Analysis of Extremes in Climate Science Francis Zwiers Climate Research Division, Environment Canada. Photo: F. Zwiers.

Slides:



Advertisements
Similar presentations
C O M P U T A T I O N A L R E S E A R C H D I V I S I O N Application of Generalized Extreme Value theory to coupled general circulation models Michael.
Advertisements

Introduction to modelling extremes
Introduction to modelling extremes Marian Scott (with thanks to Clive Anderson, Trevor Hoey) NERC August 2009.
Spatial point patterns and Geostatistics an introduction
Comparing statistical downscaling methods: From simple to complex
Expert Team on Climate Change Detection and Indices (ETCCDI) started in 1999 jointly sponsored by CCl, CLIVAR and JCOMM.
Climate changes in Southern Africa; downscaling future (IPCC) projections Olivier Crespo Thanks to M. Tadross Climate Systems Analysis Group University.
Climate change detection and attribution methods Exploratory Workshop DADA, Buenos Aires, Oct 2012 Francis Zwiers, Pacific Climate Impacts Consortium,
EURANDOM & KNMI, May 2009 Analysis of extremes in a changing climate in support of informed decisions for adaptation
Literature Review Kathryn Westerman Oliver Smith Enrique Hernandez Megan Fowler.
RAMADDA for Big Climate Data Don Murray NOAA/ESRL/PSD and CU-CIRES Boulder/Denver Big Data Meetup - June 18, 2014.
Climate Change Impacts on the Water Cycle Emmanouil Anagnostou Department of Civil & Environmental Engineering Environmental Engineering Program UCONN.
Cost-effective dynamical downscaling: An illustration of downscaling CESM with the WRF model Jared H. Bowden and Saravanan Arunachalam 11 th Annual CMAS.
Mechanistic crop modelling and climate reanalysis Tom Osborne Crops and Climate Group Depts. of Meteorology & Agriculture University of Reading.
School of Civil, Environmental and Mining Engineering Life Impact | The University of Adelaide Wednesday, 4 th April 2012 Changes to sub-daily rainfall.
Detection of Human Influence on Extreme Precipitation 11 th IMSC, Edinburgh, July 2010 Seung-Ki Min 1, Xuebin Zhang 1, Francis Zwiers 1 & Gabi Hegerl.
Determining the Local Implications of Global Warming For Urban Precipitation and Flooding Clifford Mass and Eric Salathe, Richard Steed University of Washington.
Extremes ● An extreme value is an unusually large – or small – magnitude. ● Extreme value analysis (EVA) has as objective to quantify the stochastic behavior.
Uncertainty and Climate Change Dealing with uncertainty in climate change impacts Daniel J. Vimont Atmospheric and Oceanic Sciences Department Center for.
TOLERATE research meeting Progress by FMI: results & planned actions Kirsti Jylhä Thanks to material: Kimmo Ruosteenoja, Ari Venäläinen, Seppo.
Arctic Land Surface Hydrology: Moving Towards a Synthesis Global Datasets.
Alan F. Hamlet Dennis P. Lettenmaier Center for Science in the Earth System Climate Impacts Group and Department of Civil and Environmental Engineering.
Detection of anthropogenic climate change Gabi Hegerl, Nicholas School for the Environment and Earth Sciences, Duke University.
Extreme Value Analysis, August 15-19, Bayesian analysis of extremes in hydrology A powerful tool for knowledge integration and uncertainties assessment.
Assessment of Extreme Rainfall in the UK Marie Ekström
Explaining Changes in Extreme U.S. Climate Events Gerald A. Meehl Julie Arblaster, Claudia Tebaldi.
Bayesian Spatial Modeling of Extreme Precipitation Return Levels Daniel COOLEY, Douglas NYCHKA, and Philippe NAVEAU (2007, JASA)
Climate Forecasting Unit Prediction of climate extreme events at seasonal and decadal time scale Aida Pintó Biescas.
Extreme Value Analysis What is extreme value analysis?  Different statistical distributions that are used to more accurately describe the extremes of.
December 2002 Section 2 Past Changes in Climate. Global surface temperatures are rising Relative to average temperature.
10 IMSC, August 2007, Beijing Page 1 An assessment of global, regional and local record-breaking statistics in annual mean temperature Eduardo Zorita.
Physical science findings relevant to climate change adaptation Richard Jones, Met Office Science Fellow/Visiting Professor, School of Geography and Environment.
Kuala Lumpur, Malaysia, 8th-11th November 2012
Jana Sillmann Max Planck Institute for Meteorology, Hamburg
