Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
3 The Principal Observables ORDER OF IMPORTANCE To gain some insight beyond internal panel expertise as to which obser- vations make the greatest impact on current and anticipated prediction schemes, we posed the following question (number 2 of 6) to the Tropical Ocean and Global Atmosphere (TOGA) program on Seasonal to Interannual Prediction (T-POP) group (R. Knox letter to E. Sarachik, October 22, 1991): "Of the existing observations, can the T-POP say what elements are most useful, for what purposes, and why? Perhaps this question is premature, but even partial answers would help us to understand which kinds and distributions of observations are most critical to prediction efforts. We include measurements for both initialization and validation of models un- der 'prediction efforts.' What are the most significant observational gaps you see, and what, if any, are the redundancies? Is it possible to go a step further and quantify and prioritize the additional observations needed, or the redundant observations to be dropped?" The relevant portions of the response (Sarachik to Knox, January 9, 1992) follow. Essentially, the same response has appeared in report form (TOGA Numerical Experimentation Group, 1992). "The T-POP group has met (jointly with the international TOGA Nu- merical Experimentation Group) in San Francisco during the week of De- cember 9, 1991 and discussed your letter of October 22, 1991. This letter is a summary of that discussion in the form of responses to the six ques- 19
20 OCEAN-ATMOSPHERE OBSERVATIONS tions you posed in your letter. We appreciate your and your panel's effort at writing the letter-we found the discussion extremely useful. "(Question 2) We all agreed that the most important observations for the improvement of prediction skill by coupled atmosphere-ocean models are the fields of surface wind stress and sea surface temperature (SST). "We agreed that fields of subsurface thermal data are the next most useful, and this data would be used in conjunction with the wind stress and SST for initializing coupled atmosphere-ocean models. "We noted that the TOGA Tropical Atmosphere Ocean (TAO) array currently being planned for full implementation in the tropical Pacific measures the key quantities and therefore has a very high priority. (See Conclusion 1.) "We agreed that fields of sea level are the next most useful for valida- tion of ocean models and for an integral constraint on initialization. "We agreed that accurate measurements of sea level pressure are very useful in equatorial regions and especially in the southern hemisphere where not much other surface data exist. "We agreed that current meters are useful for validating ocean models but are not terribly useful for initializing them. "We also agreed that it would be useful to have near surface salinity for validating evaporation minus precipitation (E - P) in coupled models. "We pointed out that in the atmosphere, we have little information about boundary layer structure, especially in the divergent trade cumulus regions, which tend to lie in oceanic regions without upper air observa- tions. Since the ability of models to simulate and assimilate near surface data depends on boundary layer processes, we view this as a serious gap. "Another current gap is the equatorial winds over the Pacific, Atlantic, and Indian oceans, soon to be ameliorated over the equatorial Pacific by the TOGA TAO array. "It was also pointed out that some very simple and inexpensive mea- surements could be taken on existing platforms, e.g., radiometers on re- search ships or ships of opportunity, radiosondes on research vessels, etc. ''The only redundant observations that we could identify were expend- able bathythermographs (XBTs) that overlap the TOGA TAO array. Oth- erwise we viewed all the existing observational elements that we know about as complementary, rather than redundant, a statement perhaps of the sparsity of existing observations. (See Conclusion 4)." This response gives a rather clear view of the priority order of various kinds of observations, a view that generally agrees with those of panel members. It also is in general agreement with statements and priority rankings already stated or being developed by the TOGA Scientific Steering Group and by the Joint Scientific Committee/Committee on Climatic Changes and the Ocean's Ocean Observing System Development Program with regard to post-TOGA prediction efforts. The following subsections give further dis- cussion, in the same order, of the different observations mentioned above.
