FEATURE ARTICLE: HOW DO WE KNOW ABOUT CLIMATE IN THE PERIOD BEFORE INSTRUMENTS?

The first instrumental records of climate in the world date from the 17th century, while in Australia the first systematic observations took place in the mid-19th century. To draw any conclusions on climate from earlier times requires the use of indirect evidence, known to paleoclimatologists as ‘proxy’ records.

These records often require considerable interpretation, and often calibration against instrumental data. As there are uncertainties in the calibration, and also often in the dating, this often leads to substantial uncertainties in the records. It is also sometimes difficult to untangle the different climatic influences on a proxy record. For example, a change in the level of a lake may indicate a change in rainfall; a change in evaporation (which could, in turn, be related to temperature, wind or cloud cover); a change in inflows resulting from changes in a catchment area which may be a long distance away; a combination of these; or something else altogether. Nevertheless, many useful conclusions have been drawn about climate in pre-instrumental times from proxy records, both in Australia and elsewhere.

Over periods ranging from a few hundred to a few thousand years ago, there are some types of records which can provide information about conditions in each individual year. These include:

• tree rings
• other organisms with annual growth layers (e.g. corals)
• glaciers
• documentary records (e.g. of harvest dates/amounts, or freezing/thawing of rivers and lakes)

Australia is not particularly well-endowed with such records. There are no documentary records prior to European settlement, apart from a few fragments of oral history (e.g. information from Aboriginal elders which indicated that the 1849 snowfall in Melbourne was unprecedented since at least the late-18th century). The only trees suitable for multi-century tree ring analysis are in Tasmania (over most of Australia, tree growth occurs on an opportunistic basis as conditions allow, rather than on a predictable annual cycle as it does in climates with large seasonal temperature variations), and there have been no glaciers in Australia since the last ice age. As a result, detailed information about the climate of the last 1,000 to 2,000 years is much more sparse for Australia (and for the southern hemisphere generally) than it is for continental parts of the northern hemisphere. Australia does have some information drawn from sources such as boreholes and lake deposits, which are useful for indicating changes on decadal to centennial timescales but not for indicating year-to-year variability, and corals in some areas provide useful information on near-shore conditions which in some cases can be used to draw inferences about climate on land (e.g. through the effect of runoff from coastal rivers).

Going back beyond these timescales, information necessarily becomes more general. The most detailed information available is through ice and sediment cores, where isotope ratios can be used to draw inferences about local (and, by extension, global) temperatures and atmospheric composition. Some Antarctic ice cores extend back 740,000 years, and some deep ocean sediment cores for several million years. Speleothems (cave mineral deposits) can also be used for this purpose.

Fossil evidence of vegetation, sediments, lake and river levels, glacier locations, or dune activity can provide good indications of the climate which prevailed over a particular period of time. This has proven especially useful in assessing Australia’s likely climate at the time of the last glacial maximum. It is also possible, once certain changes (such as orbital changes, global mean temperatures and atmospheric composition) are known with some degree of confidence, to use these as input into a climate model to draw inferences about more detailed changes that might have occurred, such as changes in the nature of the atmospheric circulation (including the summer monsoon and the mid-latitude westerlies). Climate models can also be used in very long-term simulations of the ‘present’ climate, which does not provide specific information about the past, but can be useful in placing present-day extremes in an historical context (e.g. in determining just how unusual the post-1975 dry period in south-western Australia is likely to have been).

Beyond a few million years ago, the fossil record - including isotope ratios contained within it - is the best evidence we have. At these timescales the concept of changes in the ‘Australian’ climate becomes relatively meaningless, as the shape, elevation and location of the continent were very different to what they are today, but it is nonetheless of great interest to know what changes occurred globally.


THE WORLD'S CLIMATE BEFORE THE RECENT ICE AGES

For much of the last 400 million years, the world has been substantially warmer than it is now. Geological evidence indicates that, for most of the period between 40 and 260 million years ago, the world was entirely or almost entirely ice-free, indicating a climate several degrees (at least) warmer than that which exists now. The ocean circulation would also have been very different to the present due to different continental configurations. At earlier times still, there have been numerous ice ages, and it is possible that the entire globe was ice-covered at times, particularly around 700-800 million years ago.

