Supplemental Material for:

Projected trends in high-mortality heatwaves under different scenarios of climate, population, and adaptation in 82 US communities

G. Brooke Anderson, Keith W. Oleson, Bryan Jones, and Roger D. Peng

G. Brooke Anderson—Corresponding Author

Colorado State University, Department of Environmental & Radiological Health Sciences

Lake Street, Fort Collins, CO 80521

970-491-1889

Regional trends in community-level projections

For projected trends in high-mortality heatwaves, most communities followed national trend, although some southern communities were projected to have more high mortality heatwaves under RCP4.5 than RCP8.5 under all adaptation assumptions (Online Resource Figures 1 and 2; online version of Figure 1 available at http://brookeanderson.shinyapps.io/App-2). In communities in the northern regions (Northwest, Midwest, Northeast), all ensemble members for RCP8.5 in all communities showed an increase in high-mortality heatwaves compared to the present day, and almost all ensemble members of RCP4.5 showed an increase. Further, almost all northern communities were expected to have more high-mortality heatwaves, based on the community-specific ensemble mean, under RCP8.5 than RCP4.5, although some specific ensemble members of RCP4.5 gave projections of more high-mortality heatwaves than some ensemble members of RCP8.5 at the community level.

Southern communities varied more; some communities in the Southwest, Southeast, and Great Plains, were projected to have more high-mortality heatwaves, based on the ensemble mean, under RCP4.5 than RCP8.5 (e.g., Houston, TX; New Orleans, LA; Tampa, FL). In some of these communities, all RCP4.5 ensemble members projected higher numbers of high-mortality heatwaves than any RCP8.5 ensemble member (e.g., New Orleans, LA). This finding agrees with the findings of another paper in this special issue (Oleson et al. 2015), which found that some southern communities had more frequent, but shorter, heatwaves under RCP4.5 compared to RCP8.5.

This finding could result from longer heatwaves being projected under RCP8.5 than RCP4.5 (Oleson et al. 2015). As a result, some heatwaves under RCP8.5 could be long enough that they combine two or more periods that would have been individual heatwaves had intermediate temperatures been a bit lower. In support of this hypothesis, trends in exposure are higher under RCP8.5 than RCP4.5 in some of the southern communities with higher projected numbers of high-mortality heatwaves under RCP4.5 (e.g., Miami, FL; Knoxville, TN). However, there are some southern communities (e.g., Houston, TX; Tampa, FL) where both numbers of and exposure to high-mortality heatwaves are expected to be lower under RCP8.5 than RCP4.5 (Online Resource Figure 2).

The trend of adaptation scenario creating the largest variation in projections persisted within individual communities (Online Resource Figure 1 and online version of Online Resource Figure 1, http://brookeanderson.shinyapps.io/App-2). When considering projections using different ensemble members of the climate model within specific communities, trends were consistent across health-based models, with the biggest difference between the custom simplified model and the other three models (online version of Online Resource Figure 1, http://brookeanderson.shinyapps.io/App-2).

Online Resource Table S1. Number of and exposure to high-mortality heatwaves projected under RCP8.5 climate change scenario, using only ensemble members 2–15. This table replicates the results of Table 2 in the main text for RCP8.5, but using only the 14 ensemble members as used for RCP4.5, to ensure determine uncertainty associated with using different numbers of ensemble members for the two climate change scenarios. Numbers of high-mortality heatwaves are totaled across all 82 study communities. Exposures are yearly averages, in millions of person-days per year, totaled across all 82 communities. Ranges across ensemble members are given in parentheses, and the percent of all heatwaves or heatwave exposure projected to be from high-mortality heatwaves is given after the comma in each table cell. [Table continues on following page]

Climate scenario: RCP8.5 (ensemble members 2–15 only)
Model / Population scenario: SSP3 / Population scenario: SSP5
Number of high-mortality heatwaves over a 20-year period
No adaptation
Classification tree / 50 (47, 53), 0.39% / 50 (47, 53), 0.39%
Bagging / 49 (45, 52), 0.38% / 49 (46, 52), 0.38%
Boosting / 43 (40, 45), 0.33% / 43 (40, 45), 0.33%
Simple / 75 (70, 78), 0.58% / 75 (70, 78), 0.58%
Lagged adaptation
Classification tree / 24 (19, 30), 0.19% / 24 (19, 30), 0.19%
Bagging / 22 (18, 29), 0.17% / 22 (18, 29), 0.17%
Boosting / 19 (15, 25), 0.15% / 19 (15, 25), 0.15%
Simple / 28 (22, 34), 0.22% / 28 (22, 34), 0.22%
On-pace adaptation
Classification tree / 4 (3, 6), 0.03% / 4 (3, 6), 0.03%
Bagging / 4 (3, 5), 0.03% / 4 (3, 5), 0.03%
Boosting / 3 (2, 4), 0.03% / 3 (2, 4), 0.03%
Simple / 5 (4, 6), 0.04% / 5 (4, 6), 0.04%
Exposure to high-mortality heatwaves, in millions of person-days per year
No adaptation
Classification tree / 53 (48, 58), 0.73% / 93 (84, 102), 0.73%
Bagging / 53 (47, 57), 0.71% / 93 (84, 101), 0.72%
Boosting / 46 (41, 50), 0.63% / 81 (73, 88), 0.63%
Simple / 69 (65, 74), 0.89% / 121 (114, 130), 0.89%
Lagged adaptation
Classification tree / 33 (25, 42), 0.45% / 58 (44, 73), 0.45%
Bagging / 32 (24, 41), 0.42% / 57 (44, 72), 0.43%
Boosting / 27 (21, 35), 0.37% / 47 (36, 61), 0.37%
Simple / 40 (34, 49), 0.51% / 71 (60, 86), 0.51%
On-pace adaptation
Classification tree / 8 (5, 10), 0.10% / 13 (9, 17), 0.10%
Bagging / 7 (5, 9), 0.09% / 13 (9, 17), 0.09%
Boosting / 6 (4, 7), 0.08% / 10 (7, 13), 0.08%
Simple / 10 (8, 12), 0.13% / 18 (15, 21), 0.13%

Online Resource Figure S1. Projected trends in community-level ensemble mean frequencies under different adaptation and climate scenarios. Each line connects three projections for an ensemble member, with points for projections for present-day (left) and for two scenarios of greenhouse gas forcing (RCP4.5 [middle] and RCP8.5 [right]). Each small graph on the map mimics the graph shown in Figure 2a in the main text, with values averaged across ensemble members within each community. Color corresponds to the adaptation assumption: none (red), lagged (green), or on-pace (blue). All projections are based on the bagging health-based model and on the SSP3 population projection. An interactive version of this figure, with toggling between health-based model and between ensemble means or all ensemble members, is available at https://brookeanderson.shinyapps.io/App-2.


Online Resource Figure S2. Ensemble-level projections of number of and exposure to high-mortality heatwaves within communities given SSP3 population growth and no adaptation. Each dot gives the ensemble mean for the community and climate scenario (red: RCP8.5; blue: RCP4.5). Each small vertical line shows the projection for one ensemble member (29 members for RCP8.5; 14 for RCP4.5). The horizontal line shows the range across all ensemble members. Part (a) shows projections of the increase in the number of high-mortality heatwaves in 2061–2080 compared to present; because projections are adjusted for false positive rates for health-based models, fractions of high-mortality heatwaves are sometimes projected. Part (b) shows projections of the increase in person-days of exposure (in millions) to high-mortality heatwaves in 2061–2080 compared to present. Communities are grouped by region (Oleson et al., 2015). [Figure on following pages]

a.


b.