Increased population exposure to precipitation extremes under future warmer climates

The National Center for Atmospheric Research (NCAR) CESM low-warming simulations that were designed for assessments of climate change impacts and mitigation options under the 1.5 °C and 2.0 °C warmer climates, as well as the representative concentration pathways scenario 8.5 (RCP8.5) large ensemble simulations (Kay et al 2015, Sanderson et al 2017), were used here to evaluate the future changes in precipitation extremes and the consequent changes of population exposure around the world. CESM is a fully coupled ocean-atmosphere-land-sea ice model with the horizontal resolution of 0.9° latitude by 1.25° longitude for the atmosphere and land and 1.0° latitude by 1.0° longitude for the ocean. To reach these limited warming goals, a simple Minimal Complexity Earth Simulator model was first employed in the CESM low-warming simulations to produce the optimal emission pathways and then to force the CESM. In these two low-warming emission scenarios, a well-mixed greenhouse gas concentration is redesigned, but other anthropogenic forcing agents, including aerosol, land use, chlorofluorocarbon, and ozone, follow the RCP8.5 protocol (Sanderson et al 2017). These simulations include 11-realization ensembles covering the period of 2006–2100, and the global mean surface air temperature stabilizes at approximately 1.5 °C (2.0 °C) above preindustrial times at the end of the 21st century (figure S1, available online at stacks.iop.org/ERL/15/034048/mmedia). In this study, the 1.5 °C and 2.0 °C future climates are thus defined for the period of 2081–2100. Both the historical simulations and the RCP8.5 large ensemble simulations run a 40-realization ensemble of the CESM. The global surface air temperature is projected to increase by approximately 5.7 °C by the end of this century in relative to the preindustrial period under RCP8.5 scenario (figure S1). Here, the first 11 members from these 40 runs were selected for the analyses in order to match the members in the CESM low-warming experiments. The results of the median ensemble means from these 11 members are mainly shown in the following.

To assess the CESM performance, the observed gridded daily precipitation data derived from approximately 16 000 stations and satellite observations, which were acquired from the National Oceanic and Atmospheric Administration (NOAA) Climate Prediction Center (CPC), were used in this study. This dataset was constructed using the optimal interpolation approach, and it spanned from 1979 to the present, with a horizontal resolution of 0.5° of longitude-latitude (Chen et al 2008). For the convenience of comparison, the observed dataset was resampled to the CESM grids with a horizontal resolution of 0.9° × 1.25° using a first-order conservative remapping procedure via the Climate Data Operator (CDO). The present day here is defined as the period of 1986–2005 in both observation and historical simulations of CESM.

The population exposure to precipitation extremes generally means the number of people exposed to the heavy precipitation events, which is defined as the frequency of extremes multiplied by the number of exposed people (Jones et al 2018, Chen and Sun 2019, 2020). The population distribution in the year 2000 and the future projections under different shared socioeconomic pathways (SSPs) from 2010 to 2100 (Jones and O'Neill 2016) were used here to explore the exposure changes from the different future warming climates.

The precipitation extreme that this study interested in was defined as the daily precipitation greater than or equal to the daily 95th percentile (R95p). The threshold for the 95th percentile was calculated from the past three decades of 1976–2005 for each grid. Assessments of previous studies (e.g. Lin et al 2018) documented that the CESM can reasonably reproduce the probability distributions and time evolution of temperature and precipitation anomalies. Our evaluations further show a high capability of the CESM to reasonably capture the spatial pattern of precipitation extreme days based on comparisons to observations (figure S2), with a pattern correlation of 0.8. However, overestimation is clear across the global land areas for the extreme days by the CESM, especially over the regions of Africa, Asia, Australia, and the western coasts of North and South America (figure S2). The quantitative estimation shows approximately 47% overestimation of the extreme days that aggregated over the global land grids. Bias correction is thus urgently needed before further study is implemented.

