FLOODS OF LAKE EYRE
LA NIÑA DE AUSTRALIA
LA NIÑA DE AUSTRALIA - CONTEMPORARY AND PALAEO-HYDROLOGY OF LAKE EYRE
Kotwicki, V., Allan, R. (1998). La Niña de Australia - Contemporary and palaeohydrology of Lake Eyre. Palaeogeography, Palaeoclimatology, Palaeoclimatology,144:265-280.
Robert Allan, Climate Impact Group, Division of Atmospheric Research, CSIRO Australia, Private Bag No 1, Mordialloc, Vic, 3195, Australia
ABSTRACT The available records of inflows to Lake Eyre cover the time span of forty five years only. Their quality is questionable. As longer time series of precipitation records is available, the observed series of inflows to Lake Eyre can be extended with the help of a rainfall-runoff model. Further reconstruction of the inflow series can be achieved by examining local evidence or with the help of proxy data. However, the limitations of these extensions and reconstructions of inflows are severe. The process of inflows to Lake Eyre could be considered one of the most convincing manifestations of hydrological uncertainty. However, it is apparent that major flooding episodes in the Lake Eyre Basin are most often associated with La Niña phases of the El Niño Southern Oscillation (ENSO) phenomenon.
The paper presents a hydrological description of the Lake Eyre basin and large-scale contemporary and palaeo climatic patterns which affect water balance of the catchment. A simple water balance model has been constructed to determine storages of the lake when annual evaporation reaches equilibrium with annual runoff.
The basic business of hydrology seems to be straightforward, as its fundamental formula, the water balance equation, can be expressed in a simple statement:
Runoff (R) = Precipitation (P) - Evaporation (E)
However, a small relative error in P or E, which are normally of similar magnitude and both larger than R, can cause a large relative error in R. Hence, as Eagleson (1970) notes, sometimes it is indispensable to recollect that the true nature of the evaporation - condensation - precipitation - cycle was not fully accepted until the late nineteenths century. Even now, rainfall is measured nowhere in the world with sufficient accuracy and, as Shuttleworth (1983) remarks: ...our ability to describe and measure evaporation directly is in its infancy. These issues are of particular relevance in marginal, arid and semi-arid areas when small changes in either rain or evaporation can bring about large changes in runoff.
The hydrology of arid and semi-arid areas is generally recognised as one of the highest forms of art and science. Difficulties in the determination of rainfall, streamflow and evaporation rates are legion (Pilgrim et al.1988). These problems multiply when the hydrology of any associated lakes is taken into consideration. Paradoxically, however, because of their high responsivity to even small forcing factors, arid and semi-arid lakes may possibly serve as sensitive indicators of emerging climatic changes, another area where art and science hold equal sway in current thinking.
This paper is not looking for an evidence of palaeofloods which can be found elsewhere in this volume. Its purpose is to consider meteorological conditions - especially El Niño - Southern Oscillation - which lead to the filling of Lake Eyre, likely climatic fluctuations and their effect on the basin and to use available data in a simple and robust model to simulate the behaviour of both the basin and the lake under a variety of climatic conditions.
Lake Eyre and its basin
Hydrological characteristics of the Lake Eyre basin reflect its long and relatively uniform geological history. In the late Cretaceous, Australia separated from other continents and began drifting on a plate less disturbed than any other on Earth. There was no dramatic mountain building or volcanic activity; long periods of tectonic stability and erosion levelled the continent. Gradients became very low, with shallow and infertile soils covering old bedrock, and traces of disorganised river systems negotiating numerous dune fields and petering out in abundant sinks and flats. Chappell (1981) paints a vivid picture of late Quaternary environments and their effects upon the hydrological sensitivity of the continent.
Lake Eyre, a large depression in arid Australia which rarely fills with water, attracts the interest of limnologists, hydrologists, geomorphologists and ecologists all over the world. Voluminous literature on the subject is cited throughout other chapters of this volume. For a detailed hydrological description of the basin see Kotwicki (1986), while comprehensive characteristics are given by Kotwicki and Isdale (1991); large-scale meteorological aspects are examined by Allan (1985). Two major fillings of Lake Eyre in 1950 and 1974 are documented in separate monographs (Lake Eyre Committee, 1955; Bonython and Fraser, 1989, respectively), which also contain extensive reference lists. Isdale and Kotwicki (1987) and Kotwicki and Isdale (1991) correlate Lake Eyre inflows to proxy coral records and large-scale climatic patterns, whereas Kotwicki and Kundzewicz (1995) discuss uncertainty aspects of Lake Eyre inflow estimates.
The Lake Eyre basin (Figure 1) spreads over 1.14 million km² of mostly arid central Australia. The Lake Eyre drainage system is quite well developed and of persistent nature, due to favourable structural conditions. Many of these drainage patterns are disconnected relics from linked river systems which developed under past pluvial regimes and which became disorganised under arid conditions. Evidence of this can be seen throughout the whole basin, perhaps most prominently in its western part, in the impressive channels of the Macumba or the Finke (sometimes claimed to be the oldest river on Earth).
Almost half of the basin area receives less than 200 mm of rainfall per year. Muloorina, the rainfall station nearest to Lake Eyre, has a mean annual rainfall of 146 mm, but a median of only 118 mm for 62 years of record. The higher rainfalls with annual means of the order of 400 mm per year occur in the northern and eastern margins of the basin, influenced by the southern edges of the summer monsoon.
