1 Introduction and methods

Interdisciplinary climate change risk assessment is typically conducted either at a global scale, sometimes using a harmonised approach in terms of climate change and socioeconomic scenarios, and risk projection, but without exploring the implications for individual countries except for very large ones (for example Byers et al. 2018; Warren et al.2022; O’Neill et al. 2022). Nations, on the other hand, typically conduct their own independent interdisciplinary climate change risk assessments (for example Reidmiller et al. 2017; Betts and Brown 2021). Inevitably, different nations choose to use different approaches to selecting climate change and socio-economic scenarios, and use different models to projecting risk. Therefore, consistent sets of climate change risk assessments for different countries that are harmonised, utilising consistent sets of scenarios and methodological approaches, are in general, lacking. Without such studies, an objective comparison of levels of risk across countries at different levels of warming is challenging. In this study, such an approach is adopted for six countries.

This paper synthesises the harmonised assessment of projected future climate change risks in six countries provided by the Topical Collection, using a consistent set of climate change and socioeconomic scenarios and a single set of risk models. The countries were selected to span different levels of development, as well as different continents: Brazil, China, Egypt, Ethiopia, Ghana, and India. The analysis of risk is based on the projection of how hazard and exposure are projected to change under the scenarios considered. Vulnerability is assumed to remain constant except in the case of coastal flooding where adaptation is specifically modelled (Brown et al. 2021).

In the Topical Collection, many important potential future risks associated with alternative levels of global warming ranging from 1.5 to 4 °C above pre-industrial levels in the six countries were quantified for the year 2100 (Table S1). A case where human population remains constant at year 2000 levels, is compared with one where it evolves as in an established shared socioeconomic pathway (SSP2) (Riahi et al. 2017), with populations doubling or tripling in many African countries, but declining in China, leading to an overall global population increase from 7.2 billion in 2015 to 9.2 billion by 2100. These levels of warming span the goal set in the UNFCCC Paris Agreement of limiting warming to well below 2 °C and to pursue efforts to limit it to 1.5 °C, as well as higher levels of warming such as 3 °C which are consistent with the current national policies under the UNFCCC in terms of their Nationally Determined Contributions (NDCs) (Rogelj et al. 2016). Table S1 indicates the global climate change scenarios used, including the implications for the magnitude and rate of sea-level rise. The scenarios were generated using an established integrated assessment approach to create scenarios including or excluding different levels of climate change mitigation so as to reach approximately 1.5, 2, 3, and 4 °C warming by 2100 (Warren et al. 2021).

Risk indicators assessed in the Topical Collection include future exposure of agricultural land to drought, human exposure to drought (Price et al. 2022) and fluvial flooding (He et al. 2022), coastal flooding (Brown et al. 2021), changes in crop yields (Wang et al. 2021), and biodiversity loss (Price et al. 2020). The multi-sectoral findings from the Topical Collection are synthesised for each country, also including (for completeness) a brief summary of recent literature on heat stress, so the reader can gain a holistic picture of how climate change is expected to affect that country, and what the greatest benefits of limiting global warming (relative to pre-industrial levels) to 1.5 °C (Paris Agreement) as opposed to 3 °C (NDCs) would be.

The study facilitates a comparison of risk accrual with global warming in the six countries, and allows a comparison of levels of risk and relative increases in risk across countries and across metrics. However, it should be noted that adaptive capacity and vulnerability, which could not be included, also vary both nationally and locally. This means that a 5% change in exposure to a hazard in one country might have a more profound effect in one country than another. Similarly, there will be differential vulnerabilities to different types of hazards such that 5% change in one metric may have much larger or smaller implications than a 5% change in another.

