Visual Summaries of Ensemble Regional Projections: A Distillation Problem Carol McSweeney1, Tamara Janes1, Richard Jones1, Chris Gordon2

1 Met Office Hadley Centre, Exeter, UK 2Centre for Climate Research Singapore

CORDEX meeting, Stockholm, May 2016

Overview

•  Visualisations are a important means of communicating climate information

•  In a number of recent projects we have found it necessary to produce visual summaries of ensemble projections of varying complexity

•  Here we show some of the examples we’ve produced and discuss the reasoning behind the choices made

Summary Plots as a Distillation Problem

Drawing out key messages:

•  Making appropriate choices can enhance the dissemination of key messages

•  Some choices can be misleading!

Making sense of multi-method information: In more ‘complex’ cases, we can use summary plots as a means of showing how the types of information relate to one another.

•  Ranges from more than one GCM ensemble (MME/PPE)

•  How does range across a sub-set compare with full ensemble, what is impact of downscaling?

Example 1: Extremes indices in CMIP5 projections (See Fig 21-7/8 in IPCC AR5 WGII, Ch 21)

Baseline = 10% by definition

CMIP5 range

Max

Min

75%ile

Median 25%ile

Key  Messages:    

East  and  Sahelian  Africa:  signal  for  increases  in  heavy  rainfall  is  consistent  across  ensemble  members.      

 

Southern  and  West  Africa:  most  of  the  available  scenarios  indicate  increases,  but  those  increases  are  smaller,  and  for  some  members  of  CMIP5  heavy  rainfall  events  

decrease  overall.    There  are  also  some  regional  excepEons  in  Southern  Africa.    

Example 2: Downscaled CMIP5 projections for Singapore National projections

Example for mean wet-season (NDJ) rainfall (mm per day) 9 CMIP5 models sub-selected to span projections of future mean climate but avoid least realistic models. Extend method to show:

•  Full CMIP5 GCM range, 9-member GCM subset.

•  Impact of downscaling to 12km on projection range

Also added an indication of natural variability (1-sigma) based on 20-year periods in pi-control runs for some CMIP5 models

Full  CMIP5  range  with  selected  members  

Key  messages:    

Downscaled  rainfall  projec1ons  of  wet  season  rainfall  across  southeast  Asia,  including  Singapore,  are  uncertain  in  nature.  Projec1ons  for  rainfall  within  both  the  9  downscaled  experiments  and  the  CMIP5  ensemble  span  both  posi1ve  and  nega1ve  changes  by  the  2080s  and  therefore  represent  a  high  degree  of  uncertainty  in  the  

ensemble  median  changes  seen  in  the  central  map.        

Projected  changes  for  the  near  term  (2020s)  remain  small  compared  with  natural  variability  in  most  models.  

 

Example 3: PARCC West Africa

Project made use of existing downscaled projections for Africa:

•  5 members of HadCM3 PPE, downscaled with PRECIS, SRESA1B (Buontempo et al., 2015)

Showing CMIP5 range for wider uncertainties, but also as a reference point.

An alternative composition for these plots in order to:

•  Depict a range of plausible outcomes – show maps for the models indicating ‘high’ and ‘low’ future rainfall.

•  Show natural variability more explicitly - shown using an example time series.

Example 3: Parcc West Africa Key  messages:  

 Rainfall  projec1ons  for  Chad  contain  a  high  level  of  uncertainty,  both  within  the  downscaled  experiments  depicted  here  and  the  wider  CMIP5  ensemble.      Results  from  the  five  RCM  

experiments  suggest  increases  in  precipita1on  within  the  southern  tropical  regions  of  Chad,  however  the  wider  CMIP5  results  indicate  both  posi1ve  and  nega1ve  changes  in  precipita1on  

for  the  country.    

Summary : #1

Motivation for this paper/talk is to share our experiences/thoughts and generate discussion, not to be prescriptive!

Have aimed to: •  Represent  informa1on  faithfully  •  Draw  out  key  messages  •  Show  where  key  messages  come  from  (e.g.  Through  mul1ple  ranges  of  

projec1ons)  

Figures designed to be intuitive and informative but are still complex – the figures have information at a number of different levels. May require explanation, time to absorb - this is OK.

Summary #2

Choices in construction of figures are bespoke to specific projects – different data, different emphases, different user groups.

 Some  common  themes:  •   Placing  examples  of  spa1al  paOerns  with  spa1ally  reduced  data  to  show  uncertainty  seems  to  be  intui1ve  to  interpret  

•   Showing  some  indica1on  of  natural  variability  useful  for  interpreta1on  •   AOempted  to  use  intui1ve  colour  schemes,  signals  to  reader.  

Illustrate process of arriving at a range of projections, and how ranges relate to one another

•   important  for  interpre1ng  e.g.  different  studies  based  on  different  subsets,  or  comparing  results  with  IPCC/CMIP5  ranges.    

Questions and Answers Further Information from: carol.mcsweeney@metoffice.gov.uk

References: McSweeney, CF, Janes, T., Jones, RG and Gordon C. Summarising Ensemble Regional Projections. In preparation. Example 1: Hewitson,B et al 2014: Chapter 21: Regional Context. In: Climate Change 2014 :Impacts, Adaptation, and Vulnerability. Part B: Regional Aspects. Contribution of Working Group II to the Fifth Assessment Report of the IPCC. Example 2: Jones, R., G Redmond, T. Janes, C. McSweeney, WK Cheong, S. Sahany, S.-Y Lim, M.E.Hassim, R.Rahmat, S.Y Lee, C. Gordon, C. Marzin and E. Buonomo. Chapter 5: Climate Change projections. In: Singapore 2nd National Climate Change Study – Phase 1. See also: Marzin, C et al 2015: Singapore’s Second National Climate Change Study – Phase 1. Web link: http://ccrs.weather.gov.sg/publications-second-National-Climate-Change-Study-Science-Reports Example 3: Janes, T., Jones, R. and Hartley, A. 2015. Projections of change in ecosystem services under climate change. UNEP-WCMC Technical Report. See also: Buontempo, C., Mathison, C., Williams, K., Wang, C. and McSweeney, C. (2014) An ensemble climate projection for Africa. Climate Dynamics DOI 10.1007/s00382-014-2286-2 Jones R., Hartley A., McSweeney C., Mathison C. and Buontempo C. 2012. Deriving high resolution climate data for West Africa for the period 1950-2100. UNEP-WCMC technical report