China’s growing human displacement risk caused by floods under 1.5 °C and 2.0 °C global warming and beyond

Human displacement is one of the most pressing global issues, and China has the world’s largest population affected by floods. Yet, the spatial and temporal variations of China’s Human Displacement Risk (HDR) caused by floods remain unclear. Here, we investigated China’s HDR caused by extreme floods under different global warming levels, including 1.5 °C, 2.0 °C and beyond. We developed an approach to estimate human displacement caused by floods in China. Based on this method, our findings indicate that China’s HDR will increase by ∼10.7 (∼11.0) times under 1.5 °C (2.0 °C) warming, and each 0.5 °C warming will increase HDR by 3 million on average. These great increases are mainly driven by climate change rather than population variations. Our results also reveal that the relationship between human displacement and increasing percentage of flood protection levels follows an exponential function. Additionally, we found that increasing China’s current flood protection standard by ∼46% (1.5 °C and 2.0 °C) and ∼59% (4.5 °C) would reduce future HDR to the historical period level. This study provides valuable insights into China’s HDR, which can aid in adaptive flood risk management amid the trend of shifting to a warmer and more extreme climate.


Introduction
Human displacement is a sort of environmental refugees (Hoffmann et al 2020) and is one of the most concerned issues in the world (Black et al 2011).It can have severe consequences on ecosystems, public health, and human well-being, leading to social unrest and economic fallout (Paprotny et al 2018, Willner et al 2018b).The United Nations and the Federation of Red Cross have both recognized the significance of this issue.The American Red Cross has donated over $3.8 million since 2011 to aid displaced populations.Many factors can lead to human displacement, such as extreme flood events, wars, conflicts, extreme heat and pandemic (Winsemius et al 2016, McLeman 2019, Niva et al 2021, Schutte et al 2021).However, floods are the primary driver of human displacement globally, with over 140 million people being displaced since 2008 (much more than wars/conflicts) (Kam et al 2021), as the global population continues to move towards rivers (Mård et al 2018, Dryden et al 2021).Therefore, understanding how human displacement caused by extreme floods will change under the 1.5 • C and 2 • C Paris Agreement warming levels is of great interest and importance globally.
Studies on extreme flood induced human displacement remain limited.The methods used include statistical approaches, physical models, and empirical methods (Kakinuma et al 2020, Kam et al 2021).The flood hazard data used include discharge, water levels and flooding area from gauges, numerical models (Hirabayashi et al 2013, Dottori et al 2018) and satellite observations (Smith et al 2019, Tellman et al 2021).Some research investigated population exposure to extreme floods (Alfieri et al 2017, Willner et al 2018a), and historical population data were used for future periods without considering future population variations (Hirabayashi et al 2013, Alfieri et al 2017, Willner et al 2018a).However, population exposure is not equivalent to human displacement (Kakinuma et al 2020).Some studies investigated the benefits of adaptation to future extreme flood risk based on economic loss (Ward et  Here, we examined the risk of human displacement in China caused by extreme floods under global warming levels of 1.5 • C, 2.0 • C, and beyond.This analysis includes an assessment of spatial and temporal variations, as well as the potential benefits of adaptation measures, such as increasing flood protection standards to mitigate the impact of intensifying floods.To estimate the risk, we employed a large-scale hydrological and hydraulic coupled model, along with a human displacement estimation approach developed specifically for this study.We used different Representative Concentration Pathways (RCP) and Shared Socioeconomic Pathways (SSP) data to represent future climate and population variations, including RCP2.6-SSP1, RCP6.0-SSP3, and RCP8.5-SSP5.This study is unique in that it provides valuable insights into how China's human displacement risk may change under different global warming scenarios.Our findings are informative for adaptive flood risk management and can help support long-term social and economic development.

