The Sun's influence on modern-day global warming may have been overestimated, a study suggests.
Scientists found unexpected patterns in solar output in the years 2004-2007, which challenge existing models.
However, they caution that three years of data are not enough to draw firm conclusions about long-term trends.
Writing in the journal Nature, they say it may become necessary to revise the way that solar influences are dealt with in computer models of the climate.
But, they add, the research does not challenge the role of humanity's production of greenhouse gases as the dominant long-term driver of modern-day climate change.
"What we can't really do at this stage is to extrapolate from this three-year period to any longer period - we can't even say that [what we've seen] has happened on previous solar cycles," said principal researcher Joanna Haigh from Imperial College London.
"If you could extrapolate... the climate models have been over-estimating the Sun's effect on temperature [rise]."
The Intergovernmental Panel on Climate Change (IPCC) concluded that humanity's emissions over the 20th Century were about 10 times more important as a driver of temperature rise than the slow upward trend in average solar output.
This new study does not change that basic picture, Professor Haigh said, despite the claims of some observers that solar factors have been underestimated as a cause of modern-day climate change.
"If the climate were affected in the long term, the Sun should have produced a notable cooling in the first half of the 20th Century, which we know it didn't," she said.
The Sun's output waxes and then wanes on an 11-year cycle.
This periodic 11-year fluctuation rides on top of a much longer-term trend - and for most of the 20th Century, that trend was upwards, leading to the net warming influence on the Earth's temperature that the IPCC documented.
One of the things that varies most during the solar cycle is the Sun's output of ultraviolet radiation.
It is hard to measure from the Earth's surface, as the atmosphere absorbs much of the UV energy - but satellites can do the job.
One of the more recent satellites studying the Sun is the Solar Radiation and Climate Experiment (Sorce), launched by Nasa in 2003, with data collected and collated at the University of Colorado in Boulder, US.
Instruments on board the satellite provide readings of how the Sun's total energy output is changing, and breaks that down into various components across the spectrum from infrared through visible light into the ultraviolet.
The first years of readings from Sorce, covering 2004-2007, co-incided with the waning phase of the last solar cycle.
Scientists expected to see a slight decline across the spectrum. Instead, they saw a distinct fall in UV output, but an increase in emissions at visible wavelengths.
The UV fall - about six times as big as anticipated - was consistent with changes in ozone concentrations observed with other satellites. Ultraviolet radiation produces ozone in the upper atmosphere.
While these ozone concentrations can affect weather and longer-term conditions at the Earth's surface, so can the unexpected increase in energy at visible wavelengths, which penetrates down through the atmosphere.
Putting these various factors together, Professor Haigh's team calculates that over this three-year period, solar influences produced a net warming - not the net cooling that previous observations and theory predicted.
The observations, and the analysis, appear to have raised more questions than they have answered.
Is something awry with the satellite readings? That is unlikely, given that the UV changes were seen with two of its instruments and that they are consistent with ozone measurements; but it cannot be ruled out.
The trade-off between UV and visible radiation has not been seen by previous satellites. Is that because Sorce is better, or because what it has turned up is specific to that one solar cycle?
It is possible, contended Mike Lockwood of Reading University, that there was something special about the last solar cycle - that it could mark the end of an extended phase of relatively high output, and the transition into a less active phase.
"If you look back... 9-10,000 years, you find oscillations of the Sun between 'grand maxima' and 'grand minima'," he said.
"It's now emerging that the 'space age' has been a grand maximum; so my view is that the Sun is due to fall out of this and into a 'grand minimum', so I would not be surprised if in 50 years' time we find ourselves in conditions like the 'Maunder Minimum' [of the late 17th and early 18th Centuries] associated with the 'Little Ice Age'."
Professor Lockwood was not involved in the Nature paper; but his research has shown that even though short-term changes in solar output may not affect the global big picture, they can have a powerful impact on local weather patterns, particularly over Europe and Eurasia.
"So we might have the ultimate paradox that in a globally warming world we'd have cold winters in Europe. But it would be an awful lot warmer in Greenland," he said.
Professor John Shepherd, who studies climatic change at the National Oceanography Centre in Southampton, added:
"The observations do show that solar radiation does some peculiar and interesting things, which will hopefully be revealed through future research.
"As with all other known solar effects since measurements began, these effects are subtle and tiny - certainly nowhere near enough to explain any of the climate changes that we observe."
The Sorce satellite, meanwhile, continues to assimilate readings, and these may in time shed light on whether these three years of observation have thrown up a facet of how the Sun behaves generally and so whether the models need to be re-written with a lower value for solar influences - or whether there was indeed something unusual about the solar cycle we have just witnessed.