To explore some aspects of differential vulnerability between countries, the regional economic implications of our simulated changes in crop yields and fluvial flooding were estimated within the Topical Collection (Wang et al. 2021; Yin et al. 2021) under both the constant population scenario and SSP2. In the latter case, estimates reflect changes in population, labour force, national GDP, and capital stock within SSP2 growth trajectories (2086–2115). Yin et al. (2021) estimate direct economic losses of fluvial flooding, assuming no adaptation, by linking the spatially explicit daily flood hazard data from He et al. (2022) with country and sector specific flood depth-damage functions. These values then input to an economic Input–Output model for the estimation of indirect losses. Indirect losses reflect business interruption losses of affected economic sectors, the spread of losses towards other initially non-affected sectors, and the costs of economic recovery.

The analysis encompasses scientific uncertainty in the regional spatial pattern of climate change projection at a given level of warming. It does so by utilising 21 alternative spatial patterns of regional climate change emerging from 21 alternative climate models. The percentage of additional risk accrual avoided with respect to the risk reference baseline period of 1961–1990, if global warming (above pre-industrial levels) is limited to a lower rather than a higher level, is calculated separately for each regional climate model pattern analysed. For each risk metric, the mean percentage avoided risk across these regional climate model patterns is derived. Applied methods and associated limitations, including the calculation of sea-level rise and associated coastal flooding, are described in the Supplementary Material (SM). Further details may be found in the sister publications within the Topical Collection.

Risk indicators were estimated for all scenarios, except that for the 1.5 and 2 °C levels of warming, analysis was in the case of fluvial flooding instead carried out for the scenarios 1b and 2b, that is < 1.5 °C and < 2 °C, in which there is 66% probability of warming remaining below these levels rather than 50% (see Warren et al. (2021) for further details). Throughout this synthesis, unless otherwise indicated all comparisons use the ‘exact’ 1.5 and 2 °C warming scenarios 1a and 2a, rather than 1b and 2b. Spatially explicit climate scenarios were produced for each country using pattern scaling as described in the Supplementary Information, including the scenarios 1a and 2a, which were produced by pattern scaling assuming these precise levels of warming in 2100.

Table S2 indicates the metrics used in this synthesis (see Supplementary Material for detail).

2 Results

The results allow comparison of risk accrual at different levels of synthesised in two ways:

  1. (i)

    A cross-sectoral comparison within each of the six countries, identifying which of the climate-related risks are of greatest concern (Sect. 2.1).

  2. (ii)

    A cross-country comparison within each of the six countries, assessing in which countries the risk in various sectors accrues the most as the climate warms, and which countries benefit the most from climate change mitigation (Sect. 2.2).

2.1 Risk accrual across sectors within each country

Figure 1 summarises our projections of multi-sectoral climate change related risk accrual for warming of 1.5–4 °C by 2100 by country, detailing projected changes in crop yields, agricultural land exposed to severe drought, and areas with increasing frequency of severe flooding. The numerical median and 10–90% ranges of various risk indicators are provided in Figures S1–S6 and Tables S3–8, with the latter also including projected changes in human exposure to flooding and drought for both population scenarios, and for some countries the land remaining suitable for growing some key cash crops.

Fig. 1
figure 1

Median growth in multi-sectoral additional climate change risk relative to 1961–1990, for global warming levels of 1.5–4 °C (above the 1850–1900 mean, a proxy for the pre-industrial baseline). Risk accrual is quantified using modelled percentage changes in risk indicators. Metrics shown are detailed in Table S2 and represent median changes as follows: loss of areal extent of climate refugia for plants, areal exposure of agricultural land to drought, areal exposure of land to decreasing Q100 return period, maize yield decline, rice yield decline, soybean yield decline, and wheat yield decline. Drought is defined as the probability that any given month will be classified as having a drought of magnitude − 1.5 (SPEI 12). Solid colour indicates risk levels at 4 °C global warming; dark grey, 3 °C warming; pale grey, 2 °C; and outlined, 1.5 °C

Full size image

Figure 2 shows the mean proportion of additional risk accrual avoided if warming is limited to 1.5 °C as compared with 3 °C or 2 °C by 2100. Metrics shown match those shown in Fig. 1 and represent avoided mean % changes in risk indicators relative to 1961–1990. In the majority of cases, there are substantial avoided risks although climate change is projected to benefit rice and soy yields in some countries. Limiting warming to 1.5 °C as compared with 3 °C is projected to deliver very large benefits in the avoidance of severe drought (SPEI12 − 1.5) exposure in all six countries, in terms of both agricultural land exposed, and people exposed. The reductions in exposure of agricultural land to severe drought are 61%, 43%, 18%, and 21% lower in Ethiopia, China, Ghana, and India, respectively. Human exposure to severe drought is reduced by 20–80% across countries in a constant population scenario in 2100 under SSP2. The findings for each country are detailed below and summarised in Table 1.