The human displacement estimation approach
Human displacement estimation considered the flood inundation fraction, population, inundation water level, and water level threshold that could lead to human displacement.The human displacement was estimated based on equation ( 1) where D is means the human displacement number; i means the ith gird; j means the jth day of the kth year; m means the number of the days of a year; POP means the population in the gird; frac means the simulated daily flood inundation fraction in the gird, and frac = 0 when water level (W depth ) is less than or equal to the human displacement water level threshold (W thres ); max means the maximum human displacement in a year; total human displacement (TDis) in the year k in a region with a total grid number of n was the sum of the maximum human displacement in each grid.The human displacement water level threshold equals to the water level corresponding to flood protection standards in this study (see details in the supporting information).
To attribute changes in displacement risk, we used two scenarios.The first scenario considered future population and climate changes, represented by the RCP-SSP scenario.The second scenario held the future population constant at the level of the last year of the baseline period  and only considered future climate change, represented by the RCP-2015 scenario.The first scenario represents the displacement risk caused by both climate change and population variations, while the second scenario represents the displacement risk caused by climate change only.The difference between the first and second scenarios represents the displacement risk caused by population variations.The attribution was based on equation (2): C pop > 0) (see results); at the sub-national scale, when C pop < 0, Ratio will be negative (i.e.Ratio < 1), indicating climate change has a positive effect on human displacement increase and population changes has a negative effect on human displacement increase.

The adaptation benefit analysis
When increasing the flood protection standard, the human displacement risk would decrease, which represents the benefit of adaptation.Human displacement and increasing percentage in the flood protection standard are fitted based on an exponential function, because we found the relationship between human displacement and flood protection standard follows an exponential function (see results and supporting information): where D is represents the human displacement number; Percentage i represents the ith increasing percentage in the current flood protection standard to adapt to intensifying floods; a and b are the parameters of the exponential function (exp).The required improvement percentage (I per ) in the flood protection levels to reduce future human displacement risk to the historical period level was calculated based on equation ( 4) where Dis His means the annual average human displacement number in the historical period.The human displacement number after considering Iper was estimated based on the hydrological and hydraulic coupled model and the human displacement estimation approach; the effectiveness of Iper was validated by comparing the estimated human displacement number with Dis His .

The datasets and model
In  et al (2022b).The WEB-DHM-SG model was calibrated based on discharge from hydrological gauges in China.The model could simulate discharge variations well with Nash-Sutcliffe Efficiency being above 0.7 in most regions.More information on the calibrated model parameters, calibration and validation approaches, and model performance in the calibration and validation can be found in the study by Qi et al (2022a).The model was also validated using satellite observed water occurrence and the documented human displacement caused by floods from the International Displacement Monitoring Center (IDMC) (www.internal-displacement.org/database/ displacement-data).In the baseline period, the average of documented human displacement is 5.9 million, and the average of estimation is 6.1 million (relative bias = 3.2%) indicating good performance (see details in the supporting information).More information about the data used and methodology can be found in the supporting information.

Human displacement in the whole of China
In the near future (before 2050), human displacement is expected to increase dramatically, with the relative contribution from population growing steadily (see figure 1).Differences between the RCP-SSP and RCP-2015 scenarios are expected to widen in the near future, indicating that increasing population would exacerbate the rising human displacement risk to some degree (up to 42%) (see figures 1(d)-(f)).However, climate change is expected to contribute more than population variations to changes in human displacement risk (at least 58%) (see figures 1(d)-(f)).The increasing displacement risk is due to substantial increases in flood inundation area (see figure 1(b)), induced by considerably intensifying precipitation extremes (see figure 1(c)).By the year 2050, human displacement is expected to range from 49 to 94 million, considering both RCPs and SSPs, while a much smaller range (32-56 million) is expected when population is fixed at the same level as in 2015.
In the distant future (post-2050), human displacement is expected to vary significantly, with a decreasing relative contribution from population (see figure 1).In the RCP2.6-SSP1scenario, human displacement is projected to decrease dramatically after 2060, with a maximum of approximately 63.7 million displaced individuals.This decrease can be attributed to changes in population (figure 1(d)), as flood inundation area (figure 1(b)) and extreme precipitation (figure 1(c)) are not expected to decrease.Under the RCP6.0-SSP3scenario, human displacement is projected to first decrease and then quickly increase after 2075 due to changes in flood inundation area and extreme precipitation (see figures 1(b) and (c)).The maximum human displacement number is expected to reach approximately 66.9 million by the end of the century, with population variations having a slightly reduced influence before stabilizing (figure 1(e)).In the RCP8.5-SSP5scenario, human displacement is expected to decrease in the last few decades due to a considerable reduction in population contribution (figure 1(f)), with a maximum of 67.8 million displaced individuals.
The risk of human displacement is expected to steadily increase when warming levels are at or below 3.5 • C, with each 0.5 • C increase resulting in an average increase of 3 million in human displacement risk (see figure S4 in the supporting information).This increase is largely due to climate change.Under 1.5 • C (2.0 • C) warming, population and climate changes are expected to contribute approximately 40.9% (39.9%) and 59.1% (60.1%) of the increases, respectively.Human displacement risk is projected to be at least 10.7 (11.0) times higher than the baseline period under 1.5 • C (2.0 • C) warming, but only 5.9 (6.2) times higher when considering climate change alone.Under 4.5 • C warming, these numbers increase to approximately 12.4 and 10.2, respectively.