A popular correction method proposed by Watanabe et al (2012) was employed here that first corrects the statistical parameters for the baseline period and then corrects the projected variables using quantile-based mapping methods. For precipitation, the days of precipitation equal to zero have to be excluded before the estimation of statistical parameters, because the two gamma distributions applied here are only available for positive variables. Thus, a threshold, calculated as the percentile of uncorrected data at which the corresponding observed precipitation exceeds zero, was introduced in this method to treat the portion of uncorrected data less than it as zero. This threshold would be zero when the number of non-precipitation days in the uncorrected data exceeds the observed number of non-precipitation days. Then, the mean (μ) and coefficient of variation (CV) are corrected as follows:

$$\begin{eqnarray}&&{\mu }_{c}=\displaystyle \frac{{\mu }_{p}{\mu }_{o}}{{\mu }_{b}}\end{eqnarray} \tag{ 1 }$$

$$\begin{eqnarray}&&C{V}_{c}=\displaystyle \frac{C{V}_{p}C{V}_{o}}{C{V}_{b}}.\end{eqnarray} \tag{ 2 }$$

The subscripts o, p, b, and c mean the observation, projection, baseline, and correction. Using the corrected mean and CV, we estimated the two statistical parameters of the gamma distribution, α (shape) and β (scale), which were finally employed to calculate the precipitation via the two-parameter gamma distribution. The parameters of α and β are calculated using the L moments method.

$$\begin{eqnarray}&&{x}_{c}={F}^{-1}\left(F{x}_{p};\,{\alpha }_{p},\,{\beta }_{p};\,{\alpha }_{c},\,{\beta }_{c}\right).\end{eqnarray} \tag{ 3 }$$

This correction process was implemented on the daily precipitation for each month. After correction, the overestimation on the precipitation extreme days was substantially reduced across the global lands, especially for some regions of Africa and Asia, when compared with the uncorrected result (figure S2). The pattern correlation of the corrected result was up to 0.94, and the bias decreased to approximately 5% aggregated over the land. Thus, the corrected simulations were used in this study to further assess the changes of precipitation extremes and the associated exposure.

The avoided impacts due to the lower warming in the future are estimated as follows:

$$\begin{eqnarray}&&{\rm{AI}}=\displaystyle \frac{{C}_{2.0}-{C}_{1.5}}{{C}_{2.0}}\times 100 \% .\end{eqnarray} \tag{ 4 }$$

Here, AI is the result of the avoided impact, and C_1.5 and C_2.0 are the changes of precipitation extremes under the 1.5 °C and 2.0 °C warmer climates, respectively, against the present day.

The roles of climate and population changes on exposure were also investigated. As Jones et al (2015) proposed, the exposure change (ΔE) could be decomposed into three parts, including the climate effect, the population effect, and their interaction effect estimated by equation (5).

$$\begin{eqnarray}&&{\rm{\Delta }}{\rm{E}}={P}_{1}\times {\rm{\Delta }}C+{C}_{1}\times {\rm{\Delta }}P+{\rm{\Delta }}P\times {\rm{\Delta }}C.\end{eqnarray} \tag{ 5 }$$

P₁ and C₁ are the populations and precipitation extreme days at the baseline time, respectively, and ΔP and ΔC are their corresponding changes at future warmer climates with respect to the baseline time. Thus, the terms of ${P}_{1}\times {\rm{\Delta }}C,$ ${C}_{1}\times {\rm{\Delta }}P,$ and ${\rm{\Delta }}P\times {\rm{\Delta }}C$ represents the climate, population, and interaction effect on exposure change. If the equation (5) is divided by E₁ for both sides, this equation can provide an estimation of the percentage change for each component (equation (6)). Thus, we can discuss the contribution of each component to the exposure change.

$$\begin{eqnarray}&&\displaystyle \frac{{\rm{\Delta }}E}{{E}_{1}}=\displaystyle \frac{{\rm{\Delta }}C}{{C}_{1}}+\displaystyle \frac{{\rm{\Delta }}P}{{P}_{1}}+\displaystyle \frac{{\rm{\Delta }}C}{{C}_{1}}\times \displaystyle \frac{{\rm{\Delta }}P}{{P}_{1}}.\end{eqnarray} \tag{ 6 }$$