Fig. 2
figure 2

Percentage of additional risk accrual avoided if warming is limited to 1.5 °C as compared with 3 °C (a, c, e, g, i, k, and m) or 2 °C (b, d, f, h, j, l, and n) by 2100. Metrics shown match those shown in Fig. 1 and represent avoided mean % changes in risk indicators relative to 1961–1990 as follows: loss of areal extent of climate refugia for plants (a and b), areal exposure of agricultural land to drought (b and d), areal exposure of land to decreasing Q100 return period (c and e), maize yield decline (d and f), rice yield decline (e and g), soybean yield decline (k and l), and wheat yield decline (m and n). Drought is defined as the probability that any given month will be classified as having a drought of magnitude − 1.5 (SPEI 12)

Full size image
Table 1 Summary of projected changes in climate related risk indicators in the six countries. Risks are greater for higher levels of global warming, and are greatly reduced if warming is limited to 1.5°C rather than 2°C or 3°C. Limiting warming to 1.5°C rather than 3°C (or higher) reduces the adaptation required on the part of the human system, and allows biodiversity more time to adapt naturally. Because risks still remain at 1.5°C warming, adaptation will still be needed, but this will be much less challenging for 1.5°C warming than for higher levels. Exposure metrics refer to the constant population scenario.
Full size table

2.1.1 Brazil

Climate change projections for Brazil generally show a decrease in precipitation during the twenty-first century, with uncertainty in the magnitude and sign of the changes, and in small areas of the west precipitation may increase (He et al. 2022). Temperatures are projected to rise and heat stress vulnerable regions have been identified in metropolitan areas of Brazil (Lapola et al. 2019), with high heat stress conditions projected to increase regionally (Bitencourt et al. 2020). Population is projected to increase from 165 to 185 million between 2000 and 2100 in the SSP2 scenario. In other socioeconomic scenarios (SSPs), the projected population ranges from 142 to 264 million in 2100.

Despite the uncertainties in precipitation changes, there is high agreement in the literature that warming is expected to drive increases in climate variability in the form of both drought (Montenegro and Ragab 2012; Penalba and Rivera 2016; Marengo et al. 2017) and fluvial flooding (Hirabayashi et al. 2013; Alfieri et al. 2017). In this study, we also find that land exposure to severe drought increases with warming (Fig. 1) and also that human exposure to severe drought also increases with warming (Table S3a; Table S10). The probability of a given month being in an extreme drought (averaged across the country) triples in a 1.5 °C scenario, and is 5 and 7 times greater in 2 °C and 3 °C scenarios, respectively, exposing an additional 10% and 17% (respectively) of the population to drought under SSP2 assumptions in 2100 (Table S3a; Price et al. 2022). This probability reaches 50% at 4 °C warming, with some droughts exceeding 5 years in length (Price et al. 2022). The percentage of agricultural land exposed to droughts lasting for more than 1 year rises from 28 to 89% with 1.5 °C warming (Table S3b; Table S10; Price et al. 2022). With constant population, limiting warming to 1.5 °C avoids around 65% of the increase in human exposure to drought occurring with 3 °C warming (Table S10).

