Degree Days and Season Length

Why this matters

"My bill went up a third after you replaced the furnace" is a sentence every heating shop hears in a cold January. Sometimes the equipment is the problem. More often the month was colder, longer or both, and the shop has no number to show it. Degree days are that number: a single figure for how much heating or cooling a period demanded, published free for nearly every town, and the only honest way to say whether a house used more energy or the weather asked for more.

What a degree day measures

A heating degree day (HDD) counts how far a day's average temperature fell below a base temperature. The base used in U.S. weather and utility data is 65 F. The average is taken as the day's high plus its low, divided by two.

• A day with a high of 40 F and a low of 20 F averages 30 F. 65 minus 30 is 35, so that day contributes 35 HDD.
• A day averaging 70 F contributes zero HDD. Degree days never go negative.

Add the days in a period and you have that period's HDD. A month of 1,000 HDD asked for about twice the heating of a month of 500 HDD, for the part of the energy use that responds to outdoor temperature.

A cooling degree day (CDD) is the mirror image: how far the day's average rose above the base. CDD track cooling demand less tightly than HDD track heating, because a lot of cooling load is humidity and sun rather than air temperature, so a humid month of moderate CDD can drive more runtime than a dry month with more.

Why 65 F, and why it may be wrong for this house

The 65 F base is a convention from the mid-twentieth century. It assumed that a house of that era stopped needing heat when the outdoor average reached about 65 F, because people, appliances and sun made up the rest. That outdoor temperature is called the house's balance point.

A tighter, better-insulated house with more internal gains often balances lower, commonly in the mid-to-upper 50s. Using base 65 for that house counts degree days on mild days when the furnace never runs. The error is not spread evenly: it lands mostly in the shoulder months, because on a cold day every base gives a large number and on a mild day the base decides whether there is any number at all. The worked example shows how much.

For comparing a house against itself from one year to the next, the convention usually does no harm, because the same base applies to both years. It matters when the question is how many degree days a particular house actually responds to, which is the question the bill normalization sibling answers with the house's own data.

Where to get the numbers

• The local National Weather Service office's monthly climate report for each reporting station carries a daily high, low, average and HDD and CDD columns. It is the fastest source for a specific month at a specific station.
• NOAA's National Centers for Environmental Information publishes daily station data and the U.S. Climate Normals, the 30-year averages (the current set covers 1991 to 2020) that include normal degree days by month for each station. "Normal" in a weather sense means that 30-year average, not a typical year.
• The Climate Prediction Center publishes weekly and monthly degree-day summaries by city and state, including departure from normal.
• Many utility bills print the period's average temperature or degree days, sometimes against the same period last year. Check which station and base the utility uses before comparing its figures with your own.

Pick one station and keep it. Use the station nearest the house in elevation and setting, not just distance; an airport on a river plain and a house up a ridge can differ by several degrees on still nights. Then use that station for both periods you are comparing. Two stations disagree by enough to swamp a real change.

A worked example: one shoulder-season week, two bases

A customer's newly installed system "ran all week" in early spring. The tech pulls the daily highs and lows from the station report for the seven days and computes HDD at base 65 and at base 60, because the house is a recent build and likely balances lower.

What the totals say. At base 65 the week counts 144 HDD; at base 60 it counts 109. That is 35 degree days apart (144 minus 109), and base 65 reads about 32 percent higher than base 60 (144 divided by 109 is 1.32). Look where the gap sits: on day 2, the coldest, base 65 gives 35 against 30, a ratio of about 1.17; on day 7, the mildest, it gives 6 against 1, six times as much. The colder days carried the week, and the two mild days are where the base choice decides almost everything.

Against a cold week. Compare a midwinter week where every day averages 30 F. Base 65 counts 35 a day, 245 for the week; base 60 counts 30 a day, 210 for the week. Base 65 reads about 17 percent higher (245 divided by 210 is 1.17), roughly half the 32 percent gap of the spring week. Same house, same base error, very different effect on the number, purely because of how mild the days were.

