ENSO as a manifestation of the interannual variability of the oceanatmosphere system

The existence of El Nino (an anomalous rise in water temperature to the west of the equatorial zone of the Pacific Ocean accompanied by suppression of the upwelling of cold deep water, rich with biogenes, and therefore having disastrous consequences for fisheries in the coastal regions of Ecuador and Peru) and Southern Oscillation (fluctuations of surface pressure, dominating winds as well as air temperature and precipitation in the tropics and neighbouring regions of the Indian and Pacific Oceans) has been known for a long time. But the fact that El Nino and Southern Oscillations represent two sides of one and the same phenomenon became clear only in the late 1960s, due mainly to the work of Bjerknes. Since that time this phenomenon has been called El Niño/Southern Oscillations or in abbreviated form ENSO. We reproduce Bjerknes' chain of arguments, taking subsequent refinements into account (see Philander and Rasmusson, 1985, in particular).

For time scales of several weeks (the relaxation time of the statistically steady state of the atmosphere) and longer, the large-scale atmospheric circulation in the tropics is controlled by a system of Walker cells, every one of which is characterized by the ascent of air over a heated continent and a neighbouring area with high sea surface temperatures, by air transport in the upper layers of the troposphere towards cold water areas, by descending air over this area, and by trade winds closing a circulation cell in the lower layers of the troposphere (Figure 5.11). In the Pacific Ocean, ascending branches of the Walker cell are located over the Indonesian archipelago and the northern part of Australia, along the inter-tropical convergence zone to the north of the equator and along the Pacific convergence zone south-west of the Pacific Ocean; the descending branch of the Walker cell is located over the south-east part of the tropical zone where the sea surface temperature is less than in the western part by 8 °C in March and by 13 °C in September. The Walker cell is closed by the south-eastern trade winds.

The south-eastern trade wind is are subject to marked seasonal oscillations. It is strengthened and penetrates far to the north in August and September when the inter-tropical convergence zone is situated near the parallel 12 °N, and, on the other hand, it is weakened and displaced to the south in February and March when the inter-tropical convergence zone is in the vicinity of the equator. The westward component of the south-eastern trade winds leads to the accumulation of warm water in the western part of the Pacific Ocean and to upwelling of cold deep water in the east. Accordingly, the thickness of the upper mixed layer changes: it exceeds 150 m in the western part of the tropical zone and decreases, down to zero, in fact, in the vicinity of the South American coast. It is clear that the strengthening of the south-eastern trade winds in

S.America Africa Indonesia.

S.America Africa Indonesia.

Figure 5.11 Schematic representation of Walker cells in the equatorial belt of the Earth.

August and September must contribute to a reduction in the atmospheric temperature, a rise in the surface pressure, a weakening of convection and a decrease in precipitation in the eastern part of the tropical zone, and to opposite changes in the west.

Now, imagine that the surface temperature rise has occurred in the south-eastern part of the tropical zone of the Pacific Ocean, that is, El Niño has appeared. It should lead to a drop in the zonal gradient of surface pressure, to a weakening of the south-eastern trade winds, to a rise in the mean ocean level in the eastern part and a drop in the western part of the tropical zone, to a decrease in inclination of the free ocean surface and the thermocline underlying the upper mixed layer, to a weakening of the South Equatorial Current and an amplification of the easterly propagating subsurface water mass transport, to a westward displacement of the boundary of the intensive convection area in the atmosphere (this boundary is easily identified from satellite measurements of outgoing long-wave radiation and is assumed to coincide with the 240 W/m2 contour or with the 27.5 °C surface isotherm), and to an increase in precipitation in the central part of the equatorial zone of the Pacific Ocean. These are the consequences of El Niño's compliance with the warm phase of ENSO.

Next comes the cold phase of ENSO, with its inherent anomalous reduction in surface temperature in the south-eastern part of the tropical zone of the Pacific Ocean. This phase is identified with the opposite of El Niño which was given the name La Niña. The consequences of La Niña are opposite to those described above: the drop in water temperature in the south-eastern part of the tropical zone determines a local rise in surface pressure, strengthening of the south-eastern trade winds, a rise in the mean ocean level in the west and a decrease in the east, an increase in the zonal inclination of the free surface and thermocline, intensification of the Southern Equatorial Current, and weakening of easterly propagating subsurface water mass transport, a shift to the west of the convective zone in the atmosphere and a decrease in precipitation in eastern and central parts of the equatorial zone of the Pacific Ocean.

