CLIPER - supposedly inaccurate - see Rita @ 9/21/2005

[RBDnhm]

Please accept my apologies in advance if this a stupid question or posted in the wrong place.

As a Rita evacuee, I was wondering if any of the computer models have been modified to compensate for their obvious errors (I understand landfall was within the expected landfall zone but I bet it was just barely from a statistical standpoint and not very close from a real world standpoint) in predicting the actual landfall of Rita and if so, how?

I vividly recall CLIPER being more accurate at the time it really mattered (9/20 - 9/22/2005) than the other models. There is is at least one article by Derick Ortt (I think that is his name) from about May 2006 identifying and acknowledging the nature of the errors.

[ncweatherwizard]

Not a bad question at all.

We all know models will have significant errors sometimes, and they probably will at least as long as I'm alive. Regarding your question, forecast models aren't fixed based on many specific downfalls, such as inaccurate forecasting on a storm like Rita. A couple of models in particular: GFDL, and the new Hurricane Weather Research and Forecasting model (HWRF) are set on a grid. These are improved over time every year or every other year or so, often by increasing resolution (the amount of space between the grids), and making changes to some complicated equations in the model to improve how the models handle atmospheric phenomena.

Regarding CLIPER, it isn't a model in the sense that the other models use some sort of physics and require initialization of a tropical cyclone for them to run. CLIPER is a statistical model: using the tracks of previous storms in similar situations, it provides a forecast of a possible track based on other previous storms. Typically though, you never have the same exact setup happen twice in a row, so if CLIPER is right, it was either a really easy forecast or plain luck.

I think someone once posted a good thorough explanation of different models, but I don't know where it is now; perhaps someone else does.

I see it's your first post...welcome aboard!

Scott

[RBDnhm]

Thanks for the reply. I will try to find the description of the models you mention elsewhere in here, although they have disabled the "search" function so it may take a while.

Are the formulas (and variables) for any of the models disclosed publicly? or is each some super secret proprietary thing?

[RBDnhm]

Please disregard my last post. Am headed back to Tropical Reference section now.