The future climate simulations of the CMIP5 ensemble documented that the occurrences of precipitation extremes are projected to clearly increase over most regions across the global lands at the end of this century (Sillmann et al 2013, Chen et al 2014, Donat et al 2016). However, how many days of precipitation extremes could be avoided if the surface air temperature increase was limited to 1.5 °C rather than 2.0 °C in the future still remains an open issue. Thus, figure 1 first shows changes in the probability of the occurrence of precipitation extremes (R95p) at the end of the 21st century under different future warmer climates with respect to the present day using the CESM low-warming simulations. The precipitation extremes are projected to increase in the future if the probability rates shown in the panels are larger than 1.0. Our estimation firstly shows that both the mean and coefficient of variation of precipitation are reported to be uniformly increased across the global lands in response to the future warming. Meanwhile, the probabilities of occurrence of precipitation extremes are projected to increase across the world under different future warmer climates, including the 1.5 °C warmer climate, the 2.0 °C warmer climate, and the highest emission scenario (hereafter called 'RCP8.5 warmer climate'), when compared with that at the present day.

Zoom In Zoom Out Reset image size

However, after taking a more detailed look at the regional changes of precipitation extremes, a strong regional dependence for the extreme change can be observed across the world, with a relatively larger increase occurring over the Southern Hemisphere while a relatively smaller increase occurs over the mid-high latitudes of the Northern Hemisphere, which is substantially independent of the different future warmer climates. This is especially true for the regions of Africa, in which the occurrences of precipitation extremes show the largest increase in the future. Over the regions of western and eastern Africa, the probability of occurrence of precipitation extremes would be approximately 5 times greater than that in the current climate for both the future 1.5 °C and 2.0 °C warmer climates. Under the RCP8.5 warmer climate, the likelihood of extremes would be 8 times greater over these regions. For Sahara, the regional aggregate occurrence was also projected to increase by at least 3 times under the future warmer climates, despite the probabilities over some parts of this region showing a decrease in response to the warming (figure S3). The warming also yields significant increases of precipitation extremes over some parts of Asia, including East Asia, South Asia, and Southeast Asia, in which the probability increases by approximately 4 times in the future compared to that in the present day. Additionally, the precipitation extremes over the regions of Central and South America are also projected to show profound increases, with such extremes being approximately 2 times greater in the future than under the current climate. Regions of the mid-high latitudes of the Northern Hemisphere, including North Europe, North Asia, Mediterranean, Alaska, Greenland, and North America, would experience a relatively weak increase of precipitation extremes under the future warmer climates, but with the occurrences at least doubled. Regarding the aggregate average over the global lands, the precipitation extremes would increase by approximately 2 times under the 1.5 °C and 2.0 °C warmer climates and 3 times more under the RCP8.5 warmer climate in the future compared with that at the current level.

Our above assessments show significant increases of precipitation extremes across the global lands in response to the future warmings. However, how many days of precipitation extremes can be avoided if the surface air temperature increase was limited to 1.5 °C rather than 2.0 °C remains an open question to date. Thus, the avoided increases of precipitation extreme days in response to the reduced warming are assessed and displayed in figure S4. In comparison to the 2.0 °C warmer climate, the decrease in warming by 0.5 °C in the 1.5 °C warmer future will help to avoid obvious increases in the number of days of precipitation extremes across the world, especially for the regions of Greenland, western North America, North Asia, and North Europe, in which the number of extreme days can be avoided by at least 5%. Over the regions of Asia (including Central, East, South, and Southeast Asia, as well as Tibet) with highly concentrated populations, the extreme days would be reduced by approximately 3% in response to the 0.5 °C reduced warming. In contrast, the number of extreme days over the regions of the Sahara, the Mediterranean, South Africa, Central America, and the Amazon present a weak decrease in response to the additional 0.5 °C warming in the future. Considering the global lands as whole, the 0.5 °C lower warming in the 1.5 °C warmer future would help to avoid an increase of precipitation extremes by approximately 3.6% when compared to that under the 2.0 °C warmer climate.