Where the Research Meets the Road: Climate Science, Uncertainties, and Knowledge Gaps First National Expert and Stakeholder Workshop on Water Infrastructure.
Global/European Analysis of Extremes - recent trends - Albert Klein Tank KNMI, the Netherlands 11 June 2002 acknowledgements: Lisa Alexander (Met Office,
Recent Advances in Climate Extremes Science AVOID 2 FCO-Roshydromet workshop, Moscow, 19 th March 2015 Simon Brown, Met Office Hadley Centre.
Some advanced methods in extreme value analysis Peter Guttorp NR and UW.
Photo: F. Zwiers Assessing Human Influence on Changes in Extremes Francis Zwiers, Climate Research Division, Environment Canada Acknowledgements – Slava.
Extreme values and risk Adam Butler Biomathematics & Statistics Scotland CCTC meeting, September 2007.
Components of the Global Climate Change Process IPCC AR4.
Statistical approach Statistical post-processing of LPJ output Analyse trends in global annual mean NPP based on outputs from 19 runs of the LPJ model.
Objective Data  The outlined square marks the area of the study arranged in most cases in a coarse 24X24 grid.  Data from the NASA Langley Research Center.
13 March 20074th C20C Workshop1 Interannual Variability of Atmospheric Circulation in C20C models Simon Grainger 1, Carsten Frederiksen 1 and Xiagou Zheng.
Climate Scenario and Uncertainties in the Caribbean Chen,Cassandra Rhoden,Albert Owino Anthony Chen,Cassandra Rhoden,Albert Owino Climate Studies Group.
Regional Climate Group 1 Department of Earth Sciences.
2.There are two fundamentally different approaches to this problem. One can try to fit a theoretical distribution, such as a GEV or a GP distribution,
Sea level Workshop – Paris 2006 Assessing the impact of long term trends in extreme sea levels on offshore and coastal installations Ralph Rayner Marine.
Climate Extremes PRECIS Workshop Tanzania Meteorological Agency, 29 th June – 3 rd July 2015.
WCRP Extremes Workshop Sept 2010 Detecting human influence on extreme daily temperature at regional scales Photo: F. Zwiers (Long-tailed Jaeger)
Montserrat Fuentes Statistics Department NCSU Research directions in climate change SAMSI workshop, September 14, 2009.
Chris Ferro Climate Analysis Group Department of Meteorology University of Reading Extremes in a Varied Climate 1.Significance of distributional changes.
The ENSEMBLES high- resolution gridded daily observed dataset Malcolm Haylock, Phil Jones, Climatic Research Unit, UK WP5.1 team: KNMI, MeteoSwiss, Oxford.
Indices versus Data Indices are information derived from data Indices are information derived from data Proxy for data Proxy for data More readily released.
Identifying natural hazards in climate databases Albert Klein Tank KNMI, the Netherlands 19 September 2002 acknowledgements: Lisa Alexander (Met Office,
Climatological Extremes 13 November 2002 Albert Klein Tank KNMI, the Netherlands acknowledgements: 37 ECA-participants (Europe & Mediterranean)
HYDROCARE Kick-Off Meeting 13/14 February, 2006, Potsdam, Germany HYDROCARE Actions 2.1Compilation of Meteorological Observations, 2.2Analysis of Variability.
Enabling Climate Impact Assessment in Wisconsin Chris Kucharik and Dan Vimont The Wisconsin Initiative on Climate Change Impacts (WICCI)
7. Air Quality Modeling Laboratory: individual processes Field: system observations Numerical Models: Enable description of complex, interacting, often.
Assessment of high-resolution simulations of precipitation and temperature characteristics over western Canada using WRF model Asong. Z.E
Climate Change, Texas, and the Limits of Confidence
Overview of Downscaling
Statistical Methods for Model Evaluation – Moving Beyond the Comparison of Matched Observations and Output for Model Grid Cells Kristen M. Foley1, Jenise.
Mike Staudenmaier NWS/WR/STID
Multimodel Ensemble Reconstruction of Drought over the Continental U.S
Detection of anthropogenic climate change
A Multimodel Drought Nowcast and Forecast Approach for the Continental U.S.  Dennis P. Lettenmaier Department of Civil and Environmental Engineering University.
Environmental Statistics
Multimodel Ensemble Reconstruction of Drought over the Continental U.S
Presentation transcript:

Analysis of Extremes in Climate Science Francis Zwiers Climate Research Division, Environment Canada. Photo: F. Zwiers

Space and time scales Simple indices Annual maxima Multiple maxima per year Incorporating spatial information One-off events Outline Photo: F. Zwiers