What the tech tells the customer. The week was a real heating week: 109 to 144 HDD depending on base, and the first three days alone produced 93 HDD at base 65 (31 plus 35 plus 27). The system running through those days is the weather. Whether it ran too much for the weather is a different question, answered by comparing the house's use per degree day across periods, not by this week alone.

What would change the read. If the customer had complained in a week like days 6 and 7 repeated, averaging 56 to 59 F, base 65 would have shown 15 HDD for those two days and base 60 just 5. A system running steadily in that weather on a house that balances near 60 would be a real anomaly, the kind worth a diagnostic visit rather than an explanation.

Departure from normal, and saying it on the phone

Most published summaries state a period's degree days next to its normal, and the difference is the departure from normal. Turn it into a percentage on the normal as the base: a January with 1,190 HDD against a normal of 1,050 ran 140 HDD above normal (1,190 minus 1,050), which is about 13 percent colder than normal (140 divided by 1,050). Against last January's 980 HDD it was about 21 percent colder than last year (1,190 divided by 980 is 1.21). Those are two different comparisons with two different bases, and a customer comparing bills is asking the second one.

The office can usually settle a bill call with that one figure, as long as it is honest that the figure explains the weather and not the house:

CUSTOMER: My gas bill is way up since you put the
          new furnace in. What did you do?
OFFICE:   Let me check one thing. Your January had
          about 21 percent more heating degree days
          than last January. That is the weather
          station's count of how cold it was.
CUSTOMER: So it's just the cold?
OFFICE:   It explains a rise around that size in the
          heating part of the bill. If your rise is
          a lot bigger than that, we want to look at
          it. Can you send both bills? We'll compare
          the use per degree day and call you back.

Note what the office did not say: that the furnace is fine. A rise of about the same size as the degree days is consistent with a healthy system; it is not proof of one.

Season length

Season length is the span between the first and last days the average falls below the balance point, and it moves with the base as much as degree-day totals do. Lower the base from 65 to 60 and the heating season shortens at both ends, because the mild days at its edges drop out. Two consequences for the shop:

• Maintenance timing. A heating tune-up scheduled by the calendar can land after the season has started in a cold year and weeks early in a warm one. Scheduling off the station's running HDD, or off the first nights below the balance point, puts the visit where the equipment's first real demand is.
• Annual comparisons. A heating season that started three weeks late and ended on time can carry fewer degree days than normal even if its coldest month was colder than normal. Compare whole-season totals, not the worst month.

What degree days do not tell you

Degree days measure accumulated demand, not peak demand. A week with one brutal night and six mild days can total the same HDD as a week of steady moderate cold, and only one of them tests whether the system can hold the house at design conditions. Sizing and capacity questions belong to design temperatures, not degree days. See related: Design Temperature Versus the Coldest Day Ever.

They also assume energy use rises in a straight line with degree days. That holds well for a furnace or boiler. It breaks for a heat pump below the outdoor temperature where it starts leaning on electric resistance backup, because each degree day on the coldest days costs more energy than a degree day on a mild one. A heat pump house compared across a mild winter and a severe one will look worse in the severe one per degree day, without any fault.

Climate zone maps use degree days too: the IECC's thermal climate zone definitions are written in heating degree days at base 65 F and cooling degree days at base 50 F. That use is owned by the climate-zones card. See related: Climate Zones and Design Temperatures in Plain Terms.

References

• National Weather Service local office monthly climate reports, and the Climate Prediction Center's degree-day summaries
• NOAA National Centers for Environmental Information, daily station data and the 1991 to 2020 U.S. Climate Normals
• See related: How to Weather-Normalize Two Winters of Bills; Climate Zones and Design Temperatures in Plain Terms
• See related: What a Utility Bill Can Tell You That an Inspection Can't
• See related: Design Temperature Versus the Coldest Day Ever