If compared to the results with no emission limitations under the RCP8.5 warmer climate, the number of extreme days would be drastically lowered across the global lands when the surface air temperature increase was limited to 1.5 °C (figure S4). The largest avoided increase occurs over the region of Alaska, with a reduction of 44.8% in response to the lower future warming. The smallest avoided increase occurs over the region of Central America (by 18.1%), in which the number of extreme days show a decrease under the 2.0 °C future warming with respect to that under the 1.5 °C future warming. For most of the remaining regions, the lowered warming from the RCP8.5 warmer climate to the 1.5 °C warmer climate can help to avoid an increase of precipitation extremes by approximately 30%. On a global scale, approximately 34.8% of the extreme days can be reduced in response to less warming. Thus, actions of limiting warming are encouraged and urgently needed for the mitigation of climate change.

Risks associated with climate change have been observed to significantly increase in the past decades and would be further exacerbated under the future warming scenarios (Field et al 2014). For example, the increasing heat waves in the future will result in an increase of the aggregate population exposure in African cities by 20–52 times at the end of this century when compared to the current level (Rohat et al 2019). The 0.5 °C additional warming in the 2.0 °C warmer future will lead to approximately 17% more populations exposed to extreme droughts over China in comparison to the 1.5 °C warmer climate (Chen and Sun 2019 ). In the following, we project future exposure changes to the precipitation extremes throughout the world under different warming scenarios and assess the roles of the warming limitation protocols.

Currently, exposure is the highest over regions of East Asia and South Asia, in which the highest populations live (figure S5). These areas are followed by the regions of the Mediterranean and eastern North America, which have a relatively high population that is also accompanied by a high exposure. Lower exposures can be observed over the mid-high latitudes, including North Asia, Alaska, and Greenland, as well as the desert regions of the Sahara and Australia. With the future warming, the region of Africa, except for the Sahara, exhibits the largest net increase in exposure, followed by the regions of South Asia, the coast of Australia, and North America (figure 2). In contrast, a weak decrease occurs over some parts of Europe and the Sahara. Similar changes are clear for the different future warming scenarios, but with a relatively stronger increase under the RCP8.5 warmer climate and a relatively weaker increase under the 1.5 °C /2.0 °C warmer levels. In comparison to the 1.5 °C warmer future, the 0.5 °C additional warming under the 2.0 °C warmer climate also induces the largest net increase in exposure over most regions of Africa. The other regions, except for the Sahara, the Amazon, and Central America, would also experience a positive response of exposure to the additional warming in the future.

Zoom In Zoom Out Reset image size

We further compared the exposure changes over rural and urban regions (figure S6). It is clear that the increase of total exposure is mainly contributed by the rapid increase of the exposure over urban regions across the world under future warming scenarios, while exposures over rural regions are projected to decrease or show a weak increase. The drastic increase of exposure over urban regions is closely related to the rapid urbanization process in the future that does not depend on the SSPs, though the SSP5 used here is the highest urbanization pathway (Jones and O'Neill 2016). The cities located in West and East Africa would suffer from the largest increase of exposure to precipitation extremes in the future; these cities showed an increase of more than 70 times the current level by the end of this century under the 1.5 °C/2.0 °C warmer climate and more than 120 times under the RCP8.5 warmer climate. Consequently, significant increases of total exposure are also projected over these regions, and they are estimated to increase by more than 10 times under the 1.5 °C/2.0 °C warmer climate and more than 20 times under the RCP8.5 warmer climate. In the regions of Asia with highly concentrated populations, especially for East Asia, the urban exposure also shows a rapid increase in the future, with an increase of more than 4 times under the future warmer climates, even though the populations are projected to decrease over this region by the end of this century (figure S5). In most cases, we found the future exposure increase to be predominantly driven by an increase from cities rather than rural regions, in which the exposure is projected to decrease in the future. However, an interesting finding was that the rural exposure over some regions with low concentrations of people, such as Alaska, Greenland, and the Sahara, showed an increase in the future, which is mainly contributed to the increase of precipitation extremes that are further discussed in the next section. Considering the global land as a whole, the aggregate exposure is estimated to increase by more than 3 times what it is currently by the end of this century under the 1.5 °C/2.0 °C warmer future and more than 4 times under the RCP8.5 warmer climate. This increase is mainly contributed to the increased exposure over urban regions, which showed an increase of 6–10 times under the future warming scenarios, while the exposure over rural regions showed a decrease in the future.

