Ventajas

2.2.1 Studies of mesoscale ocean processes

T he m esoscale or "eddy-containing" band is generally defined as enco m passin g oceanographic phenom ena with spatial scales o f betw een 50 and 500 km, and having

periods o f betw een 20 and 150 days [McClean et a l , 1997], although the upper lim it

o f the spatial scales is extended to 1000 km by som e authors [Fu and Cheney, 1995].

The largest errors in the altim eter measurem ent system, m ost notably due to poor orbit determ ination, are generally greater than 1000 km , m aking studies o f the m esoscale less prone to error. This m eant that even the relatively crude early altim eters such as G eos-3 and Seasat were able to study mesoscale signals w ith rem arkable success.

C heney et a l [1983] used collinear repeat tracks o f Seasat data to rem ove the tim e- in varian t geoid in order to study the sea surface height variability associated w ith dynam ic oceanographic phenom ena. Long w avelength orbit and tide m odel errors w ere rem oved by rem oving linear trends from individual tracks o f altim eter data. T h ey found that the m eanders and eddies associated w ith m ajor cu rren t system s resulted in high sea surface height variability. Low variability (< 5 cm ) was found to

dom inate over 70% o f the observed global ocean. M enard [1983] used Seasat data to

com pute the variability o f geostrophic velocities from along-track sea surface height slopes, in order to estim ate eddy kinetic energy. H igh eddy kinetic energy w as found to be associated w ith the G ulf Stream and the Kuroshio.

F u e t a l [1987] used data from the Geos-3 altim eter to study the seasonal variability o f the G ulf Stream . Time series o f sea level variations w ere constructed at crossover locations (w here the ascending and descending satellite ground-tracks intersect) to reveal a pronounced seasonal variability.

Sim ilar m ethods have been em ployed for studies o f ocean variability and eddy kinetic

energy using the m ore recent altimeters. G eosat [Sandwell and Zhang, 1989], ERS-1

[H eywood et a l , 1994] and TO PEX/POSEIDON [McClean e t a l , 1997].

2.2.2 Observations of large-scale ocean circulation

The large-scale generally refers to length scales greater than 1000 km , w hich w as the

arbitrarily defined upper lim it o f the m esoscale [Fu and C heney, 1995]. A t these

lo n g e r w av elen g th s, errors due to poor o rb it d eterm in atio n , tide m o d els, and p ro p a g a tio n d elay co rrectio n s, begin to do m inate. T he sig nal to n o ise ratio

c o n s e q u e n tly b e c o m e s s m a lle r fo r all a ltim e te r s y s te m s e x c e p t fo r TO PEX /PO SE ID O N , although recent im provements in the orbits for ERS-1 and 2 due to Scharroo and Visser [1998] have made large-scale circulation studies w ith these instrum ents a possibility.

W ith the advent o f TO PEX /POSEID O N, it is now possible to study the general ocean circulatio n through direct m easurem ents o f the tim e invariant dynam ic topography

[Nerem et a l , 1994; Stam m er and Wunsch, 1994; Tapley et a l , 1994]. H ow ever, the

geoidal variations are poorly know n at w avelengths below about 2500 km [Nerem et

a l , 1994], w hich results in a contam ination of the tim e invariant dynam ic topography at w avelengths shorter than this.

It w as possible to study large-scale variations from earlier, less accurate m issions, w ith a careful treatm ent o f orbit error. For exam ple, Chelton et a l [1990] studied the large-scale circulation o f the Southern Ocean, extracting estim ates o f the sem iannual, annual and inter-annual signal. This revealed the generally regional nature o f sea level variability in the Southern Ocean.

S everal authors have also used altim etry to estim ate variations in m ean sea level.

M inster et a l [1995] used TO PEX /POSEID O N to determ ine a m ean sea level rise in the northern hem isphere o f 4.2 ± 2.5 mm/yr, w ith a corresponding value o f 3.7 ± 1 .9

m m /y r in the southern hem isphere. A global valu e o f 5.8 ± 0.7 m m /yr w as

determ ined by Nerem [1995] using the same instrum ent.