Conditions became cooler from about 40 million years ago, but there were still numerous warm periods. One specific period which has been the subject of some investigation has been the period from 3.0-3.3 million years ago. This period appears to have had global mean temperatures 2-3 degrees Celsius (°C) warmer than the present, with the greatest warming at high latitudes. During this period, sea levels are estimated to have been several metres higher than they are at present.


THE ICE AGES AND AUSTRALIA'S CLIMATE

A succession of ice ages occurred from about 430,000 years before the present. These have taken the form of extended cold periods, lasting in the vicinity of 100,000 years, with intervening warmer (interglacial) periods lasting 10,000 to 30,000 years. The most recent ice age commenced about 115,000 years ago, continuing for more than 100,000 years.

In general, mean global temperatures during the ice ages were approximately 4-7°C colder than they are at present, with the greatest cooling at high latitudes (graph S1.1), and the least over tropical oceans, with cooling of less than 2°C in places. During interglacial periods, temperatures were generally reasonably close to present levels.

The lowest temperatures of the last ice age were reached about 21,000 years ago, known as the Last Glacial Maximum (LGM). There has been a detailed analysis of Australia’s likely climate at the LGM. This indicates that temperature decreases over much of Australia were at least as large as those recorded over the globe as a whole, with numerous proxies indicating cooling of 6-10°C. The climate was also generally drier, with a weaker monsoon in the tropics, and penetrations of tropical moisture into the central continent largely absent (e.g. silt deposits in streambeds in the Flinders Ranges indicate that the occasional severe floods, associated with incursions of tropical moisture, which occur there now were essentially unknown at the LGM). The mid-latitude westerlies, another potential source of moisture for the Australian continent, also appear to have moved further south in Australian longitudes. Permanent snow cover and glaciation was confined to the highest areas (generally above 1,800 metres in the Snowy Mountains, and 1,000 metres in Tasmania), but large parts of southern Australia would have had at least an intermittent seasonal snow cover in winter.

With sea levels at the LGM up to 120 metres below those of the present-day, the shape of the continent was very different to what it is now; in particular, Bass Strait and Torres Strait did not exist, the Gulf of Carpentaria was a lake with no outlet to the open sea, and the Queensland coast was several hundred kilometres east of its current position over much of its length. This is likely to have had a major influence on the tracks of tropical cyclones, which require warm water for their development (the number and intensity of cyclones is likely to have been reduced in any case). It would also have had a strong influence on local climates in some locations, which are now moderated by the ocean but would have been much more continental at the LGM. As an example, Melbourne would have been several hundred kilometres inland; without the influence of the ocean, summer temperatures may not have been much cooler than they are now (it is even conceivable that they may have been warmer at the most exposed coastal locations, such as the north coast of Tasmania), but the decrease in winter temperatures is likely to have been much more than the regional mean.


CLIMATE FOLLOWING THE ICE AGES

The most recent ice age ended about 12,000 years ago, although there was a sharp, short-lived cold period around 8,200 years ago, associated with a sudden influx of fresh water into the oceans as large lakes dammed by retreating ice sheets in the northern hemisphere collapsed.

There are numerous indications, particularly from glaciers, that the climate in some parts of the world was somewhat warmer at various times around 5,000 to 6,000 years ago than it is now. This warmth, however, appears to have been concentrated in mid-latitudes during summer, with little evidence of significant temperature changes in the tropics. Overall global temperatures are not believed to have been more than 0.4°C above current levels, and the limited available evidence also suggests little change in Australian temperatures. Studies of Australian vegetation from around 6,000 years ago indicate somewhat wetter conditions in south-eastern New South Wales on and east of the Snowy Mountains, drier conditions in south-western Victoria and south-eastern South Australia, and slightly drier conditions in south-western Western Australia - all of which would be consistent with a climate less strongly influenced by the mid-latitude westerlies and associated winter rainfall events than it is now. There is also some evidence that tropical winters may have been less dry than at present.