The annual potential evaporation as estimated by the few Class A evaporimeters in the basin ranges from 2400 to 3600 mm, with the value of pan coefficient for the Lake Eyre possibly in the 0.5 - 0.6 range. The annual evaporation rate for the filled Lake Eyre is estimated to range from 1800 to 2000 mm.
Mean annual runoff of the Lake Eyre basin, of the order of 3.5 mm depth (i.e. 4 km³ volume) is the lowest of any major drainage basin in the world. Its specific yield of 10 m³km-2day-1 is 10 times less than the Nile and 200 times less then the Amazon per unit area of the catchment. Sediment transport is correspondingly low.
In principle, Lake Eyre may be filled in three ways: by the Cooper Creek, by the Diamantina River or by its western, desert tributaries. One of these three sub-catchments may cause a minor filling (10 km³); however, a combination of two (Cooper and Diamantina in 1950) or all three of them (in 1974) is required for a major filling (20-30 km³).
The Cooper Creek and the Diamantina River feature extreme variability in discharge and flow duration. Nanson et al.(1986), Rust and Nanson (1986) and Nanson and Rust (1988) consider hydrological aspects of eastern tributaries. Embarrassingly little is known about the hydrology of the western side of Lake Eyre.
Rating curves for the Cooper at Innamincka have been revised recently, following gauging of the 1990 flood. With current, improved knowledge, the 1974 flood was at this location some 75% greater than previously estimated. The peak flow at Innamincka was 6400 m³ s-1 (for comparison, four times the average discharge of the Yellow River) and total annual discharge was 15 km³. Flow figures for the Cooper Creek given in Kotwicki and Isdale (1991) should be, therefore, corrected. This applies also to the log-Pearson-III flood frequency curve given by Kotwicki (1986) for the Cooper. Correct values for Innamincka are now 3000 m³ s-1 for a 1:10 year flood and 10000 m³ s-1 for a 1:50 year flood.
Average daily flow of the Cooper at Innamincka for the period 1973-1993 is 63 m³ s-1, similar to that of the Thames in the UK. Average daily flow of the Diamantina at Birdsville for the period 1967-1993 is 43 m³ s-1. Flow equal to or exceeding mean flow occurred for 9% of the time at the Cooper and 18% of the time for the Diamantina.
The dynamics of fillings seems to be changing in recent decades, or perhaps we are observing a segment of a long-term variability or random fluctuations. Whereas anecdotal evidence attributes most of the fillings to the Cooper, the majority of fillings in the 1970s were caused by the Diamantina, while the two last fillings were caused by western tributaries (The Neales River, the Warriner Creek, the Margaret River and a number of shorter tributaries). The fourth major potential contributory, the Finke River, seems to be non contributing at present.
Floods are not easy to detect. Landsat provides an image of the area every 16 days (if there is no cloud cover), so many important flash floods may be overlooked. Daily information on the state of the lake may be best provided by commercial pilots, and more detailed information may be obtained only by chartered flights, as roads are inaccessible in times of floods. There is practically no ground hydrometric instrumentation to alert the observer, monitor the rivers, or verify aerial reconnaissance.
Kotwicki (1986) and Kotwicki and Isdale (1991) tabulate both morphometric parameters of Lake Eyre North and South and flow characteristics of its main tributaries. The morphometric parameters given in this reference relate to -9.5 m Australian Height Datum (AHD), the level of a record 1974 flood, classed as an 1:100 year event. It should, therefore, be realised, that they relate to some specific peak water storage (lasting a week or two) in the lake's recent history. Most of the time the lake is empty, although quite often it holds some quantities of water (1-3 km³ ) and on average some 80% of its surface area is covered with shallow water (storage of some 8-10 km3) once in eight years. A major filling of Lake Eyre occurs on average once in twenty years. Morphometric parameters of Lake Eyre during such an event are shown in Table 1.
Table 1. Morphometric parameters of Lake Eyre at -11.0 m AHD (1:20 year flood)
Since rainfall records for the Australian interior are short, sketchy and unreliable (Lake Eyre and its neighbouring area of some 100 000 km² has no permanent rainfall station) the question arises as to how rainfall and runoff records can be supplemented to provide better information on their variability, cycles and trends.
It can be hypothesised that the climate and weather on Earth are subject to a number of both external (Kotwicki, 1991) and internal forcings. If minute atmospheric disturbancies in one part of the world can lead to major events elsewhere (Mandelbrot, 1983), the whole system is likely to respond to forcings of much higher magnitude, such as that of the El Niño - Southern Oscillation (ENSO).
Contemporary Climatic Patterns
Relationships between Lake Eyre floodings, the hydrological regime in the Lake Eyre Basin and climatic patterns over the period of historical instrumental record have been the focus of a number of studies (Allan, 1985, 1989a,b, 1990, 1993; Allan et al., 1986; Isdale and Kotwicki, 1987; Kotwicki, 1986; Kotwicki and Isdale, 1991). It is apparent that major flooding episodes in the Lake Eyre Basin are most often associated with La Niña phases of the El Niño - Southern Oscillation (ENSO) phenomenon. During such times, the continent experiences enhanced austral summer monsoon activity and incursions of the monsoonal system into central Australia (Allan, 1988, 1991). Other synoptic scale features, such as rain depressions, cut-off low pressure systems and various extensions of tropical moisture into the continent (often associated with cloud band activity), provide the bulk of the remaining precipitation sources for flooding events. Interestingly, the frequency of a number of these shorter timescale events is directly related to the broad scale conditions that prevail during La Niña phases (Evans and Allan, 1992).
