Uncategorized

Probabilistic projections of increased heat stress driven by climate change | Communications Earth & Environment – Nature.com


Projections of heat stress

To quantify the degree to which climate change will increase human heat stress, we first need probabilistic projections of global mean temperature change driven by anthropogenic CO2 emissions. Figure 1a shows the probability density functions of atmospheric CO2 concentrations in the years 2050 and 2100. These were produced using a joint Bayesian model of change in population, Gross Domestic Product, and carbon intensity by country11,12.

Fig. 1: Projections of CO2 emissions and global mean temperature change through 2100.
figure 1

a shows probabilistic projections of atmospheric CO2 concentrations in 2050 and 2100. b shows a probability distribution of transient global climate sensitivity written in terms of °C warming per 100 ppm of atmospheric CO2 change. c shows the convolutions of the probability distributions in a, b, which yields a probability distribution of global mean temperature change (relative to the 1850–1900 baseline) in 2050 and 2100.

While other greenhouse gases contribute to climate change on decadal to centennial timescales, atmospheric CO2 concentrations are highly correlated with global mean temperature change across a variety of climate change scenarios (see Table S1). We use linear best-fit regression to calculate the relationship between global mean temperature change and atmospheric CO2 concentrations in each of the 23 climate models that participated in the Coupled Model Intercomparison Project Phase 6 (CMIP6). The probability distribution of this linear relationship, which we refer to as the transient global climate sensitivity, is shown in Fig. 1b. The uncertainty in transient climate sensitivity is similar to that reported in other studies13 and reflects the different model physics and parameterizations that lead to various amounts of global warming across climate models forced by the same amount of anthropogenic emissions.

Anthropogenic emissions have already warmed the planet by roughly 1.0 °C as of 2000–2020 relative to 1850–1900 baseline14,15. Using the probability density functions in Fig. 1a, b along with the 1 °C warming already observed, we generate the probability density functions of global mean temperature change in 2050 and 2100, all relative to the 1850–1900 baseline used by the IPCC16, shown in Fig. 1c. For 2050, the [5, 50, 95] percentile changes in global mean temperature are [1.5, 1.8, 2.3] °C while for 2100 these percentiles are [2.1, 3.0, 4.3] °C. These projections indicate that there is only a 0.1% chance of limiting global average temperature change to the Paris Climate Agreement aspirational goal of 1.5 °C by 2100. Note that the full statistical model does not explicitly take into account the possibility of more aggressive policy actions such as negative emissions technologies and also does not consider overshoot strategies to achieve particular global warming targets.

To connect these probabilistic projections for the global mean temperature change to changes in local Heat Index, we calculate the ratio of local changes in temperature to global mean temperature change for each calendar month for each of the climate models and then averaged the results across the 23 climate models that we analyzed. This pattern scaling approach allows our probabilistic projections of global mean temperature change shown in Fig. 1c to be applied anywhere in space. The local mean temperature change for each month at each place in space is the product of the global mean temperature change in the scaling ratio ST shown in Fig. S1. The same procedure is applied to render the local change in relative humidity in each calendar month (see the scaling patterns SRH in Fig. S2). In Fig. 2, we show the regional changes in local temperature associated with the median global average temperature change (3.0 °C) projected for the end of this century. Changes over land regions are typically 5 °C, with greater increases in the Arctic.

Fig. 2: Local temperature change across months.
figure 2

Mean temperature change in each calendar month from the 50th percentile 2100 warming scenario (global mean temperature change of 3.0 °C).

To understand how the Heat Index changes with global warming, we first calculate the daily maximum Heat Index from 1979 to 1999 using daily observations of maximum temperature and monthly average observations of specific humidity (see Supplementary Methods). We then use six different scenarios of global mean temperature change that correspond to the [5,50,95] percentiles in 2050 and 2100 calculated from the PDF shown in Fig. 1c and the scaling patterns for temperature and relative humidity (Figs. S1 and S2) to calculate the change in the climatological mean temperature and relative humidity at each place in space for each calendar month in each of the six scenarios. This method of relating global to local mean temperature change takes advantage of a well-known pattern of temperature change seen across several generations of climate model ensembles17 and the (negative) correlation between global temperature and terrestrial relative humidity change18.

We applied the relevant local changes in climatological temperature and relative humidity to the observed daily temperature and relative humidity (1979–1998) for each of the six climate change scenarios we considered, and then used these records as inputs to calculate the Heat Index according to the Rothfusz equation19. This procedure takes into account uncertainty in both projected CO2 emissions and climate sensitivity, but not the small uncertainty associated with regional uncertainty in global climate change projections (see “Methods” and Figs. S3S5). Figures 3 and 4 show the average days per year where dangerous and extremely dangerous Heat Index thresholds are exceeded under six climate change scenarios, as well as in the observational record from 1979 to 1998 (see “Methods”).

