Armin Bunde: “The anthropogenic trend in the global temperature record for the last 100 years is between 0.2°C and 1.4°C”

"By using trend-eliminating methods we found that climate has a long memory which decays slowly with time"
Interviews
(02/08/2010)

Armin Bunde, Professor at the Institut für Theoretische Physik III of the University of Giessen, was one of the plenary speakers at the XXII Sitges Conference on Statistical Mechanics held last June, organized by professors Miquel Rubí, Giancarlo Franzese and David Reguera from the Department of Fundamental Physics at the UB's Faculty of Physics. His main research areas are theoretical solid state physics (materials science) and statistical physics, and their applications in material science, biology, climatology, hydrology, meteorology and medicine.

The general aim of the conference was to discuss recent advances in the investigation and implementation of energy conversion mechanisms in nanoscopic to macroscopic systems. Professor Bunde gave the talk "Long-term memory in climate: On the estimation of anthropogenic trends in local and global temperature records”, which focused on the estimation of anthropogenic trends in climate fluctuations, which are known to exhibit a large natural variability.
"By using trend-eliminating methods we found that climate has a long memory which decays slowly with time"
Interviews
02/08/2010

Armin Bunde, Professor at the Institut für Theoretische Physik III of the University of Giessen, was one of the plenary speakers at the XXII Sitges Conference on Statistical Mechanics held last June, organized by professors Miquel Rubí, Giancarlo Franzese and David Reguera from the Department of Fundamental Physics at the UB's Faculty of Physics. His main research areas are theoretical solid state physics (materials science) and statistical physics, and their applications in material science, biology, climatology, hydrology, meteorology and medicine.

The general aim of the conference was to discuss recent advances in the investigation and implementation of energy conversion mechanisms in nanoscopic to macroscopic systems. Professor Bunde gave the talk "Long-term memory in climate: On the estimation of anthropogenic trends in local and global temperature records”, which focused on the estimation of anthropogenic trends in climate fluctuations, which are known to exhibit a large natural variability.

Although there is a broad consensus about the anthropogenic effect on climate change, is natural climate variability also well known?

It is well known that climate exhibits a large variability, but the quantification of this variability is a difficult task since natural variability may be masked by a slow-varying anthropogenic trend. Progress has only been made in recent years with the incorporation of methods from statistical physics, which can be used to eliminate trends in climate data.
 
What are the principal features of natural climate variability? Are the models well established?
By using trend-eliminating methods we found that climate has a long memory which decays slowly (by a power law) with time. It is well known that a warm day is more likely to be followed by a warm day than by a cold day, and the same holds, due to the long-term memory, for months, years and decades. Mathematically, the strength of the persistence can be characterized by the exponent of the power law that describes how the memory decreases with time. The relevant features of records with this kind of long-term memory can be studied within a statistical model which enables the generation of a large volume of data with the desired strength of persistence.
 
When studying historical data, where are the figures taken from?
Temperature data are from documented measurements (e.g. the temperature record from Prague, which dates back to 1775) and from temperature reconstructions using sources such as tree ring data. Both types of data, which are usually monthly or annual records, are available on the Internet. By definition, reconstructed data are less reliable than observational data.
 
How can these data be used to determine the natural variability and the anthropogenic contribution?
To evaluate the anthropogenic trend in a record with a given persistency parameter, we generated a large number of synthetic records with the same memory and determined the temperature increase in each record. This way we could deduce the probability of a natural temperature increase exceeding a certain value. When this probability is equal to 5%, the value often defines the upper limit of the confidence interval, where an observed trend can still be considered of natural origin. Only when an observed trend is outside the confidence interval is it considered to be unnatural. According to this interpretation, the anthropogenic contribution to a certain temperature increase is at least the difference and at most the sum of the increase and the 5% threshold. Using this approach we have found, for example, that  the anthropogenic trend in the global temperature record for the last 100 years is between 0.2°C and 1.4°C. If one distinguishes between sea surface and land air temperatures, the anthropogenic trend is between 0.4°C and 1.4°C for the land data and between 0°C and 1.4°C for the sea surface data, which shows that the recent increase in sea surface temperature may still form part of the natural trend.