There are some indications from fossil coral records, as well as from lake deposits in South America, that the El Niño-Southern Oscillation (ENSO), which appears to have been present in some form for at least the last 130,000 years, was somewhat weaker in the early Holocene than it is now. South American studies indicate that the shift from relatively infrequent El Niño events to the present-day average of 2-3 events per decade took place over the period from 7,000 to 5,000 years ago.


THE LAST 1,000 TO 2,000 YEARS

As noted above, detailed information on Australian climate over the period from 1,000-2,000 years ago to the start of European settlement is extremely scanty, compared with what exists in many parts of the northern hemisphere. Nevertheless, there are some proxy records which give us an indication of what happened to the Australian climate over that period. The general picture which emerges is one of a climate which showed considerable variability on annual and decadal timescales, as it does now, but in most cases was not drastically different in a long-term sense to that which existed in the early instrumental period, prior to the start of substantial greenhouse warming.

The best single proxy temperature record from the Australian region is the record of tree rings from Huon pines on Mount Read in western Tasmania. This extends back to 1600 BC. It suggests that multi-decade means of temperature in the second half of the 20^th century were the highest of the last 3,600 years, but not by a large margin, with temperatures almost as high being sustained over a much longer period at various times between 900 and 1500 AD. (It should be noted that this data set ends in 1992 and instrumental data indicate further warming since.) The data also suggest that the early part of the 20^th century was rather cold by the standards of the last 3,600 years. There are only modest (typically less than 0.3°C) variations in multi-century means of temperature over the 3,600-year period, but an interesting feature is that the level of interdecadal variability of temperature between 1500 and 1900 was substantially lower than that prior to 1500, and particularly prior to 100 AD.

Data from boreholes, covering the last 500 years, indicate that the rate of warming in the Australian region from 1500 to the present is about half that observed in the northern hemisphere. Together with the Tasmanian data, this suggests that the cold period from 1500 to 1900 in the northern hemisphere, known as the ‘Little Ice Age’, has no real Australian counterpart.

Another record of interest has been derived from crater lakes in south-western Victoria. These lakes are very deep and have catchments of at most a few square kilometres (with the lake surface itself occupying most of the catchment), so the lake level is almost entirely determined by the rainfall-evaporation balance, with runoff and land use change having little or no influence. The lakes studied all show a marked decline in level from about 1840 after having been relatively stable for several centuries. This indicates a marked shift in the rainfall-evaporation balance in the decades around 1840, with a decrease in rainfall being the most likely factor, although changes in evaporation (which could arise from cloudiness, temperature, wind or a combination of these) may also have been involved.

A number of studies have examined the behaviour of El Niño and La Niña in the Pacific over the last 500 years, using proxies such as corals, tree rings (from New Zealand and western North America) and ice cores (from the Andes). While these do not draw directly on any information from the Australian continent, the importance of El Niño as an influence on the Australian climate makes this of great interest. In general, these studies show that the behaviour of El Niño in the period from 1500 to the late-1800s is similar to that which has occurred since, both in terms of the typical frequency of events (although there are weak indications that El Niño events may have become slightly more frequent and intense over the last 200 years) and in terms of the characteristic decade-to-decade variability in event frequency; periods such as the clusters of several El Niño events in rapid succession in the 1940s, or several La Niña events in the 1970s, have numerous counterparts in the pre-instrumental record. They also indicate that the relationships between El Niño and the broader south-west Pacific climate have remained reasonably robust over the last 500 years.


1840 TO 1900 - THE EARLY INSTRUMENTAL PERIOD

While there have been short-term meteorological observations in various parts of Australia since the earliest years of European settlement, the first long-term rainfall data set, in Adelaide, commenced in 1839. Until the late-1850s there were only a handful of stations, in the major cities. The observing network then grew through the 1860s in parts of New South Wales and South Australia, then through the 1870s and early-1880s in much of the rest of eastern Australia, and in some key Northern Territory locations. Apart from a few locations near the coast, there were few pre-1900 observations in Tasmania or Western Australia, nor in many of the more remote parts of central Australia. Temperature observations also occurred over this period through many of these regions, but as most of these were made using instrument shelters which are not directly comparable with current standards, they are of little use in assessing long-term trends.