Fig. 3: Projections of dangerous Heat Index values.
figure 3

a shows the average number of days per year when the dangerous Heat Index threshold was exceeded in the historical record (1979–1998). bg show the same quantity under the various climate change scenarios noted in each panel.

Fig. 4: Projections of extremely dangerous Heat Index values.
figure 4

Same as Fig. 3a–g but for exceedances of the extremely dangerous Heat Index threshold. Red contours in all panels outline regions where the extremely dangerous Heat Index threshold is exceeded more than once per year on average.

Over the period 1979–1998, the dangerous Heat Index threshold was exceeded on roughly 5% of the days in each year in the tropics and subtropics (between 30°S and 30°N), and for 10–15% of the days in each year in subtropical Africa, the Indian subcontinent, and the Arabian peninsula (Fig. 3a). In the midlatitudes, the dangerous Heat Index threshold was exceeded less often; in many places, these exceedances represented extreme events that occurred less than once per year in the 20-year record we examined. Exceedances of the extremely dangerous Heat Index threshold were rare across the globe in the 1979–1998 record (see Fig. 4a). The most frequent exceedances of the extremely dangerous Heat Index threshold were concentrated in the coastal regions of the Arabian peninsula and Northern India and occurred between once and three times per year in the historical record.

The global warming scenarios present troubling projections of increasing heat stress driven by anthropogenic emissions. In the tropics and subtropics, where the dangerous Heat Index threshold was typically exceeded on less than 15% of the days in each year between 1979 and 1998, we project that, by 2050, many people living in these regions will likely experience dangerous Heat Index values on between one-quarter and one-half of all the days in each year (Fig. 3c). By 2100, the median projection is that most regions in the tropics and subtropics will exceed the dangerous Heat Index threshold on most of the days in each year (Fig. 3f). Many regions in the midlatitudes will experience dangerous Heat Index values on between 15 and 90 days each year—in some places, this represents an order of magnitude increase in the frequency of exposure to dangerous heat stress from the 1979–1998 period.

The Heat Index rarely exceeds the extremely dangerous threshold in the current climate (Fig. 4a). In the median projection for 2100 (Fig. 4f) extremely dangerous heat stress will be a regular feature of the climate in sub-Saharan Africa, parts of the Arabian peninsula, and much of the Indian subcontinent. The extremely dangerous Heat Index threshold is likely to be exceeded on more than 15 days in each year by the end of the century in these regions, this will likely require massive adaptation measures for a large number of people. In our 95th percentile projection, which corresponds to high emissions and high climate sensitivity (see Fig. 4g), the Heat Index will exceed extremely dangerous levels on between 15 and 25% of all days in each year in some tropical and subtropical regions.

Chicago—a case study

As an example from the midlatitudes, we turn to Chicago; a major urban center whose history illustrates the dangers of extremely high temperatures. An extreme drought swept the United States during the summer of 1988, causing billions of dollars in damages to the agriculture sector across the United States20. During the drought, Dr. James Hansen gave congressional testimony that human-induced increases in greenhouse gases could increase the probability of extreme events such as summer heat waves. These events marked a turning point in the public understanding of climate change.

During the 1988 heat wave, the Heat Index in Chicago was 5°F higher than average over the 1979–1989 period, but the 103°F “dangerous” threshold was never exceeded. Seven years later, in 1995, a heat wave devastated Chicago and caused nearly 800 excess deaths21. This event consisted of 4 consecutive days (July 12–15) when the Heat Index exceeded 100°F. Such an event (4 consecutive days of maximum Heat Index >100°F) occurred only twice in the 1979–1998 record, both times in 1995, but the other 1995 event had a lower average intensity and occurred later in the summer.

By randomly sampling 1000 scenarios of global mean temperature changes from the distribution shown in Fig. 1c and using the local scaling patterns for Chicago’s place in space, we quantify the average change to Chicago’s daily Heat Index record by 2100. To do this, we augmented the 1979–1998 record of temperature and relative humidity in the same manner as was done in Figs. 3 and 4 (see “Methods”).

The 1979–1998 record shows that a daily Heat Index of 100°F was not exceeded in 11 out of 20 years. The same 20-year record modified by the median projections of temperature and relative humidity changes for the end of this century has at least one exceedance of this threshold each year. Further, heat waves like the kind that Chicago experienced in 1995 are projected to become a regular occurrence by the end of the century in our median projection: two 4-day periods with daily maximum Heat Index >100°F were found in the 20-year historical record (1979–1998); our median projection shows 32 such events in a 20-year period at the end of this century. This 16-fold increase in the number of potentially dangerous heat waves points to the kind of societal adaptation required to combat these phenomena in the midlatitudes. This order of magnitude increase in the number of heat waves in a 20-year record is reflected in our median projection of the number of days per year where the dangerous Heat Index threshold is exceeded. In the 1979–1998 record, the dangerous threshold (103°F) was exceeded four times (all in 1995), while an average of 11 exceedances of this threshold each year is likely by 2100.

LEAVE A RESPONSE

Please help keep this Site Going