About half of AGU 2011 is over. There are again more people here than were last year. And it is great to meet fellow scientists. Generally, I find this year there are also more general overview talks than previously, which can be nice. Upmanu Lall gave this morning a nice overview over the issues faced by hydrologists. Especially in countries that are currently labeled as developing countries. He coined the term “Hydromorphology” indicating how the use of water has and will change the conditions on earth and the human beings on earth. In the same session Thorsten Wagener gave also an overview presentation on hydrological models. He showed a nice example of bias (The figure is from my memory, which is also biased. Additionally, this is a discrete distribution, so there should be bars, but in the limited time available, I could only draw lines quickly).
Generally, I still see an overwhelming amount of presentations that use some sort of linear model in their scientific work. This is strange, because everywhere I look I see non-linear behavior. Again, I might be biased (also due to my involvement in NUPUS).
I do like the poster session. It’s fairly easy to get into slightly different areas of research due to the proximity of posters. So I went and saw a poster by some guy at Harvard, who has built a really simple linear regression to “model” the dependence between days with average temperature over 8C, with average temperature below 28C, growing days and crop yield. He admitted that this is not good for prediction. I think this is one example demonstrating how little statistics is used in “science”.