THE PRINCIPAL OBSERVABLES 21 This is followed by a final subsection dealing with the critically important but somewhat different (because of the existing heavy operational compo- nent) issue of atmospheric measurements. SURFACE WIND STRESS Uncertainties in the surface wind stress are recognized as one of the most serious impediments to advances in El Nino/Southern Oscillation (ENSO) prediction studies. Limited spatial coverage, gaps and irregularities in tem- poral sampling, errors in wind speed measurements, and uncertainties in the transfer coefficients can all contribute to uncertainties in the surface forc- ing. Busalacchi et al. ( 1990) noted that on seasonal time scales critical differences in tropical Pacific wind stress products, on the order of 0.2 to 0.4 dynes/cm 2 , were found in wind regimes of surface convergence and significant gradients such as the Inter-Tropical Convergence Zone and the South Pacific Convergence Zone. On interannual time scales the depen- dence on data coverage was clearly evident. Away from the major shipping lanes, root-mean-square differences among the wind stress anomalies for several products spanning 1979 to 1983 reached up to 0.5 dynes/cm 2â¢ The wind-forced ocean global circulation model experiments of Harrison et al. (1989) demonstrated that the equatorial wind changes during the 1982-1983 El Nino were so robust that most of the wind products considered resulted in satisfactory hindcasts of the equatorial ocean dynamic height. However, the same did not apply to the SST simulations or the dynamic height vari- ability off the equator. Subsequent design studies (Harrison, 1989) indi- cated that the equatorial dynamic height variability could be hindcasted successfully with observations of the wind field within the equatorial wave guide (Â± 3Â° latitude). SST and zonal current simulations required informa- tion on the wind field variability out to Â± 7Â° latitude, a reflection of how strongly the tropical oceans are wind driven. Sampling studies of the tropi- cal Pacific wind field (Harrison and Luther, 1990) suggest that the minimal resolution required would be daily measurements every 10Â° to 15Â° longitude and 2Â° to 3Â° latitude. Temporal sampling considerations are discussed by Halpern (1988) and Legler (1991). Early in TOGA it became obvious that an increase in the number of high-quality in situ wind observations was the only near-term solution if the monthly mean wind stress for the tropical Pacific was to attain an accuracy of 0.1 dynes/cm 2 [0.5 m/s, or 0.01 Pa (see Table 1)]. The TOGA TAO array is presently being implemented in response to this need for improved wind observations (see Conclusion 1). By the end of TOGA, approximately 70 TOGA TAO moorings (Figure 2) will be measuring the surface wind velocity, air temperature, SST, and upper-ocean thermal structure within 8Â° of the equator across the entire equatorial Pacific. For the purpose of
180Â° ~ 40Â°N 40Â°N :WON 2Qâ¢N EQ EO 20"S 20â¢s c ~ ~ i! c 400S 4oâ¢s ~ ::z: ~ r., g 100Â°E 120"E 140Â°E 1110â¢E 180" 1110Â°W 1400W 120"W 1000W aoow ~ :ll:i â¢ Atlas moorings o with current measurements ~ ~ c FIGURE 2 TOGA Tropical Atmosphere Ocean (TAO) array (ITPO, 1992). ~
THE PRINCIPAL OBSERVABLES 23 constructing a wind stress forcing function, conversion of the wind velocity observations still requires measurement of humidity, SST, and surface air temperature in order to calculate the stability-dependent drag coefficient. Buoys routinely measure the last two parameters and can measure the first. It is worth emphasizing that the TAO winds provide not only a more suitable spatial and temporal sampling scheme for the equatorial zone than shipboard winds but also a uniquely homogeneous data set from the point of view of technique and quality control. Ship winds can be affected by ship- to-ship differences in instrumentation, sensor installation and maintenance, perturbations to the airflow by the hull, and the effects of ship motion in a seaway on the measured wind. All of these difficulties are eliminated or greatly reduced for the TAO wind data. Measurements of the wind field in the Indian and Atlantic oceans re- main limited. Expansion ofT AO-style wind measurements into these tropi- cal oceans, or into midlatitudes, is a matter that ought to be considered and debated in the context of developing programs such as the Global Ocean- Atmosphere-Land System. The views of the panel on this question are discussed below. Space-based observations of the surface wind field have only recently become available from a number of sources. Since mid-1987, surface wind speed information has been available from the passive microwave sensors of the Defense Meteorological Satellite Program' s Special Sensor Micro- wave Imager (SSM/I) satellite. Global coverage at 35-km resolution is achieved approximately every 3 days. A second similar