Spatial distribution of human displacement
Human displacement in South China and the North China Plain is expected to be relatively higher than in other regions during both the baseline and future periods (see figures 2(a)-(d) and 3(a)-(d)), partly due to the relatively higher levels of extreme precipitation in these regions (see figure S5 in the supporting information).Northwest China is expected to have fewer human displacement cases compared to other regions.The relative changes in human displacement are expected to be larger in central China and southwest China under 1.5 • C, 2.0 • C, and 4.5 • C warming scenarios (see figures 3(e)-(g)).
The changes in human displacement risk induced by population variations are expected to be relatively larger in South China (see figures 2(e)-(g)) compared to other regions.Under 4.5 • C warming, the number of displaced individuals is expected to decrease in several provinces, such as Heilongjiang, Jilin, Liaoning and Shandong, due to population decreases under the SSP scenarios compared to 2015 (figures 2(j) and (g)).
In most provinces, the increases in human displacement caused by climate change are expected to be greater than those induced by population variations under 1.5 • C, 2.0 • C, and 4.5 • C warming scenarios (with a ratio <1 in figures 2(h)-(j)).
Human displacement in the Yangtze River is expected to be the largest (see figure 4(a)), accounting for 47.4%, 45.9%, and 48.3% of China's total under 1.5 • C, 2.0 • C, and 4.5 • C warming, respectively.The lower percentage of human displacement under 2.0 • C warming compared to 1.5 • C warming may be due to the combined effects of climate and population changes and their spatial distribution.Human displacement in the Yangtze, Pearl, Hai, and Huai Rivers is expected to be even greater than human displacement in the entire China during the baseline period (see figure 4(a)) in future warming scenarios.In comparison, human displacement in the Yellow River, Northeast China, and Southeast River is expected to be less, while human displacement in Northwest China and Southwest China is expected to be much less.
The increase in human displacement is mainly due to the projected increase in extreme precipitation under future warming climates (see figure 4(b)).The relatively lower human displacement in Northwest China is due to the much lower average extreme precipitation in this region compared to other regions.Although extreme precipitation is projected to decrease slightly in the Pearl River under 1.5 • C and 2.0 • C warming compared to the baseline period, the decrease is expected to be minor (see figure 4(c)).The combined effects of climate change and population variations are expected to result in an increase in human displacement in the Pearl River.

Adaptation
Improving flood protection levels is expected to dramatically reduce human displacement (resulting in adaptation benefits), with the reductions following an exponential function (see figures 5(a)-(c)).In order to maintain human displacement at the same level as the baseline period, the future flood protection standard should be 46%, 46%, and 59% higher than the baseline period under 1.5 • C, 2.0 • C, and 4.5 • C warming, respectively, according to the fitted exponential functions.Based on the corresponding improvement percentages, our model simulations validate that human displacement uncertainty is about 1.6%, −1.1%, and 0.4% compared to the baseline period under 1.5 • C, 2.0 • C, and 4.5 • C warming, respectively.These results suggest that the fitted exponential functions are appropriate for estimating the flood protection level improvement requirements.
In general, the required improvement percentage in flood protection standards is expected to increase from 1.5 • C to 4.5 • C warming though with fluctuations.The improvement percentage ranges from 44% to 59% (see figure 5(d)).The relative bias (RB) between estimates using the fitted exponential functions and our model simulations is relatively small (|RB| ⩽ 1.6% in figure 5(d)).The increasing percentage in flood protection levels is a little lower under 2.5 • C warming than under 1.5 • C warming.This may be because the heterogeneous spatial distributions of human displacement and flood variations, considering that the study region is large.In addition, the uncertainty under 1.5 • C and 2.5 • C warming is different (1.6% for 1.5 • C and 1.5% for 2.5 • C).These factors could influence the increasing percentage in flood protection levels, but the difference is relatively little (only 2%).The required flood protection level improvements are disproportionate to the increases in extreme precipitation (see figure 5(e)).The differences between the required flood protection improvement percentage and the percentage increase in extreme precipitation are large and decrease with increasing warming levels overall (see figure 5(e)).This result indicates that the required improvement percentage in flood protection standards cannot be solely determined by the increase percentage in precipitation intensity.

