Sep 21, 2012 10:25 PM by John Patrick
TUCSON - University of Arizona Scientists have found some big problems when it comes to the small scale weather models.
A new study by the University of Arizona shows that climate models struggle with accuracy when it comes to time frames shorter than three decades and small scale areas. Where computer climate models gain advantage is when they are applied long term global patterns.
Koichi Sakaguchi, a doctoral student who headed the study explains, "We found out is very difficult to predict climate at a much smaller scale for say Tucson."
Part of the reason for this inaccuracy is due to the resolution of the models which can only break the earth down into 60 square mile boxes. Areas such as Tucson are too small for that to get an accurate forecast.
The UA graduate students tested the prediction skill of the computer models with a process called hindcasting.
Michael Brunke, UA graduate student, explains the process of hindcasting, "We compared what the model simulated for the twentieth century to the climate records we already know for the twentieth century."
By applying the hindcasting method to the technology of today we can bring more accurate predictions down the road. "So now we found out that the models are not so good, for example say Tucson, so we should maybe add this particle detail to the model to see if it improves or not," says Sakaguchi.
Another goal of this study was to bridge the gap between forecasters and climatologists so a more precise picture of climate change can be painted. Brunke explains, "Climate models are basically the only tool that we have to try to figure out what's going to happen in the future because we don't know what the future holds."
As climate models continue to improve so will the track record with the community to trust future predictions.
This study has been published in the Journal of Geophysical Research-Atmospheres. To view the entire study and all of the research click here.