Let's put it this way,...if you have a coupled model that has some of these common systematic biases that are found today, such as, for example, double ITCZ. Then you're going to generate sea surface temperatures (STD which has significant biases in them and if you run high resolution model with sea surface temperature filled with a lot of errors in it. It's no going to look like real world, regardless horizontal resolution of the atmosphere model that you're automatically using. So, once again, you need this balanced program, trying to improve model physics, reduce model bias and maintain the flexible strategy that you're talking about Sugi's model experimentation. I think the balance is the real key for all of these things. You should not try to do one thing only. You need to try variety of approaches so in fact that you don't get hang up on a particular problem. They might be hanged up on variety of problems. That just means that you need work more and harder.
As for management of big projects, I would be happy to give you advice if I knew what to do. Since I don't have a lot of knowledge of what to do. I won't give you a lot advice except I agree with Jerry completely on his brief remarks. You do have to have a lot of integrity, here expect integrity from the people that work with you. Once it gets turned into an egocentric project, then you really loose everything. But you have to satisfy people's needs for some sort of recognition of accompliment. It's delicate plan, but my strategy so far in NCAR has been trying to encourage people to the highest standards, and also trying to encourage junior people to take more responsibility. Lets face it! A lot of us here who have gray hair, we're no going to be here when these problems are solved. And the sooner, we get the younger people involved in trying to help set the course. I take some responsibility. The better I think, we're all going to be on issues like this.
Dr. Julian McCreary (IPRC):
I had a few thoughts here that I could share with you. One is, to make any progress at all, you've got to have better parameterization processes. Now in the ocean, so many key air-sea flux problems are related to the mixed layer. But many GCM today, they even don't have a mixed layer. And there is an example of key problem you need to solve. I think the analogue the atmosphere is equally critical; having a planetary boundary layer that behaves realistically. Maybe one of the key area where that is critical, is ocean - atmosphere interaction in areas where there is not deep convection for the atmosphere is cold. Oceanographers would like to believe that the atmosphere responds to the ocean SST anomalies in north regions. And I can talk atmosphere modelers and they say just doesn't do it. But is this because the atmospheric model that they have been looking at don't have an adequate planetary boundary layer?