satellite, thereby providing twice the sampling, was launched in late 1991. A number of techniques have been proposed to convert this wind speed information to wind velocity (Atlas et al., 1991). The National Meteorological Center (NMC) is presently considering the inclusion of the SSM/I wind speeds into its analysis/forecast system. Since July 1991, wind velocity observations have been available from the Earth Remote Sensing (ERS-1) satellite's ac- tive microwave scatterometer. The assimilation of these vector wind data is now being evaluated for implementation at the European Centre for Me- dium Range Weather Forecasts (ECMWF). In 1996 the National Aeronau- tics and Space Administration scatterometer will permit grided 2-day aver- ages of wind velocity with 50-km resolution. Neither the active nor the passive microwave sensors constitute a panacea for monitoring the tropical wind field, since both instruments suffer from a degraded signal below wind speeds of 3 m/s. An important reason to extend the lifetime, and eventually the geo- graphic extent (Conclusions 1 and 3), of the TAO array is a part of broader efforts to gain experience and confidence in establishing proper transfer functions for the satellite measurements and to understand their limitations in forming a global, or even a tropical, surface stress product. While stan-
24 OCEAN-ATMOSPHERE OBSERVATIONS dard TAO stations alone do not return such essential variables as surface wave measurements and surface roughness needed for a comprehensive ef- fort at "ground truth" for scatterometer winds, they do return data on the penultimate quantity of interest (surface wind), and they do form logistical nuclei and sensor platforms around which more focused, shorter-term ex- periments can be established. It remains to be seen whether, and when, scatterometers can be relied on to provide continuous data that are as near to the desired result (surface stress) as existing data from the moored sta- tions. The TOGA TAO array will play a critical role in this evaluation pro- cess in the tropical Pacific, particularly in the low-mean-wind regime of the western tropical Pacific, where high-frequency wind fluctuations may con- tribute significantly to the stress field. Indeed, for the long term, a mix of buoy data with passive (SSM/I) and active (scatterometer) microwave data is the best approach, blending the spatial coverage of remote sensing with the local-area accuracy of in situ observations. SEA-SURFACE TEMPERATURE TOGA requirements for SST sampling and accuracy are noted in Table 1. Beginning in 1985, construction of a global SST field was taken on by the National Oceanic and Atmospheric Administration's (NOAA) Climate Analysis Center. By now this product has been widely used, not only in ENSO research but also as the standard for long-range weather prediction in opera- tional centers in the United States and Europe. A great deal has been learned at the NMC about how to construct a global SST field. Initially, the field was based on voluntary observing ships (VOSs) data reported through the Global Telecommunications System (GTS), with climatological data blended in to fill gaps. A few years later, Advanced Very High Resolution Radiometer (AVHRR) retrievals of SST were gradually added, replacing climatology. Drifting and moored buoy data (Â± 0.1Â°C accuracy) were at first used for calibration of the AVHRR retrieval algorithms. As the vol- ume of buoy observations grew, particularly with expansion of the World Ocean Circulation ExperimentrrOGA Surface Velocity Program, it became possible to verify both VOS data and the ''calibrated" AVHRR data. In this vein the World Meteorological Organization (WMO, 1991) carried out an assessment of the VOS SST data. Principal lessons from such studies are as follows: 1. In general, the VOS SST retrievals are accurate only to about 1.0Â°C. This can be improved but to do so requires changes in equipment and methods; in particular, hull-mounted sensors of known characteristics that are independent of ship intake plumbing and other ship-induced errors are
THE PRINCIPAL OBSERVABLES 25 needed. There is future promise here, and it should be pursued, but the current error situation is sobering. SST anomalies during an average or composite El Nino only range from about 0.2Â° to 1.0Â°C in the crucial west- em Pacific/eastern Indian oceans' warm pool region (Rasmusson and Car- penter, 1982). It is worth noting that VOS SSTs from XBTs are less useful because of the common near-surface transient due to the heat capacity of the probe as it enters the water. 2. VOS coverage is insufficient to span the globe. Figure 3 illustrates coverage in a typical month. Large, wholly unsampled regions in the trop- ics and southern hemisphere are obvious. 