Discussion
We have developed an approach for estimating human displacement in China.Our approach differs from the one developed by Kam et al (2021), which added an additional one meter to the flood protection standard.Our tests showed that the resulting human displacement was largely underestimated when the one meter was added.This may be because the one meter was based on experience and is not applicable to China.Using our developed approach, we found that the relationship between human displacement and flood protection level increase percentage follows an exponential function.This relationship has not been reported before and provides new insights for flood risk management.We have also developed simple and easy-to-use equations for calculating the required improvement percentage in flood protection levels to reduce human displacement risk.Our approach has the potential to be used in other regions and can contribute to flood risk management efforts.
Human displacement caused by floods can have severe impacts on human health, especially during a pandemic.For example, in 2020, the COVID-19 pandemic coincided with an extreme flooding disaster in South China, which caused social unrest (Guo et al 2020) and potentially exacerbated the spread of the pandemic due to the displacement of individuals caused by the floods.Estimation of human displacement caused by floods is helpful for proper preparation for the concurrence of extreme flooding disasters and pandemics, and could be beneficial for the development of disease spread models (Bertozzi et al 2020).Although considering population migration caused by floods remains challenging in disease spread models, estimating the number of individuals displaced during extreme floods should be the first step in assessing their influence.
With the projected increase in human displacement under future warming climates, economic disruptions could be exacerbated.The displacement of individuals in one region can potentially influence the economy in other regions through international trade.Given the common economic cooperation between China, Europe, and the USA (Willner et al 2018b), the increasing human displacement risk in China caused by extreme floods could potentially impact the economies of both Europe and the USA.Therefore, the influence of growing human displacement risk is not unique to China and is likely to be widespread elsewhere in the world, highlighting the importance of global cooperation in extreme flood management and disaster reduction efforts.The propagation of this influence also underscores the importance of our study in estimating the spatial and temporal changes in human displacement risk in China under different future global warming scenarios.
One common belief is that the low-income population is more vulnerable to the effects of natural disasters (Jongman 2018) which compromises the United Nations' sustainable development goals.Action must be taken to reduce the risk of human displacement in order to achieve sustainable development.In addition to improving flood protection levels, non-structural measures could also be used to adapt to the increasing risk of displacement (Jongman 2018, Hong et al 2021).One such measure is limiting population growth in floodplains, particularly among low-income populations (Hong et al 2021).Land use planning and urban development should take into account the risk of human displacement, and appropriate policies should be implemented to reduce this risk in the future.
In this study, the FLOPROS flood protection standard was used, which is the same as the study by Kam et al (2021), facilitating the comparisons between our study and Kam et al (2021).There are other flood protection standard datasets (Wang et al 2021), and future studies are encouraged to analyse the influence of different flood protection standard datasets on human displacement estimation.Although the FLOPROS was used in our study, our methodology evaluations show the estimated human displacement is close to documented data, and therefore using the FLOPROS dataset is acceptable in our study.Our estimation did not take into account the influence of coastal and pluvial floods.Nonetheless, we demonstrated a significant increase in human displacement risk caused by river floods resulting from climate change ranging from 1.5 • C to 4.5 • C warming, and the substantial benefits of improving flood protection levels for adaptation.

