Better Forecasting Through Mathematics

I just sat through a Powerpoint presentation for a new suite of computer modeling weather forecast tools which become operational Tuesday. My mind has been boggled.

When I started in the weather racket almost 25 years ago, we were dealing with whatever speedy computers of that era could handle. Not much as it turns out!

I used to get weather data via teletype and ‘wet paper’ fax. The fax machine was a mechanical beast with a metallic band, rapidly spinning drum and wet paper. Somehow when all was said and done, maps showed up on the paper.

These were pre-digital days. The fax machine was on a timer. Unfortunately, the maps weren’t always on time, so there were plenty of times you missed out.

The model of choice back then was the LFM for Limited Fine Mesh. I don’t remember much about it, except getting burned!

As it turns out, when the model was formulated there was a math screwup. All precipitation totals had to be divided by two to be correct! The outputs couldn’t be changed, because no one really knew what else was dependent on them.

It made little difference. Even today, precipitation amount is one of our least skillful parameters.

There have been a few radical changes since the LFM – each bringing better forecasting skill. Part of that is because of faster computers, allowing more data to be crunched. We’ve (and when I say ‘we’, I mean someone other than me) also managed to better integrate satellite, radar and other observational data into the models.

When I saw some of the graphic outputs from this new model (the NMM WRF), I was impressed. I don’t know how correct what I saw was, but it’s obviously higher resolution than what I’ve been working with.

Will this help? I hope so – and the presenter was positive in his tone.

The truth is, I’d use the Psychic Friends Network for my forecasts if they worked! Fast computers are used because they’re the best we’ve got, not because I like playing with computers.

It will still be an inexact science, but maybe a little less inexact. That would be very nice.

Leave a Reply

Your email address will not be published.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