3. AVHRR retrieval algorithms are vulnerable to 1Â° to 30C errors due to changes of aerosols in the atmosphere. Most of these changes are epi- sodic and unpredictable-volcanic events and desert dust storms that blow dust far over the oceans. Each such event is different in the amount and type of material injected into the atmosphere and in the altitude of injection, all of which affect the size, distribution, and decay time of the resulting SST bias or error. There does not seem to be any reliable way at present to calibrate through such events to useful accuracy except by reference to a network of direct surface observations. Developments in progress to deter- mine aerosol size distributions, concentrations, etc., with multiple-frequency lidars may provide means of improving the SST retrieval accuracy in the future. 4. The most accurate real-time SST data set in TOGA has come from drifting and moored buoys. NMC now relies heavily on buoy data, not only where VOS data are lacking but also to effect quality control of VOS data from well-traveled shipping lanes .. 5. The best-resolved SST products in space and time now use tech- niques that combine AVHRR data with other data. The Rosenstiel School of Marine and Atmospheric Sciences of the University of Miami has pro- duced SST fields for the tropical Pacific with joint use of AVHRR and drifter data; these can be shown to satisfy theÂ± 0.3Â°C accuracy requirement. The NMC SST field is now produced by using the buoy data to remove large-scale biases and trends in the AVHRR fields and to screen the VOS observations. Then the smaller-scale curvatures of the field are established by the AVHRR data, and all data are objectively blended to yield an SST map. From the TOGA experience it is clear that high-quality (Â± 0.1 OC) in situ observations must be made globally to produce a global SST field to the accuracy (Â± 0.3Â° to O.SOC) needed for TOGA and for future climate predic- tion research. NMC estimates that one in situ device of this accuracy per 5Â° square is about the necessary sampling, with denser requirements in some regions (Reynolds and Smith, 1993). About 1000 buoys in the global ocean
SHIP POINTAGE ~ !1: !1: " ): :l! lol lol lol lol lol 0 "' 0 N ~ "' 0 (l) 0 "' 0 <'l " 0 "' 0 (I) 0 (l) 0 N ~ 0 "' ~ ~ ~ !" < fjl i' s; EQI S i ~ c s -7 1:Â£1 u ~- z rÂ·~:-- I. 1 I ,.'Â·Â·,.:.:..,._LCÂ·,~--'-..,......----r.c . . -p- :r. ' "' ":Â·-.-Â· I I. Â·.:Â·\:,_~I . wÂ·Â·~~ , ........ j ~ ~ ~ ""' 0 c ~ ~Ci~FPfEFFtfft2J-~t 60S ::z: !:l l'rl ~ c, mois de JUILLET 1990 nombre d'observations : 123764 j !:l ~ ::! FIGURE 3 Typical VOS SST coverage in a month (IOC/WMO, 1990). c ~
THE PRINCIPAL OBSERVABLES 27 are required to meet this sampling, a factor of about 2 above the presently deployed census. Expansions of drifter deployments to reach this level in support of TOGA, the World Ocean Circulation Experiment (WOCE), and other programs are in the proposal stage now, and this momentum should be sustained after 1994 (Conclusion 1). Simultaneously, efforts to build high- accuracy SST packages for VOSs should be pursued, so that the practical accuracy and cost trade-offs between VOS and buoy SST data can later be studied on the basis of real systems. There remain some significant and fundamental physical questions about the meaning of SST, since boundary processes, diurnal variations, and skin effects can lead to differences on the order of tenths of a degree in the top meter or so of the ocean. It is not obvious that model parameterizations or measurement efforts should be cast in terms of a single SST or, if a single SST is used, how it should be defined and observed. From the standpoint of direct observations, the SST is a central independent variable in bulk for- mula methods of determining air-sea heat flux, while from a large-scale modeling standpoint the SST (really the temperature in a thick near-surface layer in most models) is the key variable by which to gauge the evolution of the coupled system. From both standpoints it is important to understand the relationship(s) of the SST (really the near-surface temperature profile) to the derived quantities-fluxes, time evolution of a model layer. This prob- lem and many of the other SST issues noted above are discussed in a recent report (JOI, 1992). UPPER-OCEAN THERMAL STRUCTURE Together with surface wind stress and SST, the field of upper-ocean thermal structure is central to the successful initialization of coupled atmo- sphere-ocean models. The primary source of upper-ocean thermal data globally is XBTs taken in large part from ships of the same VOS network (Figure 3) that supports surface observations. Only a small fraction of the VOS fleet operates XBTs. Figure 4 gives a proposed global network of XBT lines (U.S. WOCE Office, 1992). Table 3 lists the number of times some data, not necessarily a complete section, were collected along a par- ticular line during 1990. At first glance the coverage looks impressive, although it clearly falls short of including the entire proposed network. However, the data in fact are inadequate to generate monthly maps of up- per-ocean thermal structure, and only bimonthly maps (Figure 5) can be constructed. The coverage indicates large gaps in all the southern oceans. At the October 1991 meeting of the TOGA/WOCE XBT/Expendable Conductivity Temperature Depth (XCTD) Program Planning Committee, the continuation of the present network and implementation of the lines in the southern