ENSO phases (El Niño and La Niña) occur irregularly, and represent the tendency for the phenomenon to vacillate between two extremes that occur as a consequence of ocean-atmosphere interactions centred in the Indo-Pacific basin. This can best be seen in the pattern of significant mean sea level pressure (MSLP) correlation regions indicative of the atmospheric Southern Oscillation component of ENSO in Figure 2. This diagram shows the mean situation, with broad regions of positive and negative correlations between global and Jakarta (Indonesia) MSLP forming a dipole structure which vacillates between a configuration of higher MSLP in the Indo-Australasian and lower MSLP in the southeastern Pacific regions (El Niño phase), and the reverse MSLP pattern (La Niña phase). Linked to such atmospheric redistributions are interactions involving the oceans. The historical mean oceanic sea surface temperature (SST) response at the mature stage of both of the ENSO phases is shown in Figure 3. As can be seen, this involves major SST extremes in the central to eastern equatorial Pacific and a weaker but similar response across the tropical to subtropical Indian Ocean. During El Niño (La Niña) phases, SSTs are warmer (cooler) in the above regions.
The extent of El Niño and La Niña influences can be gauged by the near-global distribution of climatological, hydrological and environmental impacts resulting from these phases (Figure 4). However, it should be noted that such diagrams do not attempt to capture the wide range of biological and ecological impacts that occur during these phases in marine and terrestrial environments (Allan, 1988, 1991). Nevertheless, Figure 4 clearly shows not only the extensive modulation of broad global precipitation patterns but also major responses in prominent river systems, such as the Nile and Amazon. In the context of this paper, floodings of Lake Eyre appear as one of the main lake system responses to the La Niña phase of ENSO. Overall, this figure highlights the fact that the strongest and most coherent impacts associated with ENSO phases occur over the Indo-Pacific Basin and adjacent continents. The Lake Eyre Basin is well located with regard to ENSO influence.
Published research has also begun to evaluate longer decadal to multidecadal variations in Lake Eyre flooding and filling events and links to similar fluctuations in ENSO over the historical record (Allan, 1989b, 1991). Such patterns suggest that ENSO and lower frequency components the climate system are interrelated, and that environmental conditions in regions influenced by ENSO are thus likely to be modulated on interannual to multidecadal time frames. If these links can be resolved further in the historical record, then proxy data series in the Lake Eyre Basin may provide an important source of information that can be extended to encompass palaeo-ENSO and climate patterns. Some of the most recent developments in palaeo-ENSO fluctuations are discussed later in this paper.
The main concerns about lower frequency variations in the ENSO phenomenon centre on the implications such fluctuations have for the longer term stability of the impacts that are associated with it, and the viability of current statistically-derived forecasting schemes. As a widely used measure of ENSO is the Southern Oscillation Index (SOI) (based on the normalised MSLP difference between Tahiti and Darwin), this is a logical starting point to investigate the long-term character of the phenomenon. Figure 5 shows a running/sliding correlation map of Tahiti against Darwin MSLP since 1876. This picture is built up by mapping correlations between monthly MSLP at the two locations during 21 year periods or windows that are shifted every five years through the data series from the beginning of the record. In the most recent periods or epochs, strong negative correlations in the austral winter to summer seasons indicate the widely known persistence of ENSO influence at this time. Such relationships break down during the austral autumn, and thus limit the extent of seasonal forecasting. However, for the remaining two thirds of the record back to the 1870s, the winter to summer correlation pattern is insignificant or absent. One explanation is that the Tahiti and/or Darwin MSLP series in the earlier parts of the record are suspect. Another would suggest that the fluid nature of ENSO makes a two station measure highly suspect as a stable measure of the phenomenon over time.
Although no problems of the first type posed above have been found in either MSLP series, it is important to address other concerns. Recent studies have focused on resolving the near-global picture of ENSO through the use of historical MSLP and SST fields, and the perspective given by current understanding of the main physical dynamics underlying the phenomenon. This is best illustrated by the studies of Allan (1993, 1994), Allan and Lindesay (1995) and Allan and D'Arrigo (1995). The first two papers examine the patterns of historical MSLP and SST over the globe using filtered fields of these parameters. Filtering was applied to the raw MSLP and SST data as a result of recent research which has emphasised the dominance of two signals in oceanic and atmospheric variables defining ENSO. These signals have been found to occur in bands around 18 to 35 and 32 to 88 months, and are defined as representing quasi-biennial (QB) and lower frequency (LF) components of ENSO respectively. An example of the nature of such signals can be seen in Figure 6, where the SOI has been filtered in these two time frames in a series from 1876 to near present. This diagram indicates that much of the ENSO signal is the result of the superposition of QB and LF elements, with the intensity and duration of both ENSO phases being a direct consequence of such interactions. In fact, Allan and D'Arrigo (1995) examine the nature of both 'persistent' and 'double' El Niño and La Niña events using QB and LF constituents of historical SOI, SST and rainfall records from the Pacific region.