Certainly, the sole fact of the appearance of the local anomaly of the surface temperature does not yet mean that its influence will be felt at large distances from it: it is necessary to have a combination of appropriate conditions. For example, a local rise in surface temperature in the region of downward atmospheric motions can turn out to be insufficient to establish intense convection. On the other hand, a local rise in surface temperature in a region of upward atmospheric motions favours intensification of convection, an increase in precipitation, enhancement of heat release due to the phase transitions of water vapour and subsequent intensification of upward vertical motions. Because of this the influence of the surface temperature anomaly is determined not only by its magnitude, sign and location but also by the season of the year. It is no mere chance that the mature stage of El Niño development falls in a period of maximum sea surface temperature, that is, in December-February.

The short-term rise of water temperature in the eastern part of the equatorial zone, as well as a weakening of the south-east trade winds and even the appearance of anomalous westerly winds in the central part of the equatorial zone in September-November of the preceding year serve to herald El Niño. In March-May a rise in temperature is felt over eastern and central parts of the equatorial zone of the Pacific Ocean. It is also maintained here in June-August. The first signs of the end of El Niño and of the restoration of normal conditions are found in September-November. This is demonstrated by a decrease in the South Oscillation index (the difference of surface pressure between points located in the south-eastern part and in the Indonesian-Australian sector of the Pacific Ocean), weakening of westerly winds in the central part of the equatorial zone, and suppression of atmospheric convection in the east. In December and during January and February of the following year the degeneration of El Niño continues. Positive anomalies of surface temperature in the equatorial zone disappear (warm phase of ENSO terminates) in June-August. The cold phase of ENSO starts in September-October. It is characterized by sharp strengthening of the south-eastern trade winds, by a westward shift from the 180° meridian of the intense convection zone, by a rapid increase in the difference in the surface pressure between Tahiti and Darwin, and by the appearance of negative anomalies of surface temperature in the equatorial zone of the Pacific Ocean. In June- August of the second year after the occurrence of a mature phase of El Niño the negative anomalies of surface temperature spread over the whole equatorial zone and stay there until the end of the year. Thus, the duration of the complete cycle of ENSO is about three years.

The sequence of events described above is typical of ENSO, although it does not exclude individual pecularities in certain years. For example, the El Niño of 1982-3 differed from all the others by an extremely large rise in temperature in the eastern part of the equatorial zone of the Pacific Ocean. Suffice it to say that for the preceding 30 years the ocean surface temperature east of the 140 °W meridian had never been higher than 29 °C. In 1982 the 29 °C isotherm reached the coast of South America. This led to a reduction, practically down to zero, of the difference in surface temperatures between the eastern and western parts of the equatorial zone of the Pacific

Ocean, to a sharp strengthening of westerly winds, to an increase in frequency of typhoons and thunderstorms in the central part, to a degeneration of the easterly subsurface water mass transport, and to the disappearance of thermocline inclination along the equator.

No two cycles of ENSO ever repeat each other. The same can be said with respect to the interval between two subsequent cycles of ENSO, as attested by the chronology of El Niño events. Indeed, since 1935 El Niño has appeared 13 times: in 1940-2, 1946-7, 1951-2, 1953-4, 1957-8, 1963-4, 1965-7, 1969-70, 1973-4, 1977-8, 1982-3, 1986-7 and 1991-2. In two cases (in 1940-2 and in 1982-3) it was extreme: the respective South Oscillation index exceeded the amplitude of seasonal oscillations of surface pressure by 2.5 times. Therefore, ENSO represents a distinct aperiodic phenomenon.

All models of ENSO can be divided into two types. The basis of the models of the first type is the assumption that the ocean-atmosphere system in the tropics has two quasi-equilibrium states complying with warm and cold phases of ENSO. It is also assumed that the transition from one state to another is determined by an internal instability inherent in the system which acts like a trigger and is excited by high-frequency (compared to the ENSO frequency) stochastic disturbances. But even in the presence of such disturbances the instability occurs only when the oscillations of the system become critical. Because of this the appearance of the instability and, hence, the tendency towards the development of events and the duration of the existence of one or another state depend not only on high-frequency stochastic disturbances but on low-frequency deterministic oscillations of the system.

