A case study of avoiding the heat-related mortality impacts of climate change under mitigation scenarios

Simon N. Gosling1 and Jason A. Lowe2

1 Walker Institute for Climate System Research, University of Reading 2 Met Office Hadley Centre

Outline

• Methods– The health models.

– Climate change scenarios.

• Business as usual impacts.• Avoided impacts.• Limitations and conclusions.

Methods

The health models

• Six city-specific empirical-statistical models for:

– Boston, Budapest, Dallas, Lisbon, London, Sydney.

• Described and validated in Gosling et al. (2007) and previously applied in Gosling et al. (2009).

• Assume no demographic changes with CC.

• Assumes no acclimatisation/adaptation

– Therefore impacts are indicative of a requirement for adaptation, rather than definitive values.

Gosling SN, McGregor GR, Páldy A (2007) Climate change and heat-related mortality in six cities Part 1: model construction and validation. International Journal of Biometeorology 51: 525-540.

Gosling SN, McGregor GR, Lowe JA (2009) Climate change and heat-related mortality in six cities Part 2: climate model evaluation and projected impacts from changes in the mean and variability of temperature with climate change. International Journal of Biometeorology 53: 31-51.

Climate change scenarios (1)

Scenario Pathway to peak

Date of peak Rate of decline in emissions

Emissions floor

A1B-2016-2-H A1B 2016 2% per year High

A1B-2016-4-L A1B 2016 4% per year Low

A1B-2016-5-L A1B 2016 5% per year Low

A1B-2030-2-H A1B 2030 2% per year High

A1B-2030-5-L A1B 2030 5% per year Low

Climate change scenarios (2)

• ClimGen uses patterns for a given GCM by fitting a regression, for each month, variable and GCM grid cell, between climate variable and global-mean temperature.

– Gives estimated change in climate per degree change in global-mean temperature ΔT, for a given GCM.

– Patterns obtained for 21 GCMs from IPCC AR4.

• Global-mean temperature change from MAGICC for each emissions scenario applied to ClimGen.

• ClimGen downscales to 0.5x0.5 degree resolution for each GCM pattern.

• Monthly means downscaled to daily temperature data using Mac-PDM.09 (Gosling and Arnell, 2010).

• Applied delta method (mean future – mean present) + observations.

Gosling SN, Arnell NW (2010) Simulating current global river runoff with a global hydrological model: model revisions, validation and sensitivity analysis. Hydrological Processes, in press.

Business as usual impacts

A1B impact & inter-study comparisons

Mortality Attributable to CC

McMichael et al. (2003) Sydney 2050s = 149% inc. (ECHAM4 Hi scenario);125 - 240%.

Dessai (2003) Lisbon 2080s = 59.5 - 173.1 (2xCO2 with 1 GCM & 1 RCM, 30% inc. in pop); 17-56.

Donaldson et al. (2001) UK 2080s = 350% inc. (Med-Hi scenario); 280 - 350%.

• Impacts broadly agree with previous assessments.

Avoided impacts

Absolute avoided impacts (2016-5-L)

GCM uncertainty

Time

• Magnitude of avoided impact varies considerably with GCM.– GCM uncertainty can be greater than difference between Δtime

Relative avoided impacts (2016-5-L)

• GCMs in agreement upon magnitude of relative avoided impacts.• Magnitude of avoided impact increases with time.• Mitigation reduces, but does eliminate impacts of climate change.

Comparing policies (ensemble mean)

• Benefits in 2030s are minor.

• Emissions-peaking year is a greater driver of avoided impact than emissions reduction rate.

• Up to 70% of A1B impacts could be avoided by 2080s.

Comparing policies (2080s distributions)

• Width of distribution is lower with 2016-peaking policies than 2030-peaking policies.

• Little differences across same emissions-reductions policies.

Comparing policies (policy vs. A1B)

• Width of distribution is lower with time into the future.

• Approximately, policy delays 2050 BaU impacts by 30 years.

Caveats, limitations and conclusions

Caveats and limitations

• Limitations of health models.– No adaptation. – No demographic changes.

• Pattern-scaling assumes that the relationship between global temperature change and local climate response is linear and invariant.

• Pattern-scaling not validated for emissions reductions.

• All 21 GCMs considered equally credible.

• Don’t consider relative probabilities of achieving policy scenarios.

• Delta method means frequency of extremes (e.g. heat waves) is same under CC as in present.

Conclusions

• Impacts under A1B are consistent with previous studies.

• Only one other study has considered potential benefits of mitigation policy for temperature-related mortality (Hayashi et al. 2010).

• Magnitude of benefits increase towards end of century.

• GCM uncertainty means absolute avoided impacts are considerably different across GCMs.

• Year of peak emissions is a greater driver of avoided impact than rate of emission reduction.

• Up to ~70% of impacts could be avoided, but not 100%.

Hayashi A et al. (2010) Evaluation of global warming impacts for different levels of stabilisation as a step toward determination of the long-term stabilisation target. Climatic Change 98: 87-112.

Thank you for your time