NCAR Talk Debrief

“If I want high performance code written, I have an electrical engineer do it, not a computer scientist.”

  • Dr. Richard Loft . – April 25, 2018

I attended Dr. Rich Loft’s presentation at NCAR (National Center For Atmospheric research) tonight.  He is the computing CTO at NCAR, and is a computer scientist/physicist – not a climate scientist. The first 55 minutes of his talk was great, but the last five minutes was the mandatory climate BS.

One of his first slides was a 2002 drawing by David “Children Just Won’t Know What Snow Is” Viner, explaining how atmospheric models are structured.

This slide showed the massive wiring interconnect at NCAR’s Cheyenne computing facility.

Then he showed what their supercomputers can do, and pointed out that the NCAR model failed to predict Hurricane Sandy at the end of October, 2012 – due to “incorrect initial conditions.”

Finally came the mandatory climate BS. (Arctic ice-free in 50 years, Jupiter sized hurricanes, etc. etc.)  Again,  Rich isn’t a climate scientist, and is just parroting what he hears from climate scientists in the office.

After the presentation, I asked him why he thought models which failed to predict Hurricane Sandy and break down 72 hours in the future,  would be able to predict 50 years in the future.  He replied with “our models are good for seven to ten days, and it is the difference between weather and climate.”  I asked again why he trusts models forecasts 50 years in the future, at the same time he believes the models are only good for seven ten days. He came back with “the climate models are statistical models.”  To which I replied “isn’t the core of the climate models the same as the core of the weather model?”  He said they were.

I didn’t pursue it any further than that.  We have had the same conversation many times in the past. His interest is high performance computing, which NCAR has always led the way at.

Rich was really happy to see me, and introduced me as a “colleague from way back.” We talked for a long time afterwards, and I think we will likely be collaborating on some WebGL based visualization work – which is my main focus these days and is exactly what NCAR needs for some of their future product plans.

About Tony Heller

Just having fun
This entry was posted in Uncategorized. Bookmark the permalink.

Leave a Reply

Your email address will not be published. Required fields are marked *