PX ~ 0 ~ ~ ~ ~ ~ ro., - - - - - High DeneltY IJnM -----Low Density LineS ~ ~ W.it.fMWf.@fi%ifE TRANSPAC ~ ~ ~ FIGURE 4 Proposed global network of XBT lines (U.S. WOCE Office, 1992). TRANSPAC, Thermal Structure Monitoring Program 0 in the Pacific. ~
THE PRINCIPAL OBSERVABLES 29 TABLE 3 Occupation During 1990 of the VOS Lines Shown in Figure 4. (The first entry, PX-11 2, means that line PX-1 in Figure 4 was occupied two times in the year. Not all such enumerated "occupations" resulted in sampling along the complete line.) Pacific Lines Indian Lines Atlantic Lines PX-ln IX-ln6 AX-1/18 -2n2 -2/1 -2/18 -3no -3/14 -3/24 -4/8 -6/8 -4/13 -S/41 -1n -sn3 -6/0 -8/0 -6/0 -8/40 -9/10 -7/36 -9n2 -Ions -8/37 -10n2 -11n -9n4 -11m -12/13 -lOllS -12/44 -14/0 -11131 -12NO -lS/0 -12/6 -13/10 -18/0 -13/0 -14/3 -19/0 -14/0 -lS/3 -21/8 -lS/24 -16/6 -22/11 -16/0 -17/19 -24/0 -17/6 -18/40 -2S/O -18/3 -20/21 -26/0 -19/1 -21/9 -20/13 -22/S -21/0 -23/0 -26/0 -24/S -27/23 -2S/l -26/1S3 -27/0 -28ns -29/0 -31/167 -34/12 -36/0 -37n3 -38/12 -41/2 -43/10 -44/10 -47/11
~ GLOBAL DISTRIBUTION OF OBSERVATIONS c E ~ ~ c ~ _.,...., .............. _______ i , __ _ _ ~ ~=-~---~~~ ~ --- _c ~ .., 90!E 60 90 120 150 180 150 120 90 60 30 0 30E ~ ~ :::! c FIGURE 5 Typical bimonthly XBT coverage in Pacific and Indian oceans. ti
THE PRINCIPAL OBSERVABLES 31 hemisphere indicated on Figure 4 were given top priority. As part of NOAA's Climate and Global Change Program, efforts are under way to implement several of the lines in the South Atlantic and Pacific oceans. Regionally, the TAO array provides subsurface thermal data in the equatorial waveguide of the Pacific Ocean (Figure 2). A rare potential for observa- tional redundancy arises here since both TAO and XBT measurements pro- vide basin-scale coverage of low-frequency subsurface thermal variations between go N and go S. While XBT measurements may provide relatively high along-track and vertical resolution, TAO time series data have the advantage that they are not aliased by high-frequency fluctuations and thus more accurately represent low-frequency changes. Some adjustments in XBT sampling in the equatorial waveguide may be required when TAO becomes fully implemented, and both modeling and empirical studies are under way to determine the optimal mix of TAO and XBT measurements. As a practical matter, starting and stopping XBT measurements be- tween go N and go S with volunteer observers would be difficult if not impossible to establish as a reliable part of VOS crew routine, so that such adjustments in XBT sampling strategies may have to await the incorpora- tion onto VOSs of technological advances such as automatic VOS/XBT launchers. This means that there will be a useful period of VOS and TAO overlap in the equatorial waveguide, data from which ought to be used to study the problem of redundancy as carefully as possible. The combined WOCE/TOGA XBT network of Figure 4 comprises lines of two kinds. Most are directed toward broad-scale spatial mapping of the upper-ocean thermal structure and heat content on seasonal to interannual time scales. Sampling intervals are in the range of a few hundred kilome- ters along tracks. A small subset of these lines, defining a set of closed areas into which net fluxes of mass and heat can be studied, is designated for high-density sampling (eddy resolving, typically 10 to 50 km). Here the objectives are to determine spatial statistics of the temperature, salinity, and geostrophic velocity fields and to measure changes in large-scale circula- tion and transport. The two very different sampling requirements derive from the different scientific goals and form a clear example of why the development of any piece of "the observing system" must be considered from a range of scientific perspectives (Recommendation 4 ). A considerable amount of effort has gone into assessing the mapping capability of this network or portions thereof, the associated errors, the effects of deleting lines, etc. Instructive recent papers include Festa and Molinari's (1992) study of the Atlantic network; Meyers, Phillips, Smith, and Sprintall's (1992) work on space and time scales of tropical Pacific thermal variability; and similar work by Phillips, Bailey, and Meyers (1990) for the Indian Ocean. These papers and the references contained in them give a good overview of the current state of thermal mapping capability.