Importantly, Figure 6 also carries information about the longer decadal to multidecadal nature of ENSO. This is most obvious when examining the LF curve and contrasting the 1920-1930s period with both earlier and more recent epochs. During the 1920-1930s, the LF component of ENSO is particularly weak and coincides with both a period of few distinct ENSO phases and broad global precipitation anomalies. The most documented environmental impacts during this time are droughts in the Great Plains of the USA, Australia and southern Africa and reduced boreal summer monsoon conditions over India. It is also a period during which no major floods occurred in Lake Eyre (Allan, 1985, Kotwicki, 1986). However, the recent and earliest parts of the LF series show very active low frequency patterns, and correspond to periods of more vigorous ENSO fluctuations. Such changes have been examined in a wider spatial and temporal framework by Allan (1993, 1994) and Allan and Lindesay (1995). The results of these studies are highlighted in Figure 7, where the patterns of correlation between global MSLP and the MSLP at Darwin are shown for the 1886-1907, 1921-1942 and 1968-1989 epochs in the LF band. From these diagrams (noting that correlations greater than +0.75 and less than -0.75 are significant at the 95% level), it is clear that the 1921-1942 period experienced a major contraction in the two dipoles indicative of the Southern Oscillation component of ENSO. In contrast, both the earlier and later epochs show more extensive and robust correlation fields. This pattern is characterised by significant positive correlations extending into Africa and India from the positive centre over Australasia, and a more coherent centre in the southeastern Pacific. From such evidence, it is apparent that the MSLP dipoles, indicative of the 'centres of action' of ENSO, wax and wane over the historical period.
In summary, recent research on longer term decadal to multidecadal fluctuations in near-global ENSO patterns indicates that the phenomenon is characterised by interactions between its QB and LF components, and displays a tendency to vacillate between periods of stronger and weaker activity in the historical record. During more active epochs, both 'centres of action' in the Indo -Pacific region are robust and extensive, while these dipoles wane and become fragmented and less coherent during weaker periods. This behaviour has the effect of expanding or diminishing the regions influenced by the phenomenon. During the most notable weak epoch in the 1920-1930s, Lake Eyre did not experience major flooding events. In the historical and contemporary records before and after the above epoch, the ENSO phenomenon was most active and strong links to Lake Eyre floodings through enhanced Australasian summer monsoon conditions are reported in the literature (Allan, 1985, 1989a, 1990, Allan et al., 1986).
Palaeohydrology and recorded floods
The quest for an "inland sea" in Australia was a driving force for numerous expeditions penetrating central Australia in the XIX century. Whereas the proposition was basically correct, the timing was not. Laseron (1955) suggests that the existence of a large lake in central Australia is traceable to the middle Jurassic (150 million years before present (BP)) when freshwater Lake Walloon occupied the central eastern part of the continent: in the late Cretaceous (85-75 million years BP) it shrunk to a smaller, 5 x 105 km² Lake Winton, which in the early Miocene (21 million years BP) assumed a position closely resembling the recent location of Lake Eyre. Lake Dieri, a Pleistocene "greater Lake Eyre", underwent large-scale climatic variations, entering into the Recent epoch as a dry salt pan subject to occasional flooding. Investigation of both sediments and palaeoclimates may provide important clues on palaeohydrology of the lake. Recent years have brought more research and better understanding of palaeoenvironments of arid central Australia (Chivas and de Decker, 1991) and Lake Eyre in particular (this volume).
Human record is vague. The Lake Eyre area is a desolate, absolute desert and has been sparsely inhabited. Europeans have never lived within sight of the lake, and even so, due to very low bottom gradients and atmospheric deceptions, an explorer venturing to the shore would have difficulties in seeing water at a distance when the lake is only partially filled. Hence, since its discovery as a dry pan in 1840, until its first recorded filling in 1949 the lake was considered permanently dry, and eventual reports on the existence of water were dismissed as observational errors. On numerous occasions, such as in 1869, 1875, 1882-85, 1891, 1895, 1906, 1908, 1918, 1922 and 1936, large volumes of water were reported in the upper or central reaches of the main tributaries but in each case an absence of evidence was considered to be the evidence of absence of the filling. Although such an approach is not truly scientific, for a long time it formed an established body of knowledge.
After 1949, a sequence of wet and dry spells have been observed. Amid minor isolated floodings a major flood event was recorded, which began in 1973, reached its peak in 1974 and persisted until 1977. It is estimated that the peak water storage in the lake during this event reached 32.5 km³.
Two recent fillings of Lake Eyre in 1984 (Allan et al, 1986) and 1989 are unusual in that they were caused by very large flows in its western desert tributaries, mainly the Neales. These are desert rivers which have previously been shown to only carry flow of any significance less than once in a decade. On these two separate occasions, Lake Eyre was filled within a few days by inflows ranging up to 30 000 m³ s-1. This is figure based on estimates of lake volume, as measuring of flows in this area was never attempted or even contemplated. Difficulties are many. In general, flooding events result in the whole countryside becoming a gigantic sheet flow, and gently defined channels are too numerous to gauge. Few opportunities exist to contemplate any rating of the flow: probably one of the best is the Neales River at the abandoned railway bridge.
Numerous salinas in the area, which receive runoff only from local catchments, fill less frequently. During the event in March 1989, Lake Torrens, the second largest salina in the world, was completely filled for the first time since 1878 (maximum storage 5 km³) by a runoff of 170 mm from its 30 000 km² catchment. The specific yield for this event was some 30 times greater than the annual average specific yield of the Amazon. Such large fillings are nearly always associated with cyclone activity.
Until recently, one used to consider the fillings of the Lake Eyre as rare and independent events. Now they are increasingly being looked at as a potentially predictable manifestation of global circulation patterns (e.g. El Niño - Southern Oscillations phenomena).
