Thursday, June 3, 2010

Changes to CRU Database

I recently wrote about the controversy regarding climate change and malaria. I've now applied a couple of different models to both the data we used in our 2002 Nature paper and data from the latest version of the CRU database. Our Nature paper used an older version of the CRU database. One of the models I used was the basic structural model also used by Chaves and Koenraadt. The other one is a test for trends in time series that is robust irregardless of the true model underlying the data.

My estimates of the basic structural model came out nothing like those of Chaves and Koenraadt casting doubt on their results but on reflection I don't think this is a particularly useful model for addressing this question. I don't really want to get into that here but would be happy to answer questions. The trend test finds though that:

1. There is no significant increase in temperature in the data we used in our 2002 paper in line with our conclusions there.

2. But that there is a significant increase in temperature for the new CRU dataset, especially for Kericho in Kenya. It's easy to see the differences between the old and the new series.

3. Adding post 1995 data increases the effect further.

The key point is that CRU have changed their data. Chaves and Koenraadt appear at this stage to be correct that there was an increase in temperature at Kericho. But it appears that that is due to the changes in the dataset and not due to the specific modeling reasons that they cite in their paper.

No comments:

Post a Comment