making certain the uncertainties Climate change making certain the uncertainties Group 3: Bok Wen Xuan, Anabella Fong, Goh Jie Sheng
outline: What is climate Climate projections: Energy balance model (EBM) General circulation model (GCM) Quantifying uncertainties Uncertainties neglected Conclusion
what is climate? Average weather experienced there Interested in changes in variability, and in particular in extremes “The distribution of weather”
climate projections How are climate projections made?
climate projections how are they made? Need certainty Climate system is very complex - need simplification of reality Two examples of models: Energy balance model (EBM) General circulation model (GCM)
Energy balance Model (EBM) Change in stored energy = input energy - output energy Have minimal resolutions Simple enough - implemented using a calculator or spreadsheet Crucial role in our understanding of the gross features of climate
Energy balance Model Limitations Largely phenomenological Can give only limited guidance about future climate Eg: Affects temperature Warmer climate Reflect visible light back into space Clouds Subtleties Absorbing and re- radiating infrared light from the surface of the Earth Atmospheric water vapour
General circulation model (GCM) Represents the physics of climate processes Can determine the net outcome of competing effects Potential to provide climate information at regional scales
General circulation model Limitations
General circulation model (GCM) Dynamic of the climate system defined by: dy/dt = f(y,x) y(t) : values of all prognostic variables at time t x(t) : a collection of forcing variables GCM attempts to solve this equation for a given initial configuration y(t0) and forcing scenario {x(t): t > t0). t0 : time in the past at which the climate system was relatively stable, such as 1850
General circulation model Limitations Many approximations are involved GCMs work with a subset of elements of y(t) - impossible to enumerate completely the state of the system at any time Not possible to represent all of the interrelationships between these elements Cannot be solved analytically, must use numerical methods - usually involve time and spatial discretion
General circulation model Spatial discretion HadCM3: Composed of 2 components: the atmospheric model HadAM3 and the ocean model HadOM3. Simulations use a 30 day calendar, where each month is 30 days. Produce simulations for periods of over a thousand years, showing little drift in its surface climate
General circulation model Spatial discretion HadGEM1: Developed in 2006 A significant scientific advancement from HadCM3 Provides a basis for further development of models, particularly involving enhanced resolution and full Earth System modelling
General circulation model Spatial discretion HadCM3 has coarse spatial resolution of the grid HadGEM1 has enhanced resolution Output on a grid can only approximate the spatial variation in y(t) Provide credible simulations; much of our detailed understanding of the climate system derives from this type of simulator
Quantifying uncertainty Structural uncertainty: simplification and approximation rule out a perfect prediction of prognostic variables source of uncertainty in future projections
Quantifying uncertainty Earth system quantities: Describe what the Earth is like at a selected moment Involved in inferring the prognostic variables Treated as simulator inputs Come in 3 ways: Values for Earth system components are fixed Historical values of forcing variables Future values of forcing variable
Quantifying uncertainty Correct values of parameters Do not have an analog in the climate system Not easy to operationalise in the context of an imperfect simulator
Quantifying uncertainty Input uncertainty - uncertainty about simulator inputs Parameter uncertainty - uncertainty about parameters Input Prognostic variables Observations Parameters
Uncertainties neglected Structural uncertainty: Often treated as negligible Historical input values are replaced by point estimates Parameter values were derived from a combination of physical reasoning and a very limited amount of tuning Incorporate as much physics as possible, to obtain what they consider the best possible point estimate of future climate - at the expense of quantifying uncertainties
Uncertainties neglected 2 improvements: #1. Have multiple simulator evaluations over different values for both the historical inputs and the simulator parameters #2. Use the distribution from a collection of different simulators, termed a multi-model ensemble
Uncertainties neglected Ensemble of simulations from HadCM3 Observed climate data UKCP09 Climate projections Marine and coastal projections Key parameters varied within ranges of uncertainty
Uncertainties neglected International Panel on Climate Change Fourth Assessment Report 23 GCMs Different structural assumptions Different evaluations
Uncertainties neglected #2 limitations Easy to treat the spread of values as uncertainty - happens when evaluations are displayed together Cone underestimates uncertainty Multi-model ensembles cannot be interpreted probabilistically in any conventional sense
conclusion Statisticians have an important role to play in representing and quantifying uncertainty about future climate Progress requires multidisciplinary collaboration
References Met Office, FitzRoy Road, Exeter, Devon, EX1 3PB, United Kingdom. (2016, April 15). Met Office climate prediction model: HadGEM1. Retrieved February 26, 2017, from http://www.metoffice.gov.uk/research/modelling-systems/unified-model/climate-models/hadgem1 HadCM3 GCM Model Information. (2011, May 16). Retrieved February 26, 2017, from http://www.ipcc-data.org/sres/hadcm3_info.html Using Climate Projections. (2015, August 21). Retrieved February 26, 2017, from http://ukclimateprojections.metoffice.gov.uk/21678