Despite widespread projected drought increases, projected flood risk also increases in many areas. We project that at 1.5 °C warming ~ 47% (median) of Brazil’s river basins experience a decrease in return period of the twentieth-century one-in-100-year flood (Q100), ~ 54% (median) with 4 °C warming (He et al. 2022). The non-linear response of hydrological systems to increased forcing can result in flood frequency first increasing and then decreasing with warming or vice versa (Fig. 1, leftmost bars of uppermost panel). Limiting warming to < 1.5 °C avoids approximately 90% (SSP2) or 84%(constant population) of the additional total direct and indirect economic losses that are projected to occur by 2100 with 3 °C warming due to fluvial flooding (Table S9a, b; Yin et al. 2021).

Additional people at risk from coastal flooding without additional protection (climate change component only and with no socio-economic change) in 2100 range between 1.1 and 1.3% of national population per annum (from < 1.5 °C 50th percentile to 4.0 °C 50th percentile). Annual sea flood damage costs are projected to increase from 1.1% (1.5 °C 50th percentile) to 1.4% (4 °C 50th percentile) in 2100 (Brown et al. 2021) under the SSP2 scenario and without additional adaptation (Table S3a). Defences can reduce these costs. However, sea-level rise and subsequent salinisation still induce a risk to ecosystems in deltaic areas (e.g. the Amazon) if freshwater volumes or active sedimentation are not greater than the sea-level rise. Slower rates of sea-level rise (aligning to a 1.5 °C scenario) allows greater time to respond, and thus a reduction in risks to ecosystems.

Projected declines in maize (6, 13%), wheat (5, 12%), soybeans (4, 9%), and rice (2, 5%) for (1.5 °C, 3 °C) warming, respectively, relative to 1961–1990 (Table S3b), are consistent with earlier declining yield projections (Costa et al. 2009; Margulis and Dubeux 2011). Yield changes were projected to cause increasing declines in wheat production and loss to sectoral value added (showing the change in the contribution of a given sector to overall GDP). For wheat, sectoral value added declined by 1.3% at 1.5 °C and 3.9% at 4 °C (Wang et al. 2021). For rice, declines in production and sectoral value added were projected between 1.5 °C (0.19%) and 3 °C (0.30%), with losses then reducing in severity at 4 °C (0.27%) reflecting changes in both yields and commodity prices (Wang et al. 2021). Related changes to welfare were found to become increasingly positive, projected to be 1.4% at 1.5 °C and 12.8% at 4 °C, suggesting limited effects of changing rice and wheat domestic and import prices, although if the model were to consider changes in other dominant crops like soybean, then given projections from other studies (e.g. Margulis and Dubeux 2011) benefits to welfare may well weaken or become negative.

With 3 °C warming, the projected area of land suitable for growing a key cash crop, coffee, declines by about 66% for both the arabica and robusta varieties (Table S3b), as compared with approximately a 33% decline at 1.5 °C warming. This is consistent with existing literature projecting declines in suitability of between 25 and 60% by 2050 (Ovalle-Rivera et al. 2015; Tavares et al. 2018). Similarly, the area suitable for growing sugar cane is also projected to decline (Table S3b), whilst the literature variously projects that sugar cane yields might either increase or decrease in Brazil depending on the role of changes in water use efficiency (Marin et al. 2013; Carvalho et al. 2015).

Similarly, changes in climate lead to large declines in plant refugia for biodiversity with 1.5 °C warming with only 33% projected to remain, but with additional declines by 2 °C, and further declines by 3 °C with only 3.6% refugia remaining (Figs. 1 and 2; Table S3b), with 80–100% of natural land exposed to droughts in excess of 1 year in length at 3 °C (Price et al. 2022). However, 33% of the plant refugia are projected to be preserved if global warming were limited to 1.5 °C rather than very few at 3 °C (Fig. 2a)..

