ClimatePrediction.net

from Wikipedia, the free encyclopedia
ClimatePrediction.net
Area: meteorology
Target: Predicting climate change
Operator: Oxford University
Country: Great Britain
Platform: BOINC , own client
Website: http://www.climateprediction.net/
Project status
Status: active
Start: 09/12/2003
The End: still active

ClimatePrediction.net ( CPDN ) is a volunteer computing project at Oxford University , with which the climate for the next 50 to 100 years is to be calculated and thus statements about the extent of global warming can be made.

Climate models place high demands on the performance of the computer hardware used and have traditionally been reserved for the fastest supercomputers and clusters . The rapid increase in performance of commercially available PC hardware makes it possible that such models can now also be run on comparatively inexpensive conventional computers in a reasonable time. The ClimatePrediction.net client uses unused computing capacity of a computer to calculate a climate model.

CPDN BOINC client in graphical mode

Project goals

In addition to the direct goal of using a large number of simulations to predict how the world climate will most likely develop in the next 50 to 100 years, ClimatePrediction.net also wants to make a contribution to improving the climate models used. There are great uncertainties in the predictions of a climate model, which arise on the one hand from the fact that not all relationships and interactions in the climate are known, and on the other hand due to the values ​​of the parameters used. Many physical values ​​are known to have significant effects on the climate. However, which range of values ​​is possible and useful can only be found out through observation and a large number of tests. This is where ClimatePrediction.net comes in. The computers participating in the project calculate with slightly different parameterized climate models and therefore come to different results.

history

ClimatePrediction.net is based on a comment by Myles Allen entitled “Do-it-yourself climate prediction”, which appeared in 1999 in the journal Nature . He showed the possibility of running modern climate models such as HadCM3 on reasonably modern home computers in order to be able to narrow the parameter space. At that time, SETI @ home , one of the first projects for distributed computing, was on everyone's lips. Someone would be able to tell their grandchildren later that it was he who calculated the most accurate forecast of the global mean temperature for 2050 on a $ 1,650 personal computer.

By the year 2000 the project partners had found each other, including the universities of Reading and Oxford and the British Met Office . The project originally ran under the name Casino-21 and was later renamed ClimatePrediction.net. The public launch took place on September 12, 2003 with a client for Windows. On August 26, 2004 the project was converted to a new infrastructure based on the BOINC developed by the SETI @ home team . At the same time, clients for Linux and Mac OS were published for the first time.

Project progress

Several experiments are carried out within the ClimatePrediction.net project.

The first experiment served to limit the parameter space. For this purpose, the HadSM3 model from the British Hadley Center was used, which is also used for commercial weather forecasts in Great Britain (but then with a higher resolution). This model enables a detailed simulation of atmospheric processes, but only has a simplified ocean. Each participating computer works through three phases one after the other with this model: In the first phase, the model is calibrated . The main aim is to calculate a heat flow value with which the “dummy ocean” can be kept at a constant temperature. In addition, it should be determined whether the selected parameters allow a stable climate at all. If it turns out in this phase that the climate is becoming unstable (figuratively speaking, the earth either becomes an ice ball or is boiling), the simulation is terminated at this point. Otherwise, two further phases are followed in which the ocean can change its temperature, but the CO 2 content of the atmosphere is kept constant. In the second phase with a pre-industrial CO 2 content, the global mean temperature should ideally remain stable. In the third phase, double the CO 2 content is used. The climate should settle at a new stable level.

The second experiment started in 2005. A new climate model with a fully simulated ocean was used (initially HadCM3). Within this experiment, the climate was calculated for the years 1950 to 2000. During this time it is known from records how the climate developed. The combinations of starting conditions and parameters were sought for which the forecast was closest to the actual climate development. The models were classified into a ranking based on their results.

The third experiment, active since February 2006, is an actual forecast for the years 2000 to 2100. The results of the various model combinations are again weighted using the ranking. The result should be a relatively accurate prediction of what, according to the rules of probability, will happen to the climate in the next 100 years. In addition, special models are sent to a number of selected computers in between, for example with a simulation of the sulfur cycle.

Side projects

In addition to the main project, CPDN also operates two side projects that also use the BOINC infrastructure.

In August 2005, the Seasonal Attribution Project was started, the aim of which is to research the influence of man-made climate change on local storms . As an example of such a local extreme weather event, the extremely rainy autumn of 2000 over England and Wales , which led to flooding in large areas of Great Britain, is used.

The second side project was the BBC Climate Change Experiment , which was started in cooperation with the British broadcaster BBC and was supposed to simulate the entire world climate for the period from 1920 to 2080. The first results of the simulations were planned for May 2006 as BBC documentation for publication.

First results

This article or section needs to be revised. More details should be given on the discussion page under " Section" First Results " ". Please help to improve it , and then remove this flag.

On January 27, 2005, the CPDN team published initial results in the science magazine Nature. The results of 1148 “stable” simulations indicated global warming of at best 2 to a maximum of 11.5 ° C with a doubling of the CO 2 content in the atmosphere. This range is much larger than the climate sensitivity of 2 - 4.5 ° C, which was predicted by the models used in the fourth IPCC report . The analysis of the results indicates that the CPDN models, which calculated very high temperature changes, are not outliers that can be completely excluded, but are very unlikely. However, they show what is possible in principle. The fact that the climate reacts far more sensitively than previously calculated had been considered possible by various studies, but CPDN is the first project in which this possibility is actually shown in the results of the test ensemble.

Subsequent studies were able to specify that an increase in the global average temperature of more than 4.5 ° C occurs with a doubling of the CO 2 concentration with a probability of at most 5%.

Future developments

For the future it is planned to include other climate models not from the Hadley Center in the project, or to improve the existing models or operate them with a higher resolution. Only a small part of the result data currently on the clients is actually scientifically evaluated. For the future it is planned to use these valuable results to carry out further evaluations that are not directly related to the ClimatePrediction.net project.

Individual evidence

  1. ↑ Allen, Myles (1999): Do-it-yourself Climate Prediction , in: Nature, Vol. 401, October 14, p. 642 (PDF; 54 kB)
  2. ↑ project website (Engl.)
  3. ↑ Stainforth et al. (2005): Uncertainty in predictions of the climate response to rising levels of greenhouse gases , in: Nature, Vol. 433, January 27, pp. 403–406 (PDF; 738 kB)
  4. ↑ Annan, JD and JC Hargreaves (2006): Using multiple observationally-based constraints to estimate climate sensitivity , draft of January 30th (PDF) ( Memento of July 8th, 2011 in the Internet Archive )

Web links