"The overall picture of positive and negative precipitation trends was known before but I quantify for the first time how strong these trends are compared to internal variability on regional scales and when they will emerge, or 'stick out', from year-to-year noise," Douglas Maraun of the Helmholtz Centre for Ocean Research Kiel (GEOMAR) told environmentalresearchweb. "I identify regions where adaptation should be done early, and other regions where one does not really have a 'climate-change problem' but rather an 'internal-variability problem', i.e. one has to adapt to the variations one already has to cope with. These regions differ from season to season, and whether one considers mean rainfall or heavy rainfall."

Northern Europe in winter will see higher mean rainfall and more heavy rainfall within the next few decades, Maraun found. Trends will emerge on a similar timescale for mean rainfall in summer in south-western and south-eastern Europe. In other regions of the continent, however, the climate change trend for rainfall may not emerge until the end of the century or later.

"Many end users of climate-change information, e.g. agricultural modellers, still rely on one model or maybe a few," said Maraun. "Recommendations, however, suggest considering model ensembles. As these are often quite complex to handle (how do you present the output of 10 models to a decision-maker?), they are often reduced to the multi-model mean. However both approaches – considering one model or a multi-model mean – do not acknowledge and recognize the implications of uncertainties, in particular of internal climate variability."

According to Maraun, this internal climate variability means trends already observed may vanish or reverse in the long term because they were an expression of internal climate variability rather than caused by greenhouse-gas forcing. Similarly, trends expected as a result of climate change may vanish for years or decades, or might reverse, because they are overlain by very strong internal climate variability.

"My study, along with a few others, is a first step to quantifying the signal-to-noise ratio, i.e. the ratio between the actual trend we expect and the strength of the internal variations," said Maraun. "Quantifying this in terms of a time of emergence is a way to illustrate very intuitively when a trend we expect might really be relevant. In other words: how much time do we have to act in a particular place?"

Decision-makers planning climate adaptation are considering the expected climate change but not necessarily what they will really face, reckons Maraun, as the climate-change trend may take a long time to emerge from the background of internal variability.

"The expectation might be a good measure but depending on the strength of the internal variability, it might not be a good measure for quite a long time," he said. "For example, it could be that the trend will stick out only in 40 years' time. Then it might be a bad choice to start planting red-wine grapes that need a drier climate now, because it might stay as wet as it is or it might even get wetter for some time. It might be wiser to wait for several decades."

Examining rainfall over larger areas generally made trends emerge earlier. According to Maraun, if early adaptation is required, for example because grapes need to be replaced in 20 years and would last another 50 years, or because infrastructure needs replacing now but would be too expensive to replace again in 50 years' time, then the adaptation should be implemented on a larger scale.

"The internal variability averages out over this larger area, and for the total area the climate one expects is closer to what one sees," he said. "For a particular location, the experienced trend might still be strongly different from the expected trend, and thus an adaptation measure might be unsuitable. But on average the measure should be right for the larger area. Thus, if costs are shared over a large area the risk for, say, a single winegrower or community would be reduced."

To achieve his result, Maraun used simulations from 1971 to 2100 of 13 combinations of globally coupled atmosphere ocean general-circulation models and regional climate models from the ENSEMBLES project. The simulations were forced according to the A1B emissions scenario. Maraun estimated the time of emergence of the trend for seasonal total precipitation and seasonal maxima of daily precipitation. He defined the time of emergence as the year when the trend in multi-model mean exceeded a certain fraction, for example 20%, of the inter-annual variability in the year 2000.

For temperature changes, the same analysis gave a result that was obvious and boring, said Maraun. "Basically, all over Europe temperature changes will be relevant very rapidly and we will experience a considerable warming within the next 30 years."

Maraun is now focusing on understanding the causes of longer-term internal climate variability, such as the influence of the ocean on regional climate fluctuations.

Maraun reported his work in Environmental Research Letters (ERL) .

Related links

Related stories