2.1.2 China

Climate change projections for China show regionally differentiated mixed trends in precipitation during the twenty-first century, with uncertainty in the magnitude and sign of change. However, a consistent signal of projected increases in precipitation emerges in the North of the country, contrasting with parts of the South of the country where it is generally projected to decrease (He et al. 2022). Temperatures are projected to rise, with previous studies having attributed heat extremes in 2015 and 2016 to climate change (Imada et al. 2018; Sun et al. 2017) and projecting severe heat stress in China due to climate change by the 2040s (e.g. Lee & Min 2018). Despite uncertainties in precipitation changes, there is high agreement in the literature that warming is expected to drive increases in drought throughout the country, except the North (Wang and Chen 2014; Qin et al. 2014; Leng et al. 2015). Population is projected to decrease from 1.2 billion to 782 million between 2000 and 2100 in the SSP2 scenario. In other socioeconomic scenarios (SSPs), the projected population ranges from 1 billion to 580 million in 2100.

In this study, we find that regardless of changes in population, projected human exposure to drought increases greatly with warming (Table S4a). Regional warming increases evapotranspiration and can therefore offset the effects of small projected increases in precipitation. Across China, the probability of a given month being in severe meteorological drought triples in a 1.5 °C scenario, projecting droughts lasting more than2 years (Price et al. 2022). Severe drought probability quadruples at 2 °C. In a 4 °C scenario, the probability of drought in a given month has on average across the country risen to almost 50%, exposing approximately an additional 18% (range − 2 to + 51%) of the country’s population to drought and drought length may exceed 5 years. Under 3 °C warming and SSP2 population growth an additional 8% of the country’s population or 65 million (range 11.5–177.1 million) are projected exposed to severe drought than during 1961–1990 (Table S4a). Limiting warming to 1.5 °C is projected to avoid an estimated 66% of the increase in human exposure to drought occurring with 3 °C warming under a constant population scenario (Table S10), and 43% of the increase in exposure of agricultural land (Table S4b; Fig. 2). An important regional finding is that by 3 °C warming, 90% of areas with permanent snow and ice cover are projected to face severe droughts lasting longer than a year (Price et al. 2022).

The literature projects increases in flooding in China (Hirabayashi et al. 2013; Alfieri et al. 2017), although Vetter et al. (2017) highlight the discrepancies between hydrological models in the sign of projected changes in flow in major river basins. In this study, like Hirabayashi et al. (2013), we project large increases in flood risk in the North of the country, but in contrast to that study we find a mixed picture for the rest of the country. Overall, we project that with 1.5 °C warming, 66% (median) of China’s major river basins experience a decrease in return period of the one-in-100-year flood (Q100), 81% (median) experiencing a decrease in return period with 4 °C warming (He et al. 2022). Limiting warming to 1.5 °C reduces the area projected to undergo an increase in fluvial flooding by 10% as compared with 3 °C warming (Fig. 2). It avoids approximately 72% (SSP2) or 59% (constant population) of the additional total direct (and indirect) economic losses that could occur with 3 °C warming by 2100 due to fluvial flooding (Table S9a, b; Yin et al. 2021).

The number of people exposed to coastal flood risk increases with warming, even for low levels of warming. Additional people at risk from flooding without additional protection (SSP2) in 2100 range from 5.5 to 6.5% of national population per annum (from < 1.5 °C 50th percentile to 4.0 °C 50th percentile) (Table S4a). Without additional adaptation, annual flood costs are projected to be 15.4% (1.5 °C 50th percentile) to 19.3% (4 °C 50th percentile) per year in 2100 (SSP2; Brown et al. 2021). These flood costs are particularly high due to many megacities situated in low-lying coastal areas, including deltas (e.g. Shanghai (Du et al. 2020)). Impacts and costs of sea-level rise are exacerbated by high rates of land subsidence (Fang et al. 2020), particularly in deltaic areas. Coastal protection in China is already significant in length in low-lying areas, but would need to dramatically increase in highly exposed areas even with a modest rise in sea level under a scenario of aggressive climate change mitigation. This could decrease total flood costs to approximately 0.1% across all scenarios (Brown et al. 2021).