Using these data, an assessment about rainfall in mainland eastern Australia can be made from the mid-1860s onwards. The 1870-95 period was a rather wet one in most regions, generally comparable with the 1950-80 period. There were a few significant drought years (notably 1877 and 1888, along with 1865 in South Australia - which was the drought which prompted Goyder’s seminal report), but wet conditions were generally predominant. The 1885-95 period was especially wet in Queensland and much of New South Wales, with some parts of inland Queensland recording ten-year means up to 40% above the long-term average - beyond anything experienced in the wettest parts of the 1950s or 1970s. The years 1890 and 1893 were the wettest on record at many long-term sites.

Rainfall anomalies during the 1885-95 period were less extreme, although still positive, in the southern states. In this region the late-1870s and early-1880s were rather dry, following a very wet period in the early-1870s (1870 was an especially wet year). The 1870-95 period was also generally wetter than long-term averages at the few tropical sites in Queensland and the Northern Territory from which data exist. In the coastal south-west of Western Australia, the 1877-1900 period (for which some data exist) was generally drier than the first half of the 20th century, but wetter than the post-1970 period.

1895 marked the start of the ‘Federation Drought’, a prolonged dry period which affected much of eastern Australia and continued until 1903. The most severe individual drought years during the period were 1895 and 1902, but, in a similar way to the post-2001 drought, it was the lack of sustained wet periods between the individual drought years which made it an especially extreme event.


1900 TO THE PRESENT - THE INSTRUMENTAL PERIOD

As is well known, Australia has seen substantial warming since 1910, and especially since 1950. The total amount of warming (about 0.7°C) is comparable to the warming of global mean temperatures over the same period, although the pattern of warming over time is somewhat different (graph S1.2); the 1940s peak and 1940-75 levelling off of global temperatures have no direct counterpart in the Australian data.

Changes in rainfall have been more complex (maps S1.3 and S1.4). There has been an unambiguous downward trend in rainfall since 1970 in the south-west of Western Australia, with a decline of 10-20% in many locations. In contrast, many parts of north-western Australia, extending into interior parts of Western Australia, have become much wetter (up to 50% in places) since 1960.

In eastern Australia, the general pattern has been one of a relatively dry first half of the 20th century, followed by a wet period from the late-1940s through to the early-1990s (the 1950s and 1970s were especially wet decades, particularly in New South Wales and Queensland). Since the early-1990s rainfall has dropped, in most places, to levels more characteristic of the first half of the century, resulting in trends which are strongly negative if taken from 1950 but weak if taken from 1900. Two notable exceptions are southern Victoria and south-eastern Queensland, both of which have seen rainfall averages since the late-1990s which are lower than anything experienced for a comparably long period at any time in the 20th century. In the Melbourne area the 1997-2006 mean rainfall was about 10% lower than the driest 10-year period recorded at any time prior to 1996.

It is very likely that changes in global temperature have been largely driven by human-induced changes in the atmosphere, especially increased concentrations of greenhouse gases such as carbon dioxide. It is more difficult to formally attribute climate changes to particular causes over an area the size of Australia than it is over the globe as a whole, but the changes observed in Australian temperatures, and in rainfall in southern (particularly south-western) Australia, are reasonably consistent with those which climate models indicate would have been expected as a result of the known changes in greenhouse gas concentrations. The multi-decadal variations in rainfall over eastern Australia, and the rapid post-1960 increase in rainfall over north-western Australia, are more difficult to explain, although recent research indicates that heightened aerosol levels from pollution over eastern and southern Asia may be a contributing factor to the latter.

Much less information is available on other climatic elements. However, recent research indicates that there has been no significant trend in pan evaporation over Australia since 1970, while there has been a modest increase in atmospheric moisture (humidity) since 1960. The number of tropical cyclones in the Australian region has declined slightly since the 1950s, particularly in the Queensland sector (largely as a consequence of changes in the frequency of El Niño and La Niña over that time), but there is insufficient information to draw firm conclusions on the frequency of intense tropical cyclones over that period.