VISITING the giant kauri trees of Northland, on New Zealand's north-west coast, is like stepping back in time. Ancient conifer pines that over centuries have escaped damage from fire and forestry, the surviving kauris are up to 50 metres tall and five metres across - as wide as a 12-seater mini-bus is long.
Scientists estimate some have survived for two millennia and consider them the southern hemisphere's answer to California's redwoods, the world's biggest trees. Like the redwoods, age has transformed the kauris into a time capsule. Their tree rings - inner markings that reveal growth patterns through centuries - carry precise insights into changes in the world's climate conditions stretching back long before the advent of modern scientific measurement.
Since the 1970s, scientists have been extracting this knowledge using increment borers - giant corkscrews that collect long slivers of timber thin enough to slide inside a drinking straw. A decent sample will include hundreds of tree rings spanning centuries. When scientists get really lucky, they find a preserved dead tree that collapsed into a swamp and fossilised, allowing them to stitch together a picture dating back millennia.
Advertisement: Story continues below
On each data collection trip at least three samples of tree ring information are removed from each pine examined and a minimum 10 trees sampled across a tract of forest. The goal is to get a clear climate signal unaffected by natural variation between individual trees.
''Tree research is all about replication. You are looking at the growth patterns and trying to figure out whether the tree is responding to temperature or rainfall,'' says Dr Joelle Gergis, a palaeoclimatologist at the University of Melbourne's School of Earth Sciences, who spent two years collecting and analysing kauri data. Early last decade, data collected at Northland led to a peer-approved study that named the years over the past four centuries in which the climate had been most influenced by strong El Nino events over the Pacific.
This week the north New Zealand tree ring data was used to help paint a much bigger picture: what scientists say is the most complete climate record available for Australasia covering the past millennium.
The first study of its kind, it drew on 27 biological and geological records, including 12 tree ring data sets, mainly from New Zealand, 13 coral sets from Indonesia, the south-west Pacific and Western Australia and two ice cores from Antarctica. Scientists used the data to produce a long-term temperature record, and found the past five decades were very likely to have been the warmest since 1000AD.
Published in the Journal of Climate and drawing on decades of work by 30 scientists, it will help form the basis for the Australasian palaeoclimate section of the next report by the Intergovernmental Panel on Climate Change, due in 2013-14.
According to Gergis, the study's lead author, rigorous analysis had confirmed the reconstruction was robust. The numbers were crunched in 3000 different ways.
"The take-home message here is that 95 per cent of our 3000 reconstructions show that there are no 50-year periods in the past 1000 years that match or exceed the post-1950 warming," she says.
Co-author and University of Melbourne climate science professor David Karoly says the study for the first time establishes that claims there was a substantial mediaeval warm period hotter than today had no basis in Australasia. The study uses climate proxies - surrogates for the record of observed temperatures that date back to only the early 20th century.
Initially, the data from tree rings and other sites was tested for its ability to reconstruct temperatures between 1921 and 1990. The palaeoclimate records from 50 sites were compared to the actual temperature record for these years. The palaeoclimatic data that did not display a statistically significant temperature signal, but was found to have been more strongly influenced by other climate factors such as rainfall, was excluded. But the data from the 27 sites that remained collectively matched the actual temperature increase with a high correlation coefficient of 0.83, and were considered suitable for use as a proxy for the real thing to reconstruct temperatures over previous centuries.
The results matched what was known about certain historical periods. It was found early European settlers would have suffered through the coldest period of the past millennium in the 1830s and 1840s - the peak of what is known as the global little ice age. In pre-industrial times, the warmest lengthy stretch was found to be between 1238 and 1267, which the study estimates was 0.09 degrees cooler than the mid-to-late 20th century average.
But the warmest decades were found to be the last three examined: the 1970s, 1980s and 1990s. Instrumental temperature records show the first decade of the 21st century was hotter again.
The study is considered a step towards filling a gap in the long-term temperature record. To date, data from the southern hemisphere has been sparse compared with the north, in part because ocean cover is greater and also because scientific measurements on what land there is began later.
In the northern hemisphere, an earlier study had already found the second half of the 20th century was very likely to be the warmest in the past 500 years, and likely to be the warmest for the past 1300 years.
The historic reconstructions are being used as part of an international collaboration known as Past Global Changes, or PAGES. Its backers aim to improve the accuracy of future climate projections by reconstructing the past 2000 years of climate across the world. Those projections come from computer model simulations. These were also used in the PAGES work.
Dr Steven Phipps, a research fellow at the University of New South Wales Climate Change Research Centre, used three climate models to simulate the temperature patterns of the past millennium, at different times factoring in natural drivers of temperature such as volcanic eruptions and the sun and the human influence of greenhouse gas emissions.
Examined side by side, neither the average of the three climate model simulations nor the reconstructed temperature record shows any long-term trend before about 1850. In both, regional temperatures fall in line with known events, such as the 1815 eruption of Mount Tambora in Indonesia. Otherwise they bounce up and down independently. "What this demonstrates is that these variations are likely to be due to natural climate variability, which is a random process," Phipps says.
After 1850, both the model simulation and reconstruction show an unprecedented leap in warming. Phipps carried out three simulations in a bid to explain the warming, variously excluding known natural and human influences. He found the warming could only be produced through the inclusion of greenhouse gas emissions - clear evidence, he says, of human impact on Australasian temperature.
The veracity of the proxy methods used is not universally accepted. Professor Graham Farquhar, a biophysicist at the ANU's Research School of Biology, says the use of surrogates is problematic. For example, he says trees are likely to have grown faster between 1921 and 1990 due to the increased atmospheric concentrations of carbon dioxide, not just the rise in temperature. ''It is obviously very useful to have such data, but I can't see it as being definitive,'' he says.
The authors dismiss this concern. Karoly says that nothing is certain in science, but the results draw from a range of sites and using state-of-the-art statistical methods can be accepted with high confidence: ''It is reinforcing that barrage of scientific information that confirms that the climate is warming and increasing greenhouse gases are the major cause.''