Isdale and Kotwicki (1987) and Kotwicki and Isdale (1991) used coral proxy data to further reconstruct the inflows to Lake Eyre. This is possible as the coral data reflects in some way the flows of the Burdekin River, draining a catchment of around 130 000 km², directly adjoining the much larger Lake Eyre system. During high flow periods the discharge of the Burdekin River moves northwards from the river outlet due to a longshore drift, and eventually reaches the shelf-edge reefs 250 km north of the mouth. The land-derived organic compounds, like humic and fulvic acids, are transported by the river and introduced to the marine system. These compounds are taken up by corals and accommodated in their growing skeleton structures and can be detected as they fluoresce under ultraviolet light. This fluorescence intensity provides a proxy measure of adjacent river discharge in the region of high flows. Although the proxy data of the River Burdekin is shown to be statistically linked to inflows to Lake Eyre, there are significant elements of uncertainty involved.
Aspects of the palaeoenvironmental and palaeoclimatic regimes likely to have influenced the Lake Eyre region and its drainage basin are addressed in other papers in this issue. Research in this area has expanded in recent years, but is still hampered by both the temporal resolution of proxy sequences and the sparsity of datable material (Croke, 1993; Gillespie et al., 1991; Magee, 1993; Nanson et al., 1992, 1993; Wasson et al., 1991).
Several papers have attempted to use contemporary and historical relationships between Lake Eyre hydrological responses and the ENSO phenomenon as a guide to reconstruction of prehistorical and palaeoclimatic histories of ENSO (Allan, 1985; Isdale and Kotwicki, 1987; Kotwicki and Isdale, 1991). Such long-term chronologies appear to be practical only if high resolution proxy data records can be found in the basin, or regional responses to lake floodings can be determined from such proxy series in other parts of Australasia. However, this type of approach suffers from the fact that only one response in one region of the globe is used. A more comprehensive reconstruction would involve using various proxy evidence from a number of regions affected by ENSO phases in the contemporary and historical records. Such a near-global reconstruction is becoming a viable proposition as more high-resolution proxy data become available in regions influenced by the phenomenon. This type of approach has been used by Allan and D'Arrigo (1995) in a recent examination of 'persistent' ENSO phases both since and prior to instrumental records. Their particular study focused on specific tree-ring and coral core series at several locations in regions influenced by contemporary and historical ENSO episodes across the Indo-Pacific basin.
Taking these considerations a step further, Wasson and Donnelly (1994, per. comm.) have begun to investigate high-resolution proxy data from the near-global distribution of regions impacted by ENSO at various time slices back through the Holocene. The first attempt at this for the El Niño phase is shown in Figure 8. This figure shows the broad precipitation regimes (wet or dry) in a number of the regions influenced by ENSO at time slices centred on 2 000, 6 000 and 9 000 BP. In general, ENSO-like rainfall anomalies of the correct sign are found in all of the six areas sampled at 2 000 BP. However at both 6 000 and 9 000 BP, the pattern is fragmented and no clear response to ENSO is evident. If this evaluation is representative, it adds further weight to contentions that ENSO, as we understand it, evolved in the mid to late Holocene (Markgraf et al., 1992; Martin et al., 1993; Nicholls, 1989; Ortlieb and Machare, 1993).
Time series of recorded fillings of the Lake Eyre commences with the 1949-1950 flood. This short record is also subject to significant uncertainties. The existing instrumentation and observation network are not adequate. Even now, the inflows to the Lake Eyre are not measured directly in the lower course of either of its tributaries. The existing gauges measure runoff from as little as 40 per cent of the catchment area, and the information on fillings before 1949 is practically non-existent. Therefore, some means of extension of the observational records are desirable. These fall into three major groups: hydrological or meteorological modelling, geological evidence, and proxy-evidence.
Kotwicki (1986) used a rainfall-runoff model (Laurenson and Mein, 1983) for the determination of past inflows to the lake. Using forty years of available data for identification and validation of the model, he reconstructed a time series of inflows to the Lake Eyre for a century (1885-1984). However, the possibility of such a reconstruction is limited by the availability of rainfall data.
Modelling and testing of various climatological scenarios in the Lake Eyre basin encounters considerable difficulties. There are three main issues involved: modelling of atmospheric forcings, modelling of the hydrological response of the catchment and modelling of the behaviour of the lake under various inflow and evaporation scenarios.
Using clues from the modelling of the contemporary scenario (Kotwicki, 1986), a simplified new model was constructed to obtain a first approximation of the dynamics of the system. It was calibrated with 108 years of rainfall data and available evidence on fillings of Lake Eyre. Then, the model was run to estimate inflows to the lake under various hypothetical rainfall, evaporation and runoff conditions scenarios.
The model recognises three parts of the catchment: the eastern side with mean annual rainfalls of the order of 400 mm, the western part with mean annual rainfalls of 150 mm and, a part presently non-contributing (Simpson Desert, the Finke River), but which could contribute under more humid climatic conditions. Rainfall averages for the Bureau of Meteorology districts are used for 1913 - 1993 and estimates based on available rainfall stations for 1884 - 1912.
It is evident that such a model has certain limitations, the most important of them being the implied linearity which contradicts the observed non-linearity of nature. The other deficiency is the inability of the model to reflect differing rainfall distribution throughout the year. As the model has to test various assumption rather than offer the best calibration, these are considered of minor importance.