32 OCEAN-ATMOSPHERE OBSERVATIONS There are significant variations in the estimates of scales by correlation techniques. Scales tend to be shorter at depth than at the surface, so that the deep (400-m) fields, with scales on the order of 200 to 400 km, determine appropriate sample spacing. The tropical Atlantic seems to be more isotropic than the Pacific, where zonal scales exceed meridional ones. Proximity to western boundary cur- rents also leads to shorter scales. All such estimates are based on limited data, and the process is really an ongoing one in which estimates can and should be revised and improved as new data and new network elements (ship tracks) are added. In the case of thermal sampling, an ongoing scientific oversight group exists-the TOGA/WOCE XBT/XCTD Program Planning Committee-that takes account of the observational needs of more than one program and attempts on a continuing basis to make recommendations for optimizing the observing strategy. Despite its name it has considered more than just two specific instruments, as evidenced by its draft report decision to "estimate the relative contribution of different observing techniques to determination of the space/time variability of the tropical upper ocean thermal structure and recommend an appropriate overall sampling strategy." This is the sort of activity that fits squarely into our Recommendations 4 and 5 and that should be encouraged. Specific near-term recommendations of this panel are to define an appropriate sampling program for the Transpac region and to upgrade the present network in the southern hemisphere by implementing a number of specific lines. In general, the effort of this panel to address the problem of observing upper-ocean thermal structure from the perspective of more than one set of scientific goals is to be commended. We look forward to the continued refinement of a coherent sampling strategy by this panel in consultation with other groups (e.g., the TAO Implementation Panel) whose areas of interest clearly overlap. SEA LEVEL In the context of seasonal to interannual prediction, sea-level observa- tions are useful as an integral constraint on model initialization fields and as a model validation data set. Space-based techniques [e.g., GEOSAT and, more recently, the ERS-1 and the Ocean Surface Topography Experiment (TOPEX)/Poseidon altimeters] will yield routine global sea-level maps, but surface observations should be maintained until the need for them has been demonstrably superseded. The ERS-1 and TOPEX/Poseidon missions will afford a good opportunity over the next several years to demonstrate the utility of altimetry. Beyond the demonstration of utility lies the question of future continuity of a series of altimetric missions with instruments of known accuracy. Is such continuity as likely for the indefinite future as is mainte-
THE PRINCIPAL OBSERVABLES 33 nance of the conventional sea-level network? This is a question that must be answered before any action is taken on paring down the conventional network. In the Pacific (Figure 6) the station network is mature. Indeed, in keeping with the evolutionary observing system paradigm, it may be pos- sible to winnow this network to a degree in future years, at least on grounds of utility for climate predictions. This will depend on careful comparative studies of station data and altimetric data as the latter becomes more rou- tine; from this there may emerge some indications of stations that are re- dundant or that are biased by local coastal effects. Stations at which re- moval of geodetic effects is undertaken, using the Global Positioning System geodesy or other methods, are more valuable in the climate context than other stations. In the Atlantic and Indian oceans there remain several uninstrumented islands. Establishment and operation of stations on selected ones of these islands are a way in which the governments of some smaller, less developed nations can play modest but useful roles in global-scale climate research and prediction. Certain sea-level stations have records that are among the longest ex- tant geophysical series of any kind, and as such they have value beyond questions of seasonal to interannual prediction. Long-term global change considerations should be brought to bear on the continuation of these sta- tions, in keeping with the caveat attached to Conclusion 4. VELOCITY MEASUREMENTS The equatorial oceans are unique in the role that ocean dynamics play in generating and maintaining climatically significant SST anomalies. Near the equator, the horizontal components of the Coriolis force vanish, and ageostrophic flows in the surface and equatorial boundary layers become key determinants in advecting heat laterally and vertically. The post-TOGA observational network must incorporate a minimal array of direct horizontal velocity measurements in these critical boundary layers to ensure that model- based analyses of ocean circulation are in fact representative of the flow field in the real ocean. Models that cannot accurately simulate the ocean circulation controlling the evolution of ENSO SST anomalies will have limited skill in predicting ENSO. The feasibility of making such measure- ments was demonstrated before TOGA (Knox and Halpern, 1982). Long- term moored velocity measurements along the equator have been integrated into the TOGA TAO array, and the Pan-Pacific drifters program provides basin-wide estimates of the circulation in the surface boundary layer.
