Daylight Saving Time (DST) starts in spring. That’s when we have to wake up earlier because the clocks shift forward one hour. Standard Daylight Time returns everything to normal in the fall. Forecasters can use this information to build a high-performing time-of-day arrival model.
The easiest way to deal with Daylight Saving Time turning on and off is to keep two normal intraday pattern sets; one to use when daylight saving time is active, and one when it’s not. Problem solved!
But some groups are driven by the sun, like tourism. For these groups, I get more precision in my time-of-day arrival models when I’m using the first suggestion (two distinct DST intraday models for on/off) and also ramp the DST On factor up a few weeks before it happens. For example, Daylight Saving Time starts on Sunday, March 8th, and I can introduce a mix of DST On/Off for the spring season starting 10- to 25 days earlier, to avoid the “jolt” of instantly being in a completely different time zone.
Some communities are completely dictated by sun and season, like farmers who get up with the cows in agricultural areas. In these cases, I extend my normal intraday models, along with the DST-on/off models, and divide them into actual seasons, too. It ends up looking like this:
Spring = March 8 through June 20 (Summer Solstice)
Summer = June 21 through September 22 (Fall Equinox)
Fall = September 23 through November 1 (Daylight Saving Ends)
Winter = November 2 through March 7 (Daylight Saving Starts)
Daylight Saving Time begins on the second Sunday in March and ends on the first Sunday in November, except for Arizona and Hawaii, who get to stay on standard time all year long.