Projected climate changes lead to projected declines in the % of the country land area suitable for growing tea (from 15% in 1961–1990 to 9% with 1.5 °C warming and only 4% with 3 °C warming; Table S4b). Changes in climate were also projected to lead to declines in maize (20–24%) and wheat (9–12%) by 4 °C, accompanied by increases in the yield of soybeans (3–6%) and rice (5–6%) (Fig. 1; Table S4b). The literature contains mixed projections (both positive and negative) for the effects of climate change on wheat, rice, and maize in China owing to differing model parameterisations especially in relation to the treatment of CO2 fertilisation (Tao and Zhang 2013; Chen et al. 2013; ** nations (e.g., Brown et al. 2019; Dasgupta et al. 2009). Throughout all countries studied, non-climatic factors are also important in influencing impacts. These include the use of a portfolio of approaches to adaptation to reduce impacts including land claim, sediment availability which can be restricted through damming or removed through beach mining, and recognition and action into land subsidence which can exacerbate sea-level rise. Adaptation to sea-level rise must go hand-in-hand with socio-economic development, including livelihood protections (Brown et al. 2018) for which so many living and working on the coast depend. Due to the commitment to sea-level rise, many of these risks and projected levels of people at risk will happen regardless of future actions to mitigate against climate change. Potential impacts do increase from a < 1.5 to a 4 °C in a 2100 global warming scenario in all countries assessed. The lower the rate and magnitude of sea-level rise, the lower the risk of flooding and the potential for people to move away from the coast. Adaptation (e.g. building dikes) can reduce the potential number of people flooded, or those who may need to move away from at-risk locations. Furthermore, mitigating climate change reduces potential damage and adaptation costs. However, it is the longer locked-in centennial-scale implications of rising sea-levels that pose a greater threat to populations (Goodwin et al. 2018; Brown et al. 2018; Nicholls et al. 2018) justifying stringent climate change mitigation.

3 Discussion and conclusions

The necessity of selecting metrics to quantify risks creates inevitable limitations. For example the index used for drought, SPEI12, may be over-sensitive to increasing evapotranspiration in areas of high aridity, whilst the choice of Q100 (the return period of the twentieth-century 1-in-100-year flood) as an index for flood risk focuses the analysis on very large floods. Existing or planned flood defences or dams that may exist on rivers are not included, nor are potential additional defences against coastal flooding. Our statistically based projections of crop yields exclude potential positive effects of CO2 fertilisation and also negative effects of declines in crop nutrient content and increases in pests/disease. The assessment of climate change on biodiversity does not capture the potential additional effects of changes in climate variability, and for this reason we also quantified the exposure of natural land to drought (Price et al. 2022). Importantly, a full assessment of the potential role of adaptation in reducing climate change risks was beyond the scope of this study. For a further discussion of limitations, see Supplementary Material.

Despite these limitations, the study has identified important trends in climate change risk in the six countries studied. With a few exceptions, increasing warming leads to greater exposure to drought, fluvial and coastal flooding, and greater declines in biodiversity and crop yields. All countries except India are projected to experience large increases in both drought frequency and length, and hence, in terms of risk magnitude, the benefits of limiting warming to 1.5 °C rather than 3 °C are greatest in terms of reduced exposure to drought. All nations experience an increase in hazards due to sea-level rise, which is projected to rise globally, even if temperatures stabilise. However, stabilisation reduces the rate of sea-level rise, potentially reducing human migration from inundated areas as compared with the case that temperatures continue to increase globally.

This study was limited to exploration of two population scenarios: one in which population does not change and one following a middle-of-the-road scenario (SSP2). Human exposure to risks such as drought are considerably higher in Africa in the SSP2 scenario as compared with the constant population scenario, owing to the doubling or tripling of human populations in some African countries projected in this scenario by 2100. Future research might usefully explore the implications of other socioeconomic scenarios, in particular the SSP3 scenario in which populations in Egypt, Ethiopia, Ghana, and India reach 191, 287, and 256 million and 2.56 billion, respectively. Whilst increased levels of development (in some of the SSPs) could potentially reduce human vulnerability to some climate change risks, potentially by facilitating additional adaptation (for example in coastal cities; Brown et al. 2019), many risks remain large even at higher levels of development and some are unaffected or may even be exacerbated by development (for example risks to biodiversity).