Conclusion
We conducted a study on China's human displacement risk caused by extreme floods under 1.5 • C and 2.0 • C global warming and beyond for the first time.
Our study provides valuable information for flood risk management in China.The following conclusions are presented on the basis of this study.
First, we developed an approach to estimate human displacement in China caused by floods.We found that the relationship between human displacement and the increase percentage in flood protection levels follows an exponential function.
Second, China's human displacement risk would increase by ∼10.7 times under 1.5 • C warming, ∼11.0 times under 2.0 • C warming, ∼12.4 times under 4.5 • C warming, and each 0.5 • C warming would increase China's human displacement by 3 million on average.The increase is largely attributable to climate change rather than population variations.South China and North China Plain would have high human displacement risk, and the Yangtze River would account for nearly half of China's total.
Third, increasing China's current flood protection standard by ∼46% (1.5 • C and 2.0 • C) and ∼59% (4.5 • C) would reduce future human displacement risk to the same level as in the historical period.
al 2017, Burke et al 2018, Paprotny et al 2018, Vousdoukas et al 2020), population fatality (Jongman et al 2015) and reductions in population exposure (Willner et al 2018a).However, the benefits have not been studied based on reductions in human displacement.China is one of the countries most frequently affected by extreme floods globally (Internal Displacement Monitoring Centre 2018, Tellman et al 2021), with the largest population influenced by floods in the world (Jun and Melda 2020).In 2021, the Henan province in North China plain was hit by extreme flood disasters that had not been seen for decades.The floods affected about 14 million population, including one million population being displaced.Some studies examined flood induced human displacement risk on a global scale (Kakinuma et al 2020, Kam et al 2021); other study has looked into influenced population under different warming levels in the whole of China (Wu et al 2019).However, spatial and temporal variations of China's human displacement risk caused by floods remain unclear under future different global warming levels.Both climate change and population variations can influence future human displacement risk, especially in China where the population and climate are undergoing remarkably changes.However, the attribution of future flood induced human displacement risk to climate change and population variations in China has not yet been explored.Additionally, there has been no investigation into the adaptation requirements, such as improvements in the current flood protection standard, necessary to reduce future human displacement risk to the same level as in the historical period.

Figure 1 .
Figure 1.China's human displacement risk.(a), Human displacement number changes with time.(b), Flooding area (unit: km 2 ).(c), Mean annual maximum precipitation (unit: mm/day).(d)-(f), Attribution of the human displacement to population and climate variations.(g), The maximum human displacement.Results shown are the 30 year moving averages, and each 30 year moving window is indexed by its last year.Pop = Population.His = Historical period.The '2015 Pop' scenario represents that the population in the future is the same as in 2015.

Figure 2 .
Figure 2. Spatial distribution of human displacement risk and relative influence of climate and population in China.(a)-(d), human displacement risk.(e)-(g), human displacement under different RCP-SSP scenarios minus that when keeping population in 2015 (i.e.RCP-2015).(h)-(j), Ratio = population variation contribution/climate variation contribution.

Figure 3 .
Figure 3. Spatial distribution of human displacement and relative changes on the provincial level.(a)-(d), Baseline, 1.5 • C, 2.0 • C and 4.5 • C warming, respectively.(e)-(g), Relative changes compared to the baseline period.

Figure 4 .
Figure 4. Human displacement and extreme precipitation in the sub-regions of China in the baseline, 1.5 • C, 2.0 • C and 4.5 • C warming periods.(a), Human displacement.The percentage in a represents the ratio of the human displacement in the Yangzte River to the human displacement in entire China.(b), Average annual maximum precipitation (mm d −1 ).(c), Relative changes of extreme precipitation compared to the baseline period.Orange line in a represents the human displacement in the entire China in the baseline period.The error bar represents the standard deviation.Orange line in b represents the average extreme precipitation in the entire China in the baseline period.The locations of the regions can be found in the supporting information.

Figure 5 .
Figure 5. Human displacement changes when varying flood protection levels.(a)-(c), Human displacement (Y) under different flood protection level increasing percentages (X); (d), Increasing percentages in flood protection levels under the 1.5 • C-4.5 • C warming periods to keep the human displacement the same as the baseline period; (e), Differences between required improvement percentage in the flood protection levels and the intensifying percentage of extreme precipitation.See table S1 in the supporting information for the fitted exponential functions under other warming levels.The blue lines in (a)-(c) represent the human displacement in the baseline period.The black lines in (a)-(c) represent the fitted exponential functions.R 2 = Coefficient of determination.RB = Relative Bias.
Dis RCP-2015 represents the human displacement considering climate change and keeping the population the same as in 2015.The population under SSPs could be smaller than the population in 2015, leading to C pop < 0 and C Climate > 100.Ratio > 1 indicates that population changes have larger influence than climate change on human dis- placement increases; Ratio < 1 indicates that climate change has larger influence than population changes on human displacement increases; Ratio = 1 indicates their influence is the same.At the national scale, C pop is greater than zero under the difference scenarios (i.e.
the baseline period, daily precipitation and air tem- (Goldewijk et al 2017) and ISIMIP2b (2000- 2015)(Frieler et al 2017).Future population data under different SSPs were obtained from the study by Chen et al(2020).The present day flood protection standard was from the FLOPROS dataset (Scussolini et al 2016).