Whether it’s warmer or cooler than average at any given time, cold air is inevitably looming on the horizon come autumn. If you’re not a fan of freezing temperatures, your days in the sun are numbered. But by when does the first freeze usually arrive? This is a tricky metric and it’s often first elevation dependent, then later (November onwards) dependent on the strength of the cold air mass, and in many cases one’s proximity to water.
Just at National Airport in recent history, the first freeze has been as late as December 22 in 2001 and as early as the October 20 in 1992 -- basically a two-month window. While it’s true we don’t all live at the airports, this is more than likely a fairly common spread (if also exaggerated by National’s low elevation next to a warmer river) for earliest and latest freeze across the area.
Let’s take a broader look at what to expect and when to expect it...
As folks who investigate climatology know, the major airports and other NOAA climate locations — however flawed in their location — are easily the best data source available because many have very long histories.
For that reason, while providing an assortment of area-wide information — primarily in the form of the regional map — we’ll focus on the three major climate locations in the D.C. area. Both National Airport (DCA)* and Baltimore-Washington Airport (BWI)* have records dating back to the late 1800s. Dulles Airport (IAD) data goes back to the 1960s.
When reading, also note that we’re focusing on the first freeze rather than the first frost. In many cases, the first freeze and first frost overlap. Sometimes they do not, with the frost preceding a freeze by days or even weeks. However, since a frost can occur with an air temperature (by standard measure off the ground) above freezing, it is exceedingly difficult to obtain accurate data on.
At DCA, the average first freeze does not typically arrive until November 18 per new 1981-2010 norms. This is three days later than the 1971-2000 period and about as late of an average freeze one will find in the immediate Washington area. As recently as 2010, DCA’s first freeze did not come until the very end of November.
The earliest freeze on record at D.C. (at the station location prior to DCA) came on October 10, 1895 when the thermometer read 30 degrees. DCA’s earliest freeze has come twice on the same day – the 20th of October. Both of these earliest freezes did not happen that long ago either, most recently in 1992 and previously in 1972.
The latest first freeze on record came on December 22, 2001. Interestingly, both IAD and BWI recorded their first freeze of 2001 well over two months earlier that year (on October 8). This is just one illustration of DCA’s low-lying river location hardly representing freeze patterns for much of the D.C. area.
As the average freeze-by date would suggest, November is by far the most likely month for DCA to get its first freeze of the cold season. From 1981-2010, only one occurred in October, with 24 in November and six in December. This compares to the 1971-2000 tally of four in October, 21 in November and five in December.
Due partially to its inland location and higher elevation, Dulles Airport can expect its first freeze about a month before DCA. Most of IAD’s first freezes occur in October. Per the new climate norms, the average first freeze at IAD occurs by October 15. This is about two days later than the previous 30-year average, and about one day later than the long-term historical average at IAD.
Dulles Airport’s earliest freeze on record occurred twice on September 24: first in 1974, when the temperature dropped to 30 degrees, and again in 1983, when the temperature dipped to 32 degrees. These two dates are the only time a freeze occurred at IAD during the month of September. The second-earliest freeze was on October 1 in 1993 when it hit 31 degrees.
Unlike DCA, Dulles has never recorded its first freeze any later than November. The latest first freeze came on November 11, in 2005 when it fell to 31 degrees. Both 2009 and 2010 also saw their first freeze in early November. This is the only instance of back-to-back years seeing a post-Halloween first freeze since records at Dulles began 47 years ago.
Baltimore’s first freeze tends to arrive after IAD’s, but before DCA’s. This is probably intuitive given the location’s position northeast of Washington and away from the immediate influence of a river, but also lower elevation and proximity to the coast when compared to IAD.
Interestingly, given that most places in the area are now seeing later first freezes than in the past, both the average and median date of BWI’s first freeze — October 29 — remains unchanged in the new 30-year climate normals. A series of earlier-than-average first freezes since 2000 likely plays a role here.
Whereas DCA sees most of its first freezes in November, and IAD in October, BWI has about equal chances of seeing its first freeze in either month. The earliest freeze for BWI (and in the whole Baltimore climate record) occurred on October 4, 1974 when the temperature dropped to 31 degrees; the second earliest was more recently, on October 8, 2001 when it also hit 31 degrees.