In models of the second type ENSO is determined by an instability of low-frequency equatorial disturbances generated by the ocean-atmosphere system. The appearance of the instability can be explained with the help of the following qualitative considerations (see Philander, 1985). Let us assume that in the western part of the equatorial zone a local positive anomaly of surface temperature has occurred for some reason, as a result of which a convective zone with upward vertical motions and wind velocity convergence in the low layers of the atmosphere will have formed immediately over it, that is, with westerly winds towards the west and easterly winds to the east of the anomaly. The wind velocity convergence in the low layers of the atmosphere will lead to ocean current velocity convergence in the upper mixed ocean layer and, hence, to a downwelling and deepening of the thermocline. As is known, the sea surface temperature decreases from west to east. Therefore, an advective transport to the west of the anomaly will contribute to a further temperature rise and to its eastward displacement.

A different situation arises to the east of the anomaly. Here a change in temperature is determined by the competition of two factors: by upwelling created by local easterly winds, and by downwelling formed by baroclinic Kelvin waves propagating towards the east (these are induced by westerly winds at the western periphery of the anomaly). The first factor causes a decrease in surface temperature; the second leads to its increase. Domination of one factor or another depends on the relative intensity and the zonal extent of bands of westerly and easterly winds. Since, all other things being equal, the intensity and zonal extent of a band of westerly winds on the equatorial beta plane are greater than those of easterly winds, the downwelling will dominate over the upwelling and in the process of its evolution the positive anomaly will increase and be displaced to the east.

An increase in a negative surface temperature anomaly can be explained in a similar manner: a temperature drop results in suppression of convection, decrease in westerly winds, domination of upwelling over downwelling and, eventually, further temperature reduction to the east of the anomaly. If this temperature drop is accompanied by a decrease in the dimensions of the convective zone in the atmosphere and by strengthening of easterly winds then intensification of upwelling, a further temperature drop and eastward displacement of the negative anomaly are inevitable.

We illustrate the above by calculations carried out by McCreary (1986). The model used by McCreary (1986) describes the instantaneous response of the first baroclinic mode in the atmosphere to a heat inflow from the ocean, and the evolution of the ocean upper mixed layer forced by changes in momentum and heat fluxes at the ocean-atmosphere interface. In a model of the upper mixed layer the effects of mechanical and convective mixing, entrainment at the thermocline boundary and momentum and heat advection are taken into account.

The results of numerical experiments presented in Figure 5.12 are in respect of three different distributions of land and ocean. In the first experiment the ocean is approximated in the form of a zonally symmetric (with respect to the equator) band surrounding the whole Earth, in the second experiment the zonal ocean is divided into two disconnected basins with the same angular distance between boundary meridians. In the third experiment the western ocean basin decreases by half, and the remaining part of the zonal band is occupied by land adjoining a large ocean basin to the east and a small ocean basin to the west. Such a configuration is reminiscent of the system of the Indian and Pacific Oceans. In all three cases the initial distribution of temperature and upper mixed layer thickness is considered to be homogeneous, and motions in the ocean and the atmosphere are considered to be absent. The

15000 km

Figure 5.12 Time-space variability of the deviation of the upper mixed layer temperature from the thermocline temperature at the equator for different ocean configurations (a, b, c), according to McCreary (1986). Zones in which temperature deviations exceed 9 °C are shaded. Further explanation may be found in the text.

(continued)

288 Large-scale ocean-atmosphere interaction

Anime Warrion Girl Drawing

Ocean 1 Ocean 2

Figure 5.12 (continued).

Ocean 1 Ocean 2

288 Large-scale ocean-atmosphere interaction

Figure 5.12 (continued).

perturbations of the initial distributions of the desired characteristics are introduced by changes in the wind stress field during the first 100 days of integration. These changes are then excluded and perturbations generated by the model are reproduced without the superposition of any external forcings. Figure 5.12 shows the separate fragments of the time-space variability of

If 000

7500 15000 M-

Ocean 1

Ocean 2

Figure 5.12 (continued).

