Impacts of world warming on residential heating and cooling degree-days in the US

Historical (1981–2010) 30-year annual US degree-days

NOAA’s 30-year (1981–2010) US degree-day normals^31,32, calculated for 7,438 weather stations, have CDD values that differ substantially among different regions of the US. When the station data is interpolated to maps, the interpolated CDD values indicate degree-day estimates for locations that lack weather stations (Fig. 1a). The contiguous US is distinguished by a large spread in CDD values, which generally increase from the North to the South^5,20 and decrease at high latitudes. Places in the contiguous US that have the least cooling demand are dispersed across the North of the country and include the Rocky Mountains (with values ranging from 0 to 500 CDD). Conversely, sites that demand the most cooling are approximately located at (1) the intersection of California and Arizona, (2) Southern Texas, McAllen region and (3) Southern Florida, Miami area (4,000 to 4,500 CDD). To characterize CDD within larger regions, we refer to the station median values within those regions. This approach indicates that the overall highest (3,331 CDD) and lowest (3 CDD) CDD medians are found in climatically extreme Hawaii and Alaska, respectively, while an intermediate CDD value (822 CDD) is estimated for the contiguous US. The hottest regions in the contiguous US substantially differ from the areas in Alaska, where all CDD values are close to 0.

Across the US, HDD values usually vary inversely with CDD values: areas with high HDD values correspond to areas with low CDD values and vice versa. This is evident in the north-south HDD decrease across the contiguous US^5,20 (Fig. 1c). As another example, locations that have the lowest HDD values in the contiguous US, such as Southern Florida, the area of Miami (0 to 500 HDD), have the greatest CDD values. On the contrary, regions that have the highest HDD values, such as (1) western Wyoming, (2) upper North Dakota and (3) northern Minnesota (10,000 to 10,500 HDD), have the lowest CDD values. Nationally, HDD values tend to exceed^5,20,24 CDD values. For example, median HDD values in the contiguous US (5,514 HDD) and Alaska (10,958 HDD) exceed corresponding median CDD values by multiplicative factors of 7 and 3,321, respectively. The Hawaii HDD median, however, is only equivalent to 1, because the state is located in the warm tropics.

To provide a key indicator of the total heating and cooling demand and outdoor thermal comfort at various US locations, we add HDD to CDD, producing a combined degree-day metric^10,19. Other key indicators could be based on the amount of sunshine, humidity, wind and characteristics of the built environment. A map based on this interpolated HDD + CDD sum is illustrated (Fig. 2a). In the US, HDD + CDD values are shown to be largest in the northern regions, where the highest HDD values are greater than the highest CDD values in all other areas. Because the number of HDD in the contiguous US tends to predominate over the number of CDD, HDD + CDD values typically decrease from the North to the South, but exhibit a less homogenous trend than HDD values. Areas with the lowest degree-day sum in the contiguous US are located in the West coast of California (1,875 to 2,500 HDD + CDD). In contrast, locations with the greatest degree-day sum are (1) western Wyoming, (2) upper North Dakota and (3) northern Minnesota (10,000 to 10,500 HDD + CDD), which are also the regions with the largest HDD values. The correlation between high HDD values and high HDD + CDD values is evidenced by the relationship between the respective station median degree-days. Compared to the contiguous US (6,386 HDD + CDD) and Hawaii (3,439 HDD + CDD), Alaska, where HDD predominate, has the largest median degree-day sum (10,959 HDD + CDD). Overall, in the United States, HDD + CDD values tend to increase with substantial rises in elevation. This trend is evident in Wyoming and Colorado, where the highest points (Gannett Peak and Mount Elbert, respectively) are located in the same regions as the highest HDD + CDD values. For example, Denver, CO, has a larger (6,671 HDD + CDD) median degree-day sum than Nashville, TN, (5,350 HDD + CDD), which is found at a lower elevation.