2.2.3 Validation of ocean general circulation models

W ith the revolutionary advances in com puting capability over the last decade, it is now possible to run num erical high-resolution ocean general circu latio n m odels globally in a realistic configuration w ith realistic surface forcing. E xam ples are the

Parallel O cean Clim ate M odel (POCM ) of Sem tner a nd Chervin [1992] and the O cean

C irculatio n and C lim ate A dvanced M odelling P ro ject (O C C A M ) o f W ebb e t a l

[1998] bo th w ith resolutions o f 1/4% and the P arallel O cean P rogram (PO P) o f

D u ko w icz a n d Sm ith [1994], w ith its resolution o f 1/6% S trin gent m o d el-d ata com parison s are a necessary first step in order to fully use the m odel results in u n derstanding ocean general circulation, and to com bine observations w ith m odels

[Stam m er et a l , 1996].

Fu a n d Sm ith [1996] used TO PEX /PO SEID O N observations o f sea surface height to com pare the m ean dynam ic topography, the sea surface height v ariability and the

annual and inter-annual signals with the predictions o f the PO P global m odel. They found th at the sea level variance predicted by the m odel is generally a factor o f 2 low er than observed. A higher spatial resolution was deem ed to be necessary in order to fully resolve the m esoscale eddies.

S ta m m er et al. [1996] com pared the p red icatio n s o f the P O C M m o d el w ith TO PEX /PO SE ID O N observations from the same period, and found that the sim ulated am plitudes o f sea surface height variability were low er than the observations by a factor o f 2 to 4 over a broad spectral range. In a sim ilar analysis, M cC lean et al.

[1997] found that the PO CM and POP models explained 50% and 60% respectively o f the global sea surface height variability. They concluded that the low er energy content o f the PO CM m odel was a consequence of the low er resolution.

2.2.4 Development of ocean tide models

T h e a v a ila b ility o f p re c is e sea s u rfa c e h e ig h t m e a s u r e m e n ts fro m TO PEX /PO SE ID O N , along with major developments in num erical tide m odelling and data assim ilation, has resulted in the generation o f a significant num ber o f new global ocean tide m odels. Since 1994, over 20 new m odels have been developed, som e o f

w hich are assessed and intercom pared by Andersen et al. [1995] and Shum et al.

[1997]. T he tests perform ed by Shum et al. [1997] indicate that the m odels agree

w ithin 2 to 3 cm in the deep ocean, and provide a significant im provem ent over the

classical m odel o f Schwiderski [1980] by around 5 cm rm s. T hese m odels can be

used to rem ove the ocean tide signals from altim eter m easu rem en ts, allow in g im p ro v ed estim ates o f dynam ic topography and ocean v ariab ility to be m ade. S ig n ifican t d ifferences betw een m odels are ho w ever found in shallow w aters, highlighting the problem s in accurate tide prediction in those areas.

Tw o m ethods are available for the analysis of tides using altim eter data: the classical harm onic analysis, and the more recently developed response analysis, both o f w hich

are discussed in chapter 5. Cartwright and Ray [1990] used G eosat data to derive

estim ates o f the diurnal and semidiurnal ocean tides, resulting in a near-global ocean tide m odel. Their use o f the response analysis approach ensured a com plete definition o f the diurnal and sem idiurnal species.

Since the ERS-1 and 2 orbits are Sun synchronous, the response analysis approach

m ust be used to estim ate ocean tides from these m issions. A ndersen [1994] used this

technique to successfully extract the semidiurnal and diurnal tidal constituents from ERS-1 data in the northern N orth Atlantic and adjacent seas. In a global ocean tide

m odel, A n dersen [1995] com bined ERS-1 w ith T O P E X /P O S E ID O N data to y ield im provem ents over previous m odels developed solely from T O P E X /P O S E ID O N data. T he increased density o f ERS-1 ground-tracks resulted in a higher sp atial resolution for the com bined model, which is particularly im portant in coastal regions w here sm aller scale tidal effects are usually present.