Also noteworthy, the SSP scenarios exhibit less variability in exposure outcomes. Under the future warming scenarios, the total exposure to precipitation extremes presents a profound increase in the future, not depending on the SSPs (figure S7). However, a relatively greater increase can be generally found in the SSP3 scenario (with the high population growth) for most regions of Asia, Africa, and South America, while a relatively greater increase can be found in the SSP5 scenario (with the rapid urbanization process) for the regions of North America and Australia. At the global scale, the exposure also tends to significantly increase for all SSPs under different future warming climates, but with a greater increase occurring in the SSP3 scenario. Thus, the increase of exposure to the precipitation extremes shows much robustness at both the global and regional level.

The exposure increase is obvious in the future worldwide. However, the extent to which the exposure can be avoided if the surface air temperature increase is limited to 1.5 °C rather than the other high warming levels is unknown. The avoided exposures are thus calculated from the 2.0 °C and the RCP8.5 warmer climates to the 1.5 °C warmer climate, as shown in figure 3. Clearly, the largest amount of exposure that can be avoided can be seen over North Asia in response to the 0.5 °C lower warming, with a reduction of approximately 6.2%. In most cases, the lowered warming can help people be exposed less to the precipitation extremes in the future. However, there are several regions, including the Sahara, the Amazon, the Mediterranean, South Africa, and Central America, in which the exposure shows a weak decrease in response to the 0.5 °C of additional warming. Additionally, we also evaluated the avoided exposure if limiting the warming to 1.5 °C versus the RCP8.5 warmer climate. The reduction of exposure can be clearly seen across the world. The largest amount of exposure able to be avoided, approximately 55.7%, can be found over the region of Alaska in response to the reduction in warming, while the smallest avoided exposure is shown over the region of East Asia with a value of 10.6%. Our further estimation shows that the global aggregate exposure would profoundly decrease if the temperature increase was limited to 1.5 °C, with the exposure reduced by approximately 2.3% and 30.1% under the 2.0 °C and RCP8.5 warmer climates, respectively. A similar response of exposure to the additional warmings in the future can be found across the SSPs at both the global and regional scale, suggesting a weak role of SSPs on exposure change (figure S8).

Zoom In Zoom Out Reset image size

To explore the relative importance of climate and population change, we estimated the change in exposure and its components for different regions under the RCP8.5-SSP5 scenario (figure 4). At the regional scale, climate change appears more influential than the population in driving increased future exposure, as climate change yields a far greater exposure than the population change does for regions across the world, especially for regions located over Asia, Europe, and South America. For example, the increase in exposure to precipitation extremes of approximately 14.5 billion person-days over the Mediterranean region is mainly contributed by climate change, which accounts for an increase of approximately 9.6 billion person-days, while only 2.4 billion person-days and 2.5 billion person-days are contributed from the population and their interaction effects, respectively. In the other words, with the 109% increase of exposure by the end of this century, approximately 72% of the increase is sourced from the effect of climate change. Over East Asia, negative effects are clear from the population and interaction, mainly due to the projected decrease of population in the future over this region. However, climate change exerts a greater effect on exposure that offsets the negative role of the population and ultimately results in the exposure increasing by approximately 101 billion person-days by the end of this century. In contrast, for the regions of Africa, a relatively larger increase can be found from the interaction effect than from climate and population changes, suggesting the interaction effect is more significant. Over regions of North America, population change shows a slightly higher role on the exposure increase than the other two components, because SSP5 represents the high population growth scenario over North America (Jones et al 2018).

Zoom In Zoom Out Reset image size

Globally, the exposure to precipitation extremes is projected to increase by approximately 1144 billion person-days by the end of this century under the RCP8.5-SSP5 scenario, which is equivalent to an increase of approximately 474% with respect to the current time. This increase is largely contributed by climate change, which can account for approximately 332% of the increase in exposure, followed by the interaction effect that accounts for approximately 120%. In comparison, the population shows a relatively small contribution to the increase of exposure in the future, accounting for approximately 22% of the increase. Briefly, climate change is very likely to lead to increased exposure to precipitation extremes in the future, which is true globally and regionally. Thus, the positive effects of limiting the future warming are clear, and actions to promote such limiting should be encouraged regardless of the political and socioeconomic goals of a country.