Because California encompasses regions with the lowest degree-day sum in the contiguous US, the degree-days of this state are plotted at a finer scale and investigated separately (Figs 3a,c and 4a). The low HDD + CDD values (Fig. 4a) could be attributed to the favourable balance between heating and cooling demand in California, as well as its geographical location. The California CDD station median (846 CDD) is approximately equal to the national CDD station median. CDD values are higher in the hot California’s deserts and in the Central Valley, but are lower along the Pacific coast (Fig. 3a). Mild areas that surround the northern part of the Central Valley (i.e. Del Norte, Siskiyou, Modoc, etc.), as well as California’s coastal zones (i.e. San Francisco, San Mateo, Monterey, etc.) have the lowest CDD values (0 to 625 CDD), while Death Valley in Inyo County has the highest CDD values (5,000 to 5,625 CDD). The California HDD station median (2,666 HDD), however, is about half as small as the national HDD station median, which accounts for the state’s low net heating and cooling demand. In California, only the regions near the Cascade Mountains and Sierra Nevada require substantial amounts of heating (Fig. 3c). Locations with the smallest HDD values are found in eastern San Bernardino, Riverside and Imperial counties (625 to 1,250 HDD), whereas areas with the greatest HDD values are situated on the intersection of Alpine and Mono counties (7,500 to 8,125 HDD). Although Hawaii has the lowest historical station median degree-day sum in the nation, California has the lowest historical degree-day sum (4,180 HDD + CDD) in the contiguous US (Fig. 4a). Coastal areas of Ventura, Los Angeles, Orange and San Diego counties have the lowest⁸ degree-day sum (1,875 to 2,500 HDD + CDD) and areas within the northern Mono County have the highest degree-day sum (8,750 to 9,375 HDD + CDD).

Future (2080–2099) 30-year annual US degree-days

Analysing the historical (1981–2010) distribution of annual US degree-day normals^31,32, we use median projections from the CMIP5 multi-model ensemble to estimate end-of-century (2080–2099) heating and cooling demand under the “business-as-usual” RCP8.5 high emission climate scenario³⁴ (see Methods). Our results demonstrate that the spatial pattern of future (2080–2099) degree-day values is expected to alter somewhat at national scale; however, at regional scale, substantial changes are apparent (Figs 1b,d, 2b and 5). In the contiguous US, for example, the usual trend in the north-south increase in CDD values is evident (Fig. 5a). However, the map’s isolines move up, thus indicating increased cooling demand in regions with historically favourable CDD values^5,20,24. The region with the lowest CDD values will shift northward and be located in coastal Washington, Oregon and the northern part of the Rocky Mountains (500 to 1,000 CDD). The geographic areas with the highest CDD values will remain similar over time, but the CDD values in these areas increase from 6,000 CDD in the 1981–2010 period to 6,500 CDD in the 2080–2099 period. Notably, North-western Washington will have the lowest increase in CDD values (+300 to 450 ΔCDD), while Southern Texas will have the greatest increase in CDD values (+2,250 to 2,400 ΔCDD) (Fig. 5a). As the result of rising temperatures, median cooling demand can be expected to grow in all regions; new station CDD medians estimated to be 2,215 CDD for the contiguous US, 91 CDD for Alaska and 5,288 CDD for Hawaii. Areas with high historical CDD values (e.g. Hawaii, +1,962 ΔCDD) will experience a substantial rise in CDD values, while areas with intermediate or low historical CDD values (e.g. the contiguous US, +1,423 ΔCDD; Alaska, +84 ΔCDD) will experience a moderate increase in CDD values.

In the contiguous US, the trend in the north-south HDD decrease remains present, but the map’s isolines (Fig. 1d) shift up, thus indicating the potential for reduced heating demand in southern residential buildings²³. In addition to the historically warm Central and Southern Florida, for example, Southern Texas, as well as California and Arizona intersection, are expected to have the smallest heating demand (0 to 500 HDD) in the US. The northern Minnesota border and the central Rocky Mountains region in Wyoming, on the contrary, are expected to have the greatest heating demand (7,000 to 7,500 HDD). Furthermore, Southern Florida is projected to experience the smallest decrease in HDD values (−150 to 0 HDD), whilst upper North Dakota, Minnesota and Maine are anticipated to have the largest decrease in HDD values (−3,150 to 3,000 HDD). The reduction in station median HDD is projected to exceed the growth in CDD in the contiguous US (3,470 HDD; −2,011 ΔHDD) and Alaska (7,403 HDD; −3,504 ΔHDD) (Figs 1d and 5b). In the state of the largest HDD decrease, Alaska, the lowest HDD values are anticipated in the Southeast (4,000 to 5,000 HDD) and the highest HDD values are expected in the Far North (13,000 to 14,000 HDD). Yet, in Hawaii (1 HDD, 0 ΔHDD), the station CDD median is projected to exceed the station HDD median.