Rainfall volumes are calculated for contributing areas and converted to runoff figures by using a runoff coefficient (0.01...0.08) expressed as a linear function of total rainfall volume, which varies between 100 and 500 km³ year-1. The runoff figures are then routed through conceptual storages, separately for each area. The storages are calibrated using available water balances, following Kotwicki (1986); in average present climatic conditions, they remove 10.8 km³ of water for the eastern part and 4.5 km³ for the western part of the basin. Due to their physical properties - their depth is less then annual evaporation - the storages are assumed to empty every year. These storages (many of them large physical sinks - like the Channel Country or Goyder Lagoon) are further manipulated - to reflect changing climatic conditions - by a linear function of evaporation from a free water surface which is adopted as 1200 mm for the eastern part and 1800 mm for the western part of the basin.
The modelling demonstrates that changes in evaporation alone would have an insignificant effect on inflows to Lake Eyre. For the eastern part, a change of evaporation of +10 and +20% would decrease runoff by 9 and 19% respectively: corresponding figures for the western part are 4 and 8%.
A change in catchment response to rainfall (runoff coefficient) has a more pronounced effect. Increase of catchment efficiency (same rainfall but lower transmission losses) by 10, 20 and 30% would bring 12, 45 and 80% more runoff to the lake.
However, as one would expect, it is rainfall variability which has the dominant influence on the lake hydrology. Increase in rainfall figures by 10, 20 and 40% would result in inflows to Lake Eyre rising to 6, 8.5 and 15 km³ respectively. Thus, it appears that the lake is just marginal, and that a relatively insignificant increase in annual rainfall (in absolute figures) would transform it into a permanent body of water.
Combined effects of changing rainfall, evaporation and runoff patterns were not modelled because the mechanics of feed-back mechanisms in the catchment are practically unknown. The next step was to simulate the behaviour of Lake Eyre under a variety of climatic conditions.
The mean annual inflow of 4 km³ (Kotwicki, 1986) would fill the lake to a localised maximum depth of 2.1 m in its southern bays. This means that with an evaporation rate of 2100 mm the lake would partially fill and evaporate completely in a year.
If the evaporation was 1200 mm y-1 - comparable for example to coastal lakes in South Australia or to the average evaporation over the World Ocean - the lake would begin to accumulate water. An equilibrium of 9 km³ would be reached in 30 years.
With still lower evaporation - say, 600 mm, comparable to Great Lakes, or lakes in the temperate regions of Europe and Asia - the lake would attain 48 km³ after 100 years.
Such considerations are highly theoretical, as in reality, lower evaporation over the lake would be obviously accompanied by lower evaporation in the whole basin. Less water would be lost in channel transport, and in such evaporative sink holes as the Channel Country or Goyder Lagoon. The overall runoff coefficient would be higher. The effect of such scenarios is illustrated in Table 2.
Table 2. Equilibrium volume and time to reach this volume (modelled), considering various inflow and evaporation scenarios.
* potential available storage to MSL is less than this figure
The water level in Lake Eyre can be approximated (Kotwicki, unpublished) as follows:
h = Hb + (V/1.05)0.53 
where: h is water level in m AHD, Hb is Lake Eyre North bed level (-15.2 m AHD), and V is water volume in km³.
It is known that significant fluctuations of evaporation from a lake can occur from year to year. One question which arises is whether some long term trend can be expected in the foreseeable future. The other question, is how the evaporation from the lake would be affected by changing climate.
The recent, revised climatic scenario (CCG, 1992) suggests that in Australia by the year 2030, the temperature will have increased by 0.5 to 2oC, summer rainfall will increase by 0 to 20%, rainfall intensity will generally increase, and overall wind speed will increase following a strengthening of monsoon westerlies in Northern Australia and the south-east trade winds in summer.
Such changes would have a significant effect on the flow regime of Lake Eyre tributaries. Nemec (1986) estimates that an increase of 25% in precipitation and a decrease of 1E C in temperature can increase the simulated runoff by 250%. This in turn intensifies the evaporation transmission losses, where according to Budyko (1980), a change of 1oC corresponds to a 4% change in evapotranspiration.
To evaluate the effect of changes in wind speed and temperature on evaporation, Kotwicki (1999) employed the bulk-transfer model used in the study of Lake Alexandrina (terminal lake of the River Murray in South Australia) with three years of collected climatic data (10 min averages of 5 s data of water and air temperature, humidity and wind speed) assuming changes delineated above. The results show that a 2EC temperature increase would enlarge evaporation by 20%. Wind speed increase of 10% would bring a similar 10% increase in evaporation.
It should be commented here that such a forecast seems to be quite accurate, as variations are tested on real data. However, it should be remembered that the nature of multiple feed-back mechanisms affecting evaporation is not taken into account here. For example, lower wind speed should decrease evaporation; but the incoming solar radiation, not energetically depleted for vaporisation in the first instance, would increase the temperature of air and water and consequently, the evaporation. Energetical considerations are, therefore, essential for a complete description of the system.
The process of inflows to Lake Eyre involves a very high degree of hydrological uncertainties, most of which are discussed in detail by Kotwicki and Kundzewicz (1995). They conclude that despite all the uncertainties in the cause-effect links, Isdale and Kotwicki (1987) had the right to draw the following corollaries from their studies:
i) The vast area of the Australian landmass comprising both the Lake Eyre and the Burdekin River basins endure a common broad climatic forcing, which is the ENSO or some function of it and
ii) The coral proxy record may be used to hindcast annual scale paleohydrological sequences (and ENSO periodicities) for several centuries before the modern instrument period.
It is evident that for reasons related to i), and further outlined in this paper, successful modelling and forecasting of inflows to Lake Eyre would be best achieved with a global scale general circulation model. This is because the mean state of the basin is not describable by a differential equation, since local stress is not described by local conditions. The atmosphere usually interacts with itself over a wider area than will permit for such a description.