BWI’s latest first freeze occurred on November 13 (both in 1971 and 1986). More recently, BWI witnessed its first freeze during November in 2009 and again in 2010. Only three of the ten years ending in 2010 experienced a freeze before the average date.
Other locations in the immediate area feature a wide range in first freeze dates. For example, Sterling, Va., in Loudoun County, averages its first freeze on October 11. This is a week earlier than Cumberland, Md. in the higher elevations to the northwest. Somewhat oddly, Cumberland’s first freeze on average — October 18** — equals that of Damascus, Md., much closer to the city and at lower elevation.
Less than 10 miles to the southeast of Sterling, Vienna, Va. averages its first freeze on October 31. Another 10 miles to the east, we have DCA receiving its first freeze almost 3 weeks later on average. Also take, for instance, the National Arboretum. The first freeze there is at the tail end of October, again nearly three weeks prior to DCA’s. In fact, large parts of the District average an October first freeze more in line with the close-in suburbs.
Just outside the Beltway to the northeast, Beltsville, Md. averages a first freeze by October 19. Even going way southeast (and across the bay) toward Salisbury, Md. the average first freeze is around October 27 — once again, several weeks earlier than at DCA. Got the picture on this one?
With the first freeze there are some general patterns (i.e., those north and west have better odds earlier), but there is considerable local and even micro-local variance to the average first freeze date. What we do know with certainty is that in most years, the majority of the area has seen its first freeze by the second week of November, if not sooner.
The current map view has 162 locations shown. A complete list gathered by the authors contains 342 locations spread over Md., Va., W. Va., De., NJ, Pa. and NY. We initially gathered as much data as possible from both expansive lists of weather cooperatives throughout the broader region and also via mid-Atlantic and Northeast National Weather Service Offices.
Using the Inverse Distance Weighted (IDW) interpolation scheme, we were able to produce a map showing estimated first freeze dates for the D.C. area and portions of the broader region.
There are some strengths and weaknesses to this approach. A major strength is it allows for a seamless contouring of the first freeze dates across the region without having to do it by hand, while also filling in gaps of areas where data is unknown. If you want a weakness, it’s gaps in the data. As such, the product is not always correct straight out of the box.
For areas where there is a larger distance between any two points, the process has to “guess” more and more as to what value belongs between them. This assumption can cause embellishments in areas, meaning a sub-area surrounding one point that has a first freeze on a certain date may actually be larger in size than it should be.
Through our analysis, and using base contour maps, we have attempted to correct many (if not all) of the mapping errors produced during the input phase. We are confident in the overall appearance of the map, but less so on some specific details. In certain cases, like the mountains to the west, we have traced several higher peaks but not all of them. As a result, the map should be viewed as general guidance and not hard fact (even if the isolated locations which fence the contour are indeed factual).
Frequency of First Freeze by Month (1981-2010)
DCA: Sep., 0; Oct., 1; Nov., 24; Dec, 6
IAD: Sep., 1; Oct., 25; Nov., 4; Dec, 0
BWI: Sep., 0; Oct., 17; Nov., 13; Dec, 0
Frequency of First Freeze by Month (1971-2000)
DCA: Sep., 0; Oct., 4; Nov., 21; Dec, 5
IAD: Sep., 2; Oct., 26; Nov., 2; Dec, 0
BWI: Sep., 0; Oct., 17; Nov., 13; Dec, 0
First Freeze Average Temperature
DCA: 1981-2010: 31.3 degrees; 1971-2000: 30.1 degrees
IAD: 1981-2010: 31.2 degrees; 1971-2000: 31.1 degrees
BWI: 1981-2010: 30.1 degrees; 1971-2000: 30 degrees
*Both D.C. and Baltimore climate records are from two different locations in and around their respective cities. NOAA and CWG consider these one record so they will be referenced as DCA or BWI generally unless specificity is required.
**In the contour map above we have shaded in the whole area around Cumberland in the earlier period as we feel that specific location to be an outlier among colder temperatures. Several locations like this were edited in the map but their original data is shown in the attached files.
Ian Livingston and Justin Grieser are contributors to the Capital Weather Gang. Katie Wheatley is a geographic information system (GIS) analyst from the Baltimore area.