Interestingly, a new trend will emerge in the contiguous US at the end of the century (Fig. 5с): ΔHDD + ΔCDD will decrease in the North and increase in the South, creating an invisible “0 change” line that will divide the country into two approximately equal regions. Along this line, decreases in HDD will be balanced by increases in CDD. The national distribution of HDD + CDD is expected to be heterogeneous²⁰ and is thus not anticipated to decline in the north-south direction (Fig. 2b). The California coast is expected to have the lowest degree-day sum (2,500 to 3,000 HDD + CDD) in the 50 United States and upper North Dakota is expected to have the highest degree-day sum (8,500 to 9,000 HDD + CDD) in the contiguous US. The largest net degree-day reduction is projected to occur in northern Maine (−2,550 to −2,400 ΔHDD + ΔCDD), while the greatest increase in the degree-day sum is anticipated to occur in Southern Florida (+1,950 to 2,100 ΔHDD + ΔCDD) (Fig. 5c). Furthermore, as a result of anthropogenic climate change, station median degree-day sums in the contiguous US (5,872 HDD + CDD; −614 ΔHDD + ΔCDD) and Alaska (7,518 HDD + CDD; −3,589 ΔHDD + ΔCDD) are projected to decrease, with a reverse effect observed in Hawaii (5,408 HDD + CDD; +1,961 ΔHDD + ΔCDD) (Figs 2b and 5с). The mild Alaskan Southeast is projected to have the smallest degree-day sum (4,000 to 5,000 HDD + CDD), while the arctic Far North is anticipated to have the greatest degree-day sum (13,000 to 14,000 HDD + CDD). In Hawaii, the lowest degree-day sum will predominate on the shores (3,500 to 4,000 HDD + CDD) and the highest combined heating and cooling demand will occur in the Mauna Loa region (8,000 to 8,500 HDD + CDD).

Because the historical degree-day sum in Hawaii is projected to increase more rapidly than in other US regions, California, where annual HDD + CDD normals decline with time, is in the future (2080–2099) projected to achieve the lowest national combined demand for heating and cooling, which deserves a closer examination (Figs 3b,d and 4b). By the end of the century, California’s station median CDD values are anticipated to rise (2,139 CDD; +1,200 ΔCDD), while places with the lowest historical CDD values (625 to 1,250 CDD) are expected to decrease in area, shifting further north (Fig. 3b). The locations with the highest CDD (5,625 to 6,250 CDD) will be found in eastern San Bernardino, Riverside and Imperial. Notably, Death Valley will no longer be located in the region with the greatest number of CDD. The lowest increase in CDD values will occur in western Del Norte (+600 to 650 ΔCDD) and the greatest growth in CDD values will occur in eastern Imperial (+1,850 to 1,900 ΔCDD). California’s station median HDD will decrease from the South to the North (Fig. 3d). Counties within Southern California (i.e. Inyo, San Bernardino, Los Angeles, etc.) are projected to have the lowest HDD values (0 to 625 HDD), while eastern Mono (6,875 to 7,500 HDD) is expected to have the greatest HDD values. Eastern Modoc is projected to experience the smallest change in HDD values (−1,000 to −950 ΔHDD), while north-western San Diego is anticipated to experience the greatest change in HDD values (−2,350 to −2,300 ΔHDD). California station median HDD + CDD is expected to decrease (3,752 HDD + CDD; −345 ΔHDD + ΔCDD) relative to the historical HDD + CDD values (Fig. 4b). Coastal Ventura County will have the lowest degree-day sum (1,875 to 2,500 HDD + CDD) in the state; Mono County, in contrast, is projected to have the highest degree-day sum (7,500 to 8,125 HDD + CDD). Similarly to the contiguous US, California will be divided into two distinct regions: the “North”, where HDD + CDD is anticipated to decrease and the “South”, where HDD + CDD is projected increase. The largest decline in the degree-day sum will occur in eastern Modoc County (−500 to −450 ΔHDD + ΔCDD), while the greatest increase in HDD + CDD will occur in Southern Imperial (+50 to 100 ΔHDD + ΔCDD).