Very wide range of space and time scales Language used in climate circles not very precise –High impact (but not really extreme) –Exceedence of a relatively low threshold (e.g., 90 th percentile of daily precipitation amounts) –Rare events (long return period) –Unprecedented events (in the available record) Range from very small scale (tornadoes) to large scale (eg drought) Space and time scales

LocalContinental hours Space Time days month season years Regional Few observations per period (seasons to interannual) A single observation in the period of interest (multi-annual and longer) Process studies Many observations per season, many seasons “Extremes” likely to be conditioned by climate state in all cases

Simple indices Time series of annual counts or exceedences –E.g., number of exceedence above 90 th percentile Some studies use thresholds as high as 99.7 th percentile Coupled with simple trend analysis techniques or standard detection and attribution methods –Detected anthropogenic influence in observed surface temperature indices –Perfect and imperfect model studies of potential to detect anthropogenic influence in temperature and precipitation extremes Statistical issues include –“resolution” of observational data –adaptation of threshold to base period –use of simple analysis techniques that implicitly assume data are Gaussian

Indices approach is attractive for practical reasons - basis for ETCCDI strategy

Regional workshops –

Indices of temperature “extremes” Anthropogenic influence detected in indices of cold nights, warm nights, and cold days Christidis, et al 2005 Alexander, Zhang, et al 2005 DJF Cold nights JJA warm days

Some simple indices not so simple … Number of days per year in Canada with temperature above 99 th percentile Rate at which 90 th percentile is exceeded in simulated 60-year records (when threshold is estimated from first 30-years) 11% 10% Zhang, Hegerl, Zwiers, Kenyon, 2005

T max, T min, P 24-hour, etc Analyzed by fitting an extreme value distribution –Typically use the GEV distribution –Fitted by MLE or L-moments Analyses sometimes … –impose a “feasibility” constraint –include covariates –incorporate some spatial information Often used to –compare models and observations –compare present with future Annual extremes

Detection and attribution is an emerging application –include expected responses to external forcing as covariates –one approach is via Bayes Factors Main Assumptions –Observed process is weakly stationary –Annual sample large enough to justify use of EV distribution Some challenges –Data coverage –Scaling issue –How best to use spatial information –What to compare model output against –Are data being used efficiently? Annual extremes

Uneven availability in space and time Weak spatial dependence Spatial averages over grid boxes may not be good estimates of “grid box” quantities simulated by climate models Observational data rather messy Trend 5-day max pcp (data: Alexander et al. 2006)

20-yr 24-hr PCP extremes – current climate

Projected waiting time for current climate 20-yr 24-hr PCP event

Considering only annual extreme is probably not the best use of the available data resource –r-largest techniques (r > 1) –peaks-over-threshold approach (model exceedence process and exceedences) Some potential issues include –“clustering” –Cyclostationary rather than stationary nature of many observed series Has implications for both exceedence process and representation of exceedences Multiple extremes per year

Practice varies from –crude (e.g., simple averaging of GEV parameters over adjacent points) –to more sophisticated (e.g., Kriging of parameters or estimated quantiles) Fully generalized model would require simplifying assumptions about spatial dependence structure –Precipitation has rather complex spatial structure because it is conditioned by surface topography, atmospheric circulation, strength of moisture sources, etc. Using spatial information

How to deal with “outliers”? –Annual max daily pcp amount that is much larger than others, and occurs in 1885 –Recently observed value that lies well beyond range of previously observed values Both would heavily leverage extreme-value distributions (raising questions about the suitability of the statistical model) Recent events also beg the question – was this due to human interference in the climate system? Isolated, very extreme events

Fig 9.13a Surface temperature extremes Human influence: likelyHas likely affected temperature extremes May have increased the risk of extremely warm summer conditions regionally. Risk of extreme warm European summer in 1990s (1.6°C > mean): - natural forcing only - “all” forcing FAQ 9.1, Fig. 1

Summary Several methods available –Annual (or seasonal extremes), r-largest, POT, simple indices EV distributions can be fitted by moments, l-moments, mle –Latter also allows inclusion of covariates (e.g., time) Should evaluate –Feasibility –Stationarity assumption –Goodness-of-fit, etc Data limitations –quality, availability, continuity, etc –suitability for climate model assessment R-largest and POT methods use data more efficiently –Do need to be more careful about assumptions –Data may not be readily available for widespread use Formal climate change detection studies on extremes beginning to appear despite challenges … Also attempting to estimate FAR (Fraction of Attributable Risk) in the case of “one-of” events –How does one pose the question and avoid selection bias?

The End