surface moorings, which has resulted in significant losses, data gaps and data quality deterioration. There should be a discussion of why the use of surface moorings is essential at each of the present locations, in particular in the east, where vandalism is higher, given the technical advances in autonomous observations since the beginnings of PIRATA. Many of the use examples cited in the review document could be addressed by other sampling means: for example (Figures 7&9) – satellite SST can also produce the dipole index. More generally, the possibility of gliders deployed as virtual moorings should be explored, as they may be able to produce the same benefit with a far reduced risk of vandalism. Certainly the surface met and flux data have high value. However, some questions should be addressed by the TAV community: Does daily-averaged data of coarse vertical resolution from the ocean justify the quality challenges posed by vandalism to surface moorings? What is the density of moorings needed to get the surface meteorology and fluxes? Are these independent observations or are they too coarse to have any coherence?

Development of an Integrated Observing System for TAV

It is suggested that the TAV community work with the PIRATA community and the tropical Atlantic operational oceanography community to design an observing system that makes best use of all available technologies. It is suggested that OSSEs be carried out in which the importance of individual stations and sampling strategies for recovering the energetically important parts of the variability are investigated, by subsampling and reconstructing output fields of high-resolution models, such as those available from the 1/12 degree FLAME or HYCOM models. This strategy was successfully applied for investigating sampling strategies in the tropical-subtropical Indian Ocean with its strong intraseasonal variability (Vecchi et al., 2006). It is also recommended that model simulations with ocean data assimilation be tested by using different subsets of PIRATA location data.

3. Vessel support

Ship time is one of the limiting factors for successfully continuing PIRATA into the future. Through OSSEs and interaction with the meteorological/assimilation community it should be determined which surface stations with met observations are considered essential. Subsurface moorings could be deployed for one year at a time with little risk of vandalism, but there would be technical challenges in obtaining the data in real time. These systems should be tested before any transition. The impact of delayed mode access to some observations should also be weighed against the probability of ultimately more complete data sets made possible by less vandalism-prone observational techniques.

4. Technical mooring issues

This section has a number of specific questions from OOPC based on the PIRATA review document.

1) There needs to be more information provided about calibration, quality control, quality assurance, and accuracies of the sensors, instrument systems, and resulting data. There is no discussion of calibration procedures (factory or lab calibrations how often?), of how sensors degrade with time in the field (certainly the salinity and meteorological sensors do or can), of problems with ADCP data due to fish (in the TAO array they have gone to separate subsurface moorings to get ADCP data as the fish return could not be beat so how does PIRATA do QC on the ADCP data?). On page 28 the report mentions the “bad quality of the buoy data” – what does this mean, certainly the buoys cannot measure SST and make a temperature measurement at some depth below the hull which is never mentioned – so how much bias arises due to this? Page 45 talks about dust deposits-how do these affect the met sensors, especially the radiometers? Several discussions about the use of the buoy data talk about using the buoy data as a truth to identify biases and then points to differences; yet, none of this discussion has much impact as one does not know how accurate the buoy data are and thus cannot judge if any of the differences are valid and/or significant. For an operational continuation of PIRATA in the future, there must be

a thorough and careful statement of practices and procedures for QC/QA and calibration,

justification based on field comparison with other data of the in situ accuracies,

a discussion of the observational challenges (bioufouling of salinity, degradation of met sensors, fish swimming in the ADCP sampling field).

There is a reference in the text to a TAO website, but it does not carry sufficient information. It is the procedures and practices used in PIRATA that must be documented, and the accuracies and uncertainties are too important not to be discussed in some depth in this report. Page 26 uses the phrase “high quality buoy observations” but there is no substantiation of what this means. On page 40 there is a discussion of differences between the buoys and models of 0.4 °C in SST, 0.6 °C for air temp, 1.3 m/s for wind speed—are these significant? Which is correct? Certainly in low winds buoy air and sea temperatures can have at least this much error.

2) Following the above, what is the procedure for updating the real-time data with corrections based on post-deployment calibrations? How are users alerted to these corrections? A table of sampling rates is needed.

3) The document needs more detailed description of the sampling and cruises. How often are the cruises (“…at least once a year …”, page 7)? What observations are made on the land-based meteorological stations (page 15)? A further explanation of the statement on page 51 is needed, that work is being done to get higher sampling rate data sent via satellite with the result that PIRATA data sets would be “more useful and even indispensable for more skillful predictions.”

4) The utility of the telemetered data to meet some of the goals is unclear. The telemetered data is daily-averaged (top of page 15), so how does this get used to check or initialize models? The vertical spacing of the temperature sensors is coarse; is the resulting accuracy in the mixed layer depth determination good enough? What are the error bars on Figure 25 – the mixed layer salinity balance? How was the vertical sampling strategy chosen? – Argo floats give much more vertical resolution – to get a mixed layer salinity balance, would not that be a better approach?