Info.png

We've transitioned our blog to Substack

  • Tim Donohue

[Thanksgiving 2021] Sunday Nov 21 first look

We're keeping a curious eye on Newark (EWR) today, where they're landing Runway 29 as we predicted was possible. Otherwise, we're shifting our attention to "go" travel on this Sunday (not to be confused with "return" travel next Sunday). If last year's passenger traffic patterns hold, it figures to be the second busiest "go" day, with Wednesday being the busiest; as we'll see in a moment, it also figures to be a more active day in terms of weather. We think the unsettled weather affirms our work to automate and democratize these flight delay predictions - while we're happy to manually synthesize things, we're not able to dive into the 12-15 airports that the weather forecast asks for. With that said, we'll make a quick pass at the dozen airports we think are at most risk for delays on Friday in today's outlook then return tomorrow with a more narrow focus [and refined weather forecasts in hand]. Before we get underway, we'll also plug our latest explainer on the different ways the FAA can manage an imbalance between airport capacity and demand.


So what do we mean by active weather? The most notable feature is a complex low pressure system that develops over the Great Lakes and tracks towards the Northeast. Rain will develop along nearly the full length of the Mississippi Valley ahead of an associated cold front. As the front swings through the East - pushing it's broad precipitation shield along with it - rain amounts trend down for southern areas. We've previously linked to this NWS explainer on large scale weather patterns that we think is particularly applicable here (tl;dr high pressure generally promotes fair weather while clouds and precipitation are associated with low pressure). Some snow is possible for the Great Lakes on the backside of the low.


As we alluded to at the top of the outlook, the fact that such a broad swath of the country could see impacts from this system requires that we keep our analysis at a somewhat higher level. As such, we've attempted to arrive at reasonable worst (1 in 10) and/or bad (1 in 4) case scenarios for each airport, given a binary measure of ceilings/visibility as well as wind direction (in intercardinal direction, e.g. northwest, not degrees). While snow could result in de-icing delays, no accumulation is expected, so we've attempted to excluded the impacts of snow removal.


With all that said, we'll tackle things alphabetically (by airport code) and aim to spend no more than a sentence or two on each airport (sorry about some repetitive phraseology). We'll also add that our modeling is focused on arrival delays, though there's a strong correlation to departure delays (albeit with some lag and/or possible alleviation; Ctrl/Command + F to search for "correlation" in the above link). Let's get started!


We predict there's a 1 in 10 chance that Atlanta's (ATL) arrival rate falls into the 72-76 range and 1 in 4 chance a 96 rate is realized. Demand, however, peaks at 65 - in the absence of any demand overages, we predict air traffic delays will be negligible. One qualification we won't repeat for other airports: delays owing to aircraft servicing, airline staffing, network effects etc. are always lurking and not included in our modeling.


Given a southerly wind, Boston's (BOS) floor is effectively a 30 arrival rate: we estimate there's about a 1 in 5 chance it occurs. Demand peaks at 31 in the 2 p.m. hour, so minor delays are possible (though they'd be largely imperceptible).


We're betting on a 28 arrival rate at Washington Reagan (DCA), which could produce some largely imperceptible delays (29 arrivals are scheduled in the 4 p.m. and 9 p.m. hours). We should mention an 8% chance that a 26 arrival rate is instead realized, which would produce more persistent delays. This is one we'll be keeping an eye on.


While Dallas-Forth Worth (DFW) will not see rain as a result of the cold front, we are a bit worried about a wind direction change behind the front. Winds will be veering from west to north on Sunday morning - if wind speeds overachieve and the shift is later than forecast by an hour or two, the 9 a.m. hour will be problematic. 100 arrivals are scheduled in the 9 a.m. hour and, while the risk is quite low, strong enough northwesterly winds produce something like a 56 arrival rate; a more reasonable worst case scenario calls for a handful of go-arounds. Nevertheless, DFW too is one we'll keep an eye on.


In the absence of accumulating snow in Detroit (DTW), a 76 arrival rate will effectively be the floor on Sunday (1 in 5). This is an easy one: demand peaks at 37, so there should be zero flirtation with air traffic delays.


Newark. Ah, Newark (EWR). This will be another one we'll revisit tomorrow. For the moment, it's a bit difficult to decide if we set the lower bound of our capacity predictions at 32 or 36. While we don't anticipate that they'd spend all day at a depressed rate, we're confident part of the day will be spent in the low to mid 30s. Just to give a sense of worst case scenario, if they did spend all day at a 32 arrival rate, we've modeled average delays of 47 minutes for arrivals after 2:00 p.m. (peaking at 80 minutes for arrivals in the 8 p.m. hour). We'll refine this in tomorrow's update.


At least EWR get's sandwiched between two easy ones. At Washington Dulles (IAD), an arrival rate of 62-64 will effectively be the floor on Sunday (albeit a quite probable floor - at better than 50% chance). Demand peaks at 40, so we can comfortably predict air traffic delays will be negligible.


One letter (and 1,200 miles) away at IAH (Houston Intercontinental), we think a reasonable worst case (1 in 10) arrival rate for most the day will be 88. Later in the day, winds will shift to the north behind the cold front and, if we were really pessimistic, we could envision a 64 arrival rate thereafter. Regardless, arrival demand peaks at 52, so air traffic delays should be negligible. More notable is the risk for thunderstorms, which have a tendency to halt ground operations [when baggage handlers and fuelers are pulled inside]; thunderstorms can also stop arrival and departure fixes, which you have read about in our latest explainer. Because predictability of ramp closures and fix stoppages is low, we will not be revisiting IAH tomorrow, but readers transiting the airport on Sunday may want to adjust their delay expectations upward.


Unsurprisingly New York-LaGuardia (LGA) finds themselves in a similar boat to EWR, wherein it's tough to glean if the lower bound should be a 32 or 36 arrival rate. Demand is slightly relaxed relative to EWR, so that serves to mitigate delays somewhat. While we'll revisit LGA as well tomorrow, if they were to improbably spend all day at a 32 rate, average delays would peak around 33 minutes in the 9 p.m. hour.


Like DTW, Minneapolis-St. Paul (MSP) is buoyed by the lack of accumulating snow. While morning flurries may drive some de-icing delays, we can't foresee the arrival rate falling underneath 58: that comfortably accommodates the peak demand of 37 scheduled arrivals in the 2 p.m. hour.


We think Chicago-O'Hare (ORD) should be able to deliver a 114 arrival rate; we'd put the chances at less than 1 in 10 that a 92 arrival rate is delivered instead. Regardless, scheduled arrival demand peaks at 82, so our modeling produces negligible delays even at a 92 rate. We will make an important qualification here: our modeling does not see demand placed on the airport by cargo or private flying. This varies by airport (and is generally focused opposite of passenger airline flying), but does introduce some risk that delays are under-forecast. In ORD's case, non-passenger flying adds approximately 4% to demand.


We wrap up with Philadelphia (PHL), which unfortunately finds itself on the revisit list. A 32 arrival rate seems quite likely (on the order of 9 in 10), which presents a problem for the noon (35 scheduled arrivals) and 6 p.m. hour (36 scheduled arrivals). This is more of a wind direction story than a ceiling/visibility story and wind speeds are light enough to at least introduce the possibility of a 44 arrival rate. We could easily see delays approaching 25 minutes in the noon hour and 35-40 minutes in the 6 p.m. hour.


In all cases that we're worried about potential delays, there's enough uncertainty to recommend against rebooking - let's wait and see. We'll be back tomorrow with a closer look at DCA, DFW, EWR, LGA and PHL.



16 views1 comment

Recent Posts

See All