A bit of a rant…

Today’s weather models are getting better, but one thing they still struggle with is predicting the timing of the onset and ending of long-lasting weather events… hot and cold spells, dry and wet spells… esp. more than a few days into the future.

1/n
and add in the factor that the statistical techniques that are still used for predicting temperatures naturally tend toward predicting average (typical) temps farther into the future as their skill diminishes.

2/n
Why does this matter? During our cold spell, which began on Feb 2, the long-range forecasts have continually called for gradually warming temperatures. The Day-7 temp forecasts have shown -15°C (which did not happen), which get pushed back by one day every day.

3/n
more recently, the forecasts have shown dramatic warming by Day-5 or so; again, the warming has been pushed back by one day every day. the warming is coming. we just need to be patient 🙂

4/n
see the attached graph, which shows the daily high temps (thick blue line), along with each day’s extended forecast. Note how they all show temps starting to rise 2-5 days into the future.

5/n
This is not real; it simply reflects how forecasts are generated and the drop in skill in weather forecasts as one goes farther into the future.

6/n
A big problem is that we (i.e. Environment Canada) still show a single number for temps 5-6-7 days into the future, which is misleading. We don’t give any indication of the skill (or confidence) in that number, or the possible range in temps.

7/n
Partially because that’s what people (and the media) want and are comfortable with, but it’s not particularly helpful.

8/n
What would more useful is to say that the high temp 7 days from now is predicted to be -10 °C ± 2 °C (or -8 to -12 °C), 19 times out of 20. Just like surveys and polls say. While harder to interpret, it better reflects our ability to predict the temp that far in advance.

9/n
And that saying -10 °C ± 2 °C on one day and-10 °C ± 8 °C on a different day provides additional info; i.e. it reflects a greater degree in confidence. And that’s helpful for people and businesses making plans.

10/n
if you have a scientific bent, here's how the EC model "sees" the evolution of temps, cloud cover, precipitation and wind over the next 7 days... and note how it shows the spread in values that increases as one look looks forward in time (it's a "bars and whiskers" graph)

11/n
and the ECMWF (European)... the best weather model in the world.

what does this mean? when we're in a hot/cold/wet/dry spell, be wary of forecasts telling us that things will return to normal in 6-7 days

12/n
You can follow @SteveRickettsSP.
Tip: mention @twtextapp on a Twitter thread with the keyword “unroll” to get a link to it.

Latest Threads Unrolled:

By continuing to use the site, you are consenting to the use of cookies as explained in our Cookie Policy to improve your experience.