Historical (1981–2010) and future (2080–2099) degree-days in 25 US cities

We tabulate historical (1981–2010) and projected (2080–2099) annual HDD, CDD and HDD + CDD normals interpolated to locations of 25 cities chosen from a list of the 50 most populous incorporated places in the US³⁵ (Table 1; for information on the historical (1981–2010), projected (2080–2099) and Δ degree-days in all of the 50 most populous US cities, see Supplementary Tables S2 and S3). These cities were chosen to illustrate different climate regimes and emphasize geographic diversity. Among these cities, in the historical period (1981–2010), San Francisco, CA, has the lowest CDD value (163 CDD), while Phoenix, AZ, has the highest CDD value (4,608 CDD). Miami, FL, on the other hand, has the lowest historical HDD value (128 HDD) and Minneapolis, MN, has the highest historical HDD value (7,581 HDD). As the combined heating and cooling demand tends to be dominated by heating⁵, the HDD + CDD metric tends to be dominated by HDD. Consistent with this observation, the city of San Diego, CA, is estimated to have the smallest historical degree-day sum (1,946 HDD + CDD) and Minneapolis, MN, is expected to have the greatest historical degree-day sum (8,333 HDD + CDD). Based on projections of future (2080–2099) US climate, Minneapolis, MN, will continue to have the greatest degree-day sum (7,174 HDD + CDD) at the end of the century; however, in the same future period, San Francisco, CA, as opposed to San Diego, CA (3,042 HDD + CDD), is projected to emerge as the city with the lowest combined degree-day sum (2,634 HDD + CDD).

A change in annual degree-day normals may signal a change in corresponding residential heating and cooling demand, as well as outdoor thermal comfort^9,28. By comparing the historical (1981–2010) and projected (2080–2099) degree-day values at several US cities (Table 1), we provide both a qualitative and a quantitative interpretation of the potential effects of global warming on HDD, CDD and HDD + CDD. For example, in Los Angeles, CA, the historical CDD value (1,247 CDD) is projected to experience about a two-fold increase (2,666 CDD) by the end of the century (2080–2099), resembling the historical CDD value for Jacksonville, FL (2,664 CDD). In contrast, historical HDD value (1,083 HDD) of Los Angeles, CA, is anticipated to decrease (309 HDD), becoming closer to the HDD value of present Miami, FL (128 HDD). The historical combined degree-day value in Los Angeles, CA, (2,330 HDD + CDD) is not expected to increase greatly: by the end of the century, the city’s degree-day sum (2,974 HDD + CDD) is projected to be approximately equal to that of present San Francisco, CA (2,816 HDD + CDD). However, San Francisco’s historical demand is primarily for heating, whereas the demand projected for end-of-century Los Angeles is primarily for cooling. In New York City, NY, an analogous warming effect is anticipated: The historical CDD value of New York City (1,105 CDD) is projected to increase by the end of the century (2,348 CDD), approaching a CDD value that historically prevailed in the hot desert climate of El Paso, TX (2,331 CDD). The historical HDD value (4,750 HDD) in New York City is projected to decrease (3,126 HDD) to approximately the number of HDD in present Raleigh, NC (3,246 HDD). New York City’s historical degree-day sum (5,855 HDD + CDD) will decrease (5,474 HDD + CDD), resembling the historical degree-day sum in Oklahoma City, OK (5,463 HDD + CDD).

Some indication of the reliability of the model projections can be gleaned by comparing historical HDD and CDD values as estimated^31,32 by NOAA for 1981 to 2010 with the model projections interpolated to these same cities for the same years. NOAA provides estimates for 42 of the 50 largest cities in the United States. The root-mean-square (rms) difference between CDD values estimated from observations versus model results for the historical period for these cities is 382 degree-days; for HDD values, the rms difference is 586 degree days. The rms difference in CDD values between modelled future and historical climates for these cities is 1587 degree-days; the corresponding value for HDD is 1715 degree-days. Thus, projected changes in degree-days are much larger than model differences between model results and observations for the historical period. Nevertheless caution should be exercised when interpreting results of relatively coarse-resolution climate models at relatively fine spatial scales.