Overall, it is apparent that further research into high-resolution proxy data series from many sites in the Lake Eyre basin is paramount. If such evidence is accumulated, it has the potential to provide vital information on the ENSO phenomenon, and in particular the La Niña phase. Given the major focus on El Niño episodes, an emphasis on the opposite phase is not only overdue but essential if a more comprehensive understanding of the total phenomenon during the Holocene is to be achieved.
List of figures (figures are not shown on this site):
Figure 1: Lake Eyre and its tributaries
Figure 2: Regions of significant positive (stippled) and negative (hatched) correlations between annual mean sea level pressure (MSLP) at Jakarta (Indonesia) and other stations over the globe (after Whetton et al, 1990).
Figure 3: Composite of sea surface temperature (SST) anomalies over the Indo-Pacific Basin at the peak of the austral summer (JFM) during strong El Niño (Top panel) and strong La Niña (Bottom panel) phases since the late 1800s. Anomalies are from the long-term mean from 1878-1992. Contours are shown every 0.25E C (Allan, unpublished).
Figure 4: Schematic diagram showing major near-global hydrological impacts associated with (a) El Niño and (b) La Niña phases derived from historical records (Allan, 1993, 1994).
Figure 5: Running/sliding correlations between MSLP at Tahiti and Darwin since 1876. Correlations were calculated for each month in 21 year sliding windows, moved a step every five years through the data series. Regions of significant negative correlations are stippled. Contours are shown for correlation coefficients every 0.2 (Allan, unpublished).
Figure 6: Filtered Southern Oscillation Index (SOI) series in the bands 18-35 (QB) and 32-88 (LF) months since 1876 (Allan, unpublished).
Figure 7: Epoch analysis of regions of significant positive (hatched) and negative (stippled) correlations between LF band annual mean sea level pressure (MSLP) at Darwin (Australia) and other stations over the globe. Epochs shown are 1886-1907; 1921-1942 and 1968-1989. Positive correlations greater than +0.75 and negative correlations less than -0.75 are statistically significant at the 95 % level. Contours are shown for correlation coefficients every 0.25 (Allan, unpublished).
Figure 8: Palaeoenvironmental evidence for ENSO-like responses in near-global precipitation at 2 000, 6 000 and 9 000 BP time slices. Conditions are shown over the historical to contemporary pattern of hydrological impacts of ENSO (Allan, 1994).
Allan, R.J., 1985. The Australasian summer monsoon, teleconnections, and flooding in the Lake Eyre Basin. South Aust. Geogr. Pap.,2, 47 pp.
Allan, R.J., 1988. El Niño Southern Oscillation influences in the Australasian region. Prog. Phys. Geogr.,12:4-40.
Allan, R.J., 1989a. Climatic conditions during the 1973-76 Lake Eyre filling. In The Great Filling of Lake Eyre 1974. Bonython, C.W. and Fraser, A.S. (eds), Royal Geographical Society of Australasia, South Australian Branch, 10-31.
Allan, R.J., 1989b. ENSO and climatic fluctuations in Australasia. In Donnelly, T.H. and Wasson, R.J. (eds), CLIMANZ III Proceedings of Symposium , University of Melbourne, 28-29 November 1987, 49-61.
Allan, R.J., 1990. Climate. In Natural history of the north east deserts. Tyler, M.J., Twidale, C.R., Davies, M. and Wells, C.B. (eds), Royal Society of South Australia Inc., Adelaide, 81-84.
Allan, R.J., 1991. Australasia. In ENSO teleconnections linking worldwide climate anomalies: Scientific basis and societal impacts. Glantz, M.H., Katz, R.W. and Nicholls, N. (eds), Cambridge University Press, Great Britain, 535 pp.
Allan, R.J., 1993. Historical fluctuations in ENSO and teleconnection structure since 1879: Near-global patterns. Quat. Aust.,11, 17-27.
Allan, R.J., 1994. Climate change and ENSO. Chapter 5 in Jakeman, A.J. and Pittock, A.B. (eds) Climate impact assessment methods for Asia and the Pacific, Proceedings of a Regional Symposium organised ny ANUTECH Pty Ltd on behalf of the Australian International Development Bureau (AIDAB), Canberra, 10-12 Mrach 1993, pp29-36.
Allan, R.J., Bye, J.A.T. and Hutton, P., 1986. The 1984 filling of Lake Eyre South. Trans. Roy. Soc. S. Aust., 110:81-87.
Allan, R.J. and Lindesay. J.A., 1995. Low frequency fluctuations in ENSO. Nature (In preparation)
Allan, R.J. and D'Arrigo, R.D., 1995. 'Persistent' ENSO Sequences: How unusual an event was it? The Holocene (submitted).
Bonython, C.W., Fraser, A.S. (Eds.), 1989. The Great Filling of Lake Eyre in 1974. Royal Geographical Society of Australasia, A.A. Branch, 119pp.
Chappell, J., 1991. Late Quaternary environmental changes in eastern and central Australia, and their climatic interpretation. Quaternary Science Reviews,10:377-390.
Chivas, A.R., de Decker. P., (Eds), 1991. Palaeoenvironments of salt lakes. Special issue of Palaeogeography, palaeoclimatology, palaeoecology, 84(1-4).
Climate Change Group, 1992. Climate change scenarios for the Australian region. CSIRO, Division of Atmospheric Research.
Croke, J.C., 1993. Episodes of deltaic and fluvial aggradation in the lower Neales River, west of Lake Eyre, Central Australia. Quat. Aust.,11, 28.