7500 15000 M-

Ocean 1

Ocean 2

"Indian"

"Pacific"

If 000

Figure 5.12 (continued).

temperature deviations in the upper mixed layer from the thermocline temperature at the equator for the three above-mentioned ocean configurations. Initially, temperature deviations are assumed to be identical and equal to 10°. It can be seen that in the first case (a) the formation of positive and negative deviations is terminated after 1000 days, and then in fact they do not change practically but are displaced slowly to the east. In the second case (b) the situation is different: positive deviations only appear at the western boundary of the eastern ocean. Then they slowly propagate to the east and in 2800 days reach the eastern boundary. As this takes place the negative deviations are formed next to the western boundary of the eastern ocean, and, at the western boundary of another ocean, positive temperature deviations arise. These do not remain but, rather, are slowly displaced to the east. Upon reaching the eastern boundary of the western ocean the positive deviations act in the vicinities of the western boundary of the neighbouring (eastern) ocean through the westerly transport in the atmosphere. As a result, positive deviations are formed here. Then the cycle of temperature oscillations in the system of the two oceans is repeated.

Case (c) is crucially different from the first two cases. Now positive deviations after the termination of a period of relaxation are localized at the eastern boundary, and negative deviations are localized at the western boundary, of the small ocean basin. Such a state turns out to be stable in the sense that it does not change over time. In contrast to this, low-frequency oscillations with periods of about five years occur in the large ocean basin. This last peculiarity of the solution, as well as the existence of a constant temperature rise (and therefore intense convection) in the vicinity of the land dividing both oceans, and also the slow displacement of temperature deviations to the east, have much in common with observational data for the equatorial zone of the Indian and Pacific Oceans.

The calculated results presented in Figure 5.12 reproduce the successive interchange of warm and cold phases of ENSO but leave the question as to the causes of their alternation unanswered. An answer to this question is given by Graham and White's (1988) conceptual model. We describe it briefly starting from the time of the origin of a positive surface temperature anomaly in the central part of the equatorial zone of the Pacific Ocean.

It has already been mentioned that an anomalous increase in the surface temperature leads to the appearance of anomalous westerly winds in the equatorial zone and to a weakening of easterly winds beyond its limits. This, in turn, must lead to an enhancement of cyclonic vorticity in the wind velocity field, intensification of upwelling, and decrease in surface temperature in the central part of the extra-equatorial zone. The negative surface temperature anomaly formed here is transferred by baroclinic Rossby waves to the western ocean boundary where, as a result of reflection, these waves are transformed into baroclinic Kelvin waves propagating towards the equator and then along it to the east. The negative anomaly of surface temperature is carried along, together with the baroclinic Kelvin waves.

But since the phase velocity of the baroclinic Rossby waves is much less than the phase velocity of baroclinic Kelvin waves, then a temperature rise in the eastern part of the ocean, created by the transport of the anomaly from the central part of the equatorial zone, is followed by a temperature decrease caused by the transport of the negative anomaly from the central part of the extra-equatorial zone. As this takes place the lag between times of increase and decrease in temperature will be determined by the time of displacement of the negative anomaly from the central part of the extra-equatorial zone into the central part of the equatorial zone, amounting to one or two years. Similar (in terms of sequence but not sign) changes must occur when a negative anomaly of surface temperature appears in the central part of the equatorial zone.

Thus, in accordance with the concepts stated, a change in ENSO phases is explained by the influence of baroclinic Rossby waves whose domain of existence is restricted by extra-equatorial latitudes. In particular, from this it follows that any attempt to predict the complete ENSO cycle without taking into account the interacting processes in equatorial and extra-equatorial latitudes of the ocean will be condemned to failure.

Response of the ocean-atmosphere system to external forcing

After listing mathematical methods of analysis of sensitivity we shall discuss the results of numerical experiments on the equilibrium (steady-state) response of the climatic system to external forcing. In reality, the response of the climatic system is time-dependent in nature because of the large heat capacity of the ocean, hence the results obtained cannot be used directly to predict the climate evolution for a short period (on the ocean relaxation time scale). The above numerical experiments have another purpose: to give an idea of the possible limiting changes of the climatic system. That discussion will be followed by a brief description of the transient response of the ocean-atmosphere system to an increase of atmospheric C02 emission resulting from human activities.

0 0

Post a comment