~ 40"N I I to> 40Â°N 2Qâ¢N 2Qâ¢N EO EO 2Qâ¢S 2oâ¢s Â§ ~ 4oâ¢s 4oâ¢s i! <:::1 ~ ::z:: !:l t>:l 100"E 120â¢E 140â¢E 160"E 180Â° 1soâ¢w 140"W 120Â°W 100"W eoâ¢w 1il c, Â£ Operational sites with ll. Operational sites 0 Planned X Required !:l satellite transmitters ~ :::! <:::> FIGURE 6 TOGA Pacific sea-level stations (ITPO, 1992). ~
THE PRINCIPAL OBSERVABLES 35 ATMOSPHERIC MEASUREMENTS As noted above, most long series of atmospheric measurements for climate research and prediction are made for operational meteorological reasons, as part of the World Weather Watch (WWW). There are alarming negative trends in the performance of the WWW network. More than 400 WWW Basic Synoptic Network upper-air stations lie between 300 N and 30Â° S; of these, TOGA has identified 150 stations as a minimal network for resolving planetary-scale variations (wavenumbers 0 to 4) (ITPO, 1992). Data receipts at ECMWF via the GTS show that many of the stations even on this minimal list are now reporting infrequently or not at all. Similar reviews at NMC (P. Julian, personal communication) indicate that upper-air reports from tropical Africa decreased by about 50 percent from 1989 to 1992. There has been a serious proposal to reduce normal U.S. reporting practice from two observations per day to one. U.S. observations are also being cut back as a consequence of defense reductions and closures of overseas bases. Civilian aircraft meteorological reports using automatic equipment have increased, which is to be encouraged, but manually trans- mitted enroute reports have been reduced at the same time. It is now often the situation that an aircraft that is configured to give automatic enroute reports finds itself in a remote region out of range of receiving stations and thus gives no report at all. Often these remote regions are over the tropical or southern oceans, where no other compensating upper-air observations exist. This fraying of the WWW network involves losses of two distinct kinds. The first and worst is outright loss (forever) of measurements. The second is the effective loss of measurements for real-time purposes because of communication or other data transfer failures. Even if such errant measure- ments appear eventually at the major meteorological centers, they may do so too late to be incorporated into operational products such as surface wind analyses. They then become measurements without much significance, or they acquire significance only much later after a costly reanalysis effort. In the interim-perhaps many years-their use in climate studies via real-time analyses is delayed, and the archive of real-time analyses is needlessly degraded. Repairing the damage to the network is a difficult international prob- lem. Many of the failing tropical stations and data streams originate in less- developed countries, where the costs of operating meteorological stations are seen as large and less than an urgent national priority. Direct financial assistance for hardware, expendable supplies, or training, as circumstances indicate, may be the only realistic means of shoring up the network in these regions in an expeditious manner. Conclusion 2 calls for strong action in this area. Appeals to international collaboration and shared responsibility for global forecasting are less immediately compelling.
36 OCEAN-ATMOSPHERE OBSERVATIONS Despite the overwhelming importance of the operational network for atmospheric measurements, a few key TOGA stations that were established using research funding still exist. The tropical Pacific set of unmanned pulsed Doppler wind profilers includes such stations, and their continuation after 1994 should be secured. There are upper-air stations in Ecuador, the Cook Islands, and other locations that were established in response to TOGA needs. Sustaining these stations after the end of TOGA is an important objective and one not likely to be achieved if the countries involved are simply left to arrange the continuation from their own resources. Finally, priority should be given to improving the technology of mak- ing and communicating surface and upper-air observations to the major data centers. In addition to the previously mentioned problems of temporal and horizontal spatial coverage, the data reported from operational soundings do not adequately resolve vertical structures (especially in the boundary layer) that play a key role in air-sea interaction. In some cases the degradation of observed soundings is related to assumptions about data storage and trans- mission rates that are not applicable to modem data processing and commu- nications technologies. Priority should therefore be given to replacing the WWW's current technology with simple and inexpensive systems that im- prove the quality 'and availability of surface and upper-air observations from existing platforms on islands, research vessels, and ships of opportunity.