* At least 95 pct likely man causing warming - draft study
* Harder than expected to predict near-term local impacts
* Temperatures risk breaching U.N. ceiling, despite slowdown
By Environment Correspondent Alister Doyle
OSLO, Aug 16 (Reuters) - Climate scientists are surer than ever that human activity is causing global warming, according to leaked drafts of a major U.N. report, but they are finding it harder than expected to predict the impact in specific regions in coming decades.
The uncertainty is frustrating for government planners: the report by the Intergovernmental Panel on Climate Change (IPCC) is the main guide for states weighing multi-billion-dollar shifts to renewable energy from fossil fuels, for coastal regions considering extra sea defences or crop breeders developing heat-resistant strains.
Drafts seen by Reuters of the study by the U.N. panel of experts, due to be published next month, say it is at least 95 percent likely that human activities - chiefly the burning of fossil fuels - are the main cause of warming since the 1950s.
That is up from at least 90 percent in the last report in 2007, 66 percent in 2001, and just over 50 in 1995, steadily squeezing out the arguments by a small minority of scientists that natural variations in the climate might be to blame.
That shifts the debate onto the extent of temperature rises and the likely impacts, from manageable to catastrophic. Governments have agreed to work out an international deal by the end of 2015 to rein in rising emissions.
"We have got quite a bit more certain that climate change ... is largely manmade," said Reto Knutti, a professor at the Swiss Federal Institute of Technology in Zurich. "We're less certain than many would hope about the local impacts."
And gauging how warming would affect nature, from crops to fish stocks, was also proving hard since it goes far beyond physics. "You can't write an equation for a tree," he said.
The IPCC report, the first of three to be released in 2013 and 2014, will face intense scrutiny, particularly after the panel admitted a mistake in the 2007 study which wrongly predicted that all Himalayan glaciers could melt by 2035. Experts say the error far overestimated the melt and might have been based on a misreading of 2350.
The new study will state with greater confidence than in 2007 that rising manmade greenhouse gas emissions have already meant more heatwaves. But it is likely to play down some tentative findings from 2007, such as that human activities have contributed to more droughts.
Almost 200 governments have agreed to try to limit global warming to below 2 degrees Celsius (3.6 Fahrenheit) above pre-industrial times, seen as a threshold for dangerous changes including more droughts, extinctions, floods and rising seas that could swamp coastal regions and entire island nations.
The report will flag a high risk that global temperatures will increase this century by more than that level, and will say that evidence of rising sea levels is now "unequivocal".
For all that, scientists say it is proving harder to pinpoint local impacts in coming decades in a way that would help planners.
Drew Shindell, a NASA climate scientist, said the relative lack of progress in regional predictions was the main disappointment of climate science since 2007.
"I talk to people in regional power planning. They ask: 'What's the temperature going to be in this region in the next 20-30 years, because that's where our power grid is?'" he said.
"We can't really tell. It's a shame," said Shindell. Like the other scientists interviewed, he was speaking about climate science in general since the last IPCC report, not about the details of the latest drafts.
The panel will try to explain why global temperatures, while still increasing, have risen more slowly since about 1998 even though greenhouse gas concentrations have hit repeated record highs in that time, led by industrial emissions by China and other emerging nations.
An IPCC draft says there is "medium confidence" that the slowing of the rise is "due in roughly equal measure" to natural variations in the weather and to other factors affecting energy reaching the Earth's surface.
Scientists believe causes could include: greater-than-expected quantities of ash from volcanoes, which dims sunlight; a decline in heat from the sun during a current 11-year solar cycle; more heat being absorbed by the deep oceans; or the possibility that the climate may be less sensitive than expected to a build-up of carbon dioxide.
"It might be down to minor contributions that all add up," said Gabriele Hegerl, a professor at Edinburgh University. Or maybe, scientists say, the latest decade is just a blip.
The main scenarios in the draft, using more complex computer models than in 2007 and taking account of more factors, show that temperatures could rise anywhere from a fraction of 1 degree Celsius (1.8 Fahrenheit) to almost 5C (9F) this century, a wider range at both ends than in 2007.
The low end, however, is because the IPCC has added what diplomats say is an improbable scenario for radical government action - not considered in 2007 - that would require cuts in global greenhouse gases to zero by about 2070.
Temperatures have already risen by 0.8C (1.4F) since the Industrial Revolution in the 19th century.
Experts say that the big advance in the report, due for a final edit by governments and scientists in Stockholm from Sept. 23-26, is simply greater confidence about the science of global warming, rather than revolutionary new findings.
"Overall our understanding has strengthened," said Michael Oppenheimer, a professor at Princeton University, pointing to areas including sea level rise.
An IPCC draft projects seas will rise by between 29 and 82 cm (11.4 to 32.3 inches) by the late 21st century - above the estimates of 18 to 59 cm in the last report, which did not fully account for changes in Antarctica and Greenland.
The report slightly tones down past tentative findings that more intense tropical cyclone are linked to human activities. Warmer air can contain more moisture, however, making downpours more likely in future.
"There is widespread agreement among hurricane scientists that rainfall associated with hurricanes will increase noticeably with global warming," said Kerry Emanuel, of the Massachusetts Institute of Technology.
"But measuring rainfall is very tricky," he said. (Editing by Mark Trevelyan)