I was notified (by Stuart) of a really bad forecast. I was able to trace this to an unusual situation with strong winds and a dip in the advection profile due to winds from the warmer water crossing cooler land (in England) after a wind shift. This in itself wasn’t a problem, but Stuart was using auto run withOUT ‘Enforce Monotone’ checked. This led the smooth curve fit to make a really bad extrapolation to very high (almost 40 C!) temperatures after the “dip” passed by.
I think I’ve documented somewhere that Enforce Monotone is a good idea in auto run, but I never realized it could matter that much (checking it essentially cured the problem).
I’m working now on a more reliable fix for this. Meanwhile, I recommend checking Enforce Monoton in auto run.
Tom