Eagleson, P.S. (1970). Dynamic hydrology. McGraw-Hill Book Company, 462pp.
Evans, J.L., Allan, R.J., 1992. El Nino/Southern Oscillation modifications to the structure of the monsoon and tropical cyclone activity in the Australasian region. Int.J.Climatol.,12 , 611-623.
Gillespie, R., Magee, J.W., Luly, J.G., Dlugokencky, E., Sparks, R.J. and Wallace, G., 1991. AMS radiocarbon dating in the study of arid environments: Examples from Lake Eyre, South Australia. Palaeogeogr., Palaeoclimatol., Palaeoecol., 84:333-338.
Isdale, P. and Kotwicki, V., 1987. Lake Eyre and the Great Barrier Reef: A paleohydrological ENSO connection. South Australia Geographical Journal,87:44-55.
Kotwicki, V., 1986. Floods of Lake Eyre. Engineering and Water Supply Department, Adelaide, 99 pp.
Kotwicki, V., 1987. On the future of rainfall-runoff modelling in arid areas: Lake Eyre case study. In Water for the Future: Hydrology in Perspective (Proc. Rome Symposium, April 1987). IAHS Publication No 164:341-351.
Kotwicki, V. (1991). Water in the Universe. Hydrol.Sci.J., 36:49-66.
Kotwicki, V., 1993. Evaporation from Lake Alexandrina. The Flinders University of South Australia, PhD thesis.
Kotwicki, V., Clark, R.D., 1991. Aspects of the water balance of three Australian terminal lakes. Proc. XX IUGG General Assembly, Vienna, 11-24 August 1991.
Kotwicki, V., Isdale, P., 1991. Hydrology of Lake Eyre, Australia: El Niño link. Palaeogeography, palaeoclimatology, palaeoecology, 84:87-98.
Kotwicki, V., Kundzewicz, Z., 1995. Hydrological uncertainty - Floods of Lake Eyre. In Z. Kundzewicz (Ed) New uncertainty concepts in hydrology. Oxford University Press (in press).
Lake Eyre Committee (1955). Lake Eyre, South Australia. The Great Flooding of 1949-1950. The Royal Geographical Society of Australasia, S.A. Branch, 75pp.
Laurenson, E.M. and Mein, R.G., 1983. RORB VERSION 3 - User Manual, Monash University, Department of Civil Engineering.
Laseron, C., 1955. Ancient Australia - The story of its past geography and life. Revised and updated by Brunnschweiler, R.O. (1984). Angus and Robertson.
Magee, J.W. 1993. Quaternary lacustrine events and chronology in the Lake Eyre Basin. Quat. Aust.,11:67-68.
Mandelbrot, B. 1983. The fractal geometry of nature. Freeman, New York.
Markgraf, V., Dodson, J.R., Kershaw, A.P., McGlone, M.S. and Nicholls, N., 1992. Evolution of late Pleistocene and Holocene climates in the circum-South Pacific land areas. Clim. Dyn.,6 :193-211.
Martin, L., Fournier, M., Mourguiart, P., Sifeddine, A., Turcq, B., Absy, M.L. and Flexor, J-M., 1993. Southern Oscillation signal in South American palaeoclimatic data of the last 7,000 years. Quat. Res., 39:338-346.
Nanson, G.C., Page, K.J., Callen, R.A. and Price, D.M., 1993. Evidence for changes in the climate and moisture regime of Australia over the past 60 ka. Quat. Aust.,11, 69.
Nanson, G.C., Price, D.M. and Short, S.A., 1992. Wetting and drying of Australia over the past 300 ka. Geology,20:791-794.
Nanson, G.C., Rust, B.R., Taylor, G., 1986. Coexistent mud braids and anastomosing channels in an arid-zone river: Cooper Creek, central Australia. Geology,14:175-178.
Nanson, G.C., Young, R.W., Price, D.M. 1988. Stratigraphy, sedimentology and late-Quaternary chronology of the Channel Country of western Queensland. In R.F. Warner (Ed) Fluvial geomorphology of Australia, Academic Press, pp151-175.
Nicholls, N. 1989. How old is ENSO? Climatic Change,14:111-115.
Ortlieb, L. and Machare, J. 1993. Former El Niño events: Records from western South America. Global and Planetary Change,7:181-202.
Pilgrim, D.H., Chapman, T.G. and Doran, D.G., 1988. Problems of rainfall runoff modelling in arid and semiarid regions. J. Hydrol. Sci.,33:379-400.
Rust, B.R., Nanson, G.C., 1986. Contemporary and palaeo channel patterns and the late Quaternary stratigraphy of Cooper Creek, Southwest Queensland, Australia. Earth Surface Processes and Landforms, 11:581-590.
Shuttleworth, W.J. (1983). Evaporation models in the global water budget. In A. Street-Perrot et al.(Eds). Variations in the global water budget. D.Riedel Publishing Company, 147-171.
Wasson, R.J. and Donnelly, T.H. 1991. Palaeoclimatic reconstructions for the last 30,000 years in Australia - A contribution to prediction of future climate. CSIRO Division of Water Resources, Tech. Pap.,91/3, 48 pp.
Whetton, P.H., Adamson, D., Williams, M.A.J., 1990. Rainfall and river flow variability in Africa, Australia and East Asia linked to El Nino-Southern Oscillation events. In P. Bishop (Ed) Lessons for human survival: nature's records from the Quaternary. Geol.Soc.Aust.Symp.Proc.,1 , 71-82.