Downscaling precipitation extremes Rob Wilby* & Chris Dawson * Climate Change Unit, Environment Agency Department of Computer Science, Loughborough.

Slides:



Advertisements
Similar presentations
Rachel T. Johnson Douglas C. Montgomery Bradley Jones
Advertisements

The new German project KLIWEX-MED: Changes in weather and climate extremes in the Mediterranean basin Andreas Paxian, University of Würzburg MedCLIVAR.
Introduction to modelling extremes
Assumptions underlying regression analysis
The Canadian Climate Impacts Scenarios (CCIS) Project is funded by the Climate Change Action Fund and provides climate change scenarios and related information.
Chapter 4: Basic Estimation Techniques
Comparing statistical downscaling methods: From simple to complex
Weighted Least Squares Regression Dose-Response Study for Rosuvastin in Japanese Patients with High Cholesterol "Randomized Dose-Response Study of Rosuvastin.
Hydrologic Statistics Reading: Chapter 11, Sections 12-1 and 12-2 of Applied Hydrology 04/04/2006.
11-1 Empirical Models Many problems in engineering and science involve exploring the relationships between two or more variables. Regression analysis.
Lesson 10: Linear Regression and Correlation
Forecasting Using the Simple Linear Regression Model and Correlation
Regression Analysis Module 3. Regression Regression is the attempt to explain the variation in a dependent variable using the variation in independent.
Use of regression analysis Regression analysis: –relation between dependent variable Y and one or more independent variables Xi Use of regression model.
Hydrologic Statistics
Potential Predictability and Extended Range Prediction of Monsoon ISO’s Prince K. Xavier Centre for Atmospheric and Oceanic Sciences Indian Institute of.
IPRC Symposium on Ocean Salinity and Global Water Cycle Recent Trends and Future Rainfall Changes in Hawaii Honolulu, Hawaii, Presentation by.
School of Civil, Environmental and Mining Engineering Life Impact | The University of Adelaide Wednesday, 4 th April 2012 Changes to sub-daily rainfall.
Downscaling and Uncertainty
9. SIMPLE LINEAR REGESSION AND CORRELATION
Time series analysis - lecture 5
Large-scale atmospheric circulation characteristics and their relations to local daily precipitation extremes in Hesse, central Germany Anahita Amiri Department.
Quantitative Business Analysis for Decision Making Simple Linear Regression.
© 2000 Prentice-Hall, Inc. Chap Forecasting Using the Simple Linear Regression Model and Correlation.
Robert Bornstein Jamie Favors, James Thomas, Allison Charland, Shawn Padrick Department of Meteorology and Climate Science San Jose State University 4.
11-1 Empirical Models Many problems in engineering and science involve exploring the relationships between two or more variables. Regression analysis.
Correlation & Regression
Correlation and Linear Regression
Hydrologic Statistics
Linear Regression and Correlation
Oceanography 569 Oceanographic Data Analysis Laboratory Kathie Kelly Applied Physics Laboratory 515 Ben Hall IR Bldg class web site: faculty.washington.edu/kellyapl/classes/ocean569_.
Multiple Regression Analysis
1 FORECASTING Regression Analysis Aslı Sencer Graduate Program in Business Information Systems.
Downscaling in time. Aim is to make a probabilistic description of weather for next season –How often is it likely to rain, when is the rainy season likely.
Water Resources: Hydroclimatic Forensics Through Reanalysis Bryson C. Bates Leader, Pathways to Adaptation Theme 24 th June 2008 Climate Adaptation.
Regional climate prediction comparisons via statistical upscaling and downscaling Peter Guttorp University of Washington Norwegian Computing Center
Geographic Information Science
Patapsco/Back River SWMM Model Part I - Hydrology Maryland Department of the Environment.
Regression Analysis Week 8 DIAGNOSTIC AND REMEDIAL MEASURES Residuals The main purpose examining residuals Diagnostic for Residuals Test involving residuals.
1 Regression Analysis The contents in this chapter are from Chapters of the textbook. The cntry15.sav data will be used. The data collected 15 countries’
Climate Scenario and Uncertainties in the Caribbean Chen,Cassandra Rhoden,Albert Owino Anthony Chen,Cassandra Rhoden,Albert Owino Climate Studies Group.
© 1999 Prentice-Hall, Inc. Chap Chapter Topics Component Factors of the Time-Series Model Smoothing of Data Series  Moving Averages  Exponential.
Statistical Summary ATM 305 – 12 November Review of Primary Statistics Mean Median Mode x i - scalar quantity N - number of observations Value at.
Stat 112 Notes 6 Today: –Chapter 4.1 (Introduction to Multiple Regression)
WCRP Extremes Workshop Sept 2010 Detecting human influence on extreme daily temperature at regional scales Photo: F. Zwiers (Long-tailed Jaeger)
STARDEX STAtistical and Regional dynamical Downscaling of EXtremes for European regions A project within the EC 5th Framework Programme EVK2-CT
Chris Ferro Climate Analysis Group Department of Meteorology University of Reading Extremes in a Varied Climate 1.Significance of distributional changes.
WINTER OROGRAPHIC- PRECIPITATION PATTERNS IN THE SIERRA NEVADA— CLIMATIC UNDERPINNINGS & HYDROLOGIC CONSEQUENCES Mike Dettinger 1, Kelly Redmond 2, & Dan.
1 Detection of discontinuities using an approach based on regression models and application to benchmark temperature by Lucie Vincent Climate Research.
Patapsco and Back River HSPF Watershed Model Part I - Hydrology Maryland Department of the Environment.
©The McGraw-Hill Companies, Inc. 2008McGraw-Hill/Irwin Linear Regression and Correlation Chapter 13.
Topics, Summer 2008 Day 1. Introduction Day 2. Samples and populations Day 3. Evaluating relationships Scatterplots and correlation Day 4. Regression and.
BUSINESS MATHEMATICS & STATISTICS. Module 6 Correlation ( Lecture 28-29) Line Fitting ( Lectures 30-31) Time Series and Exponential Smoothing ( Lectures.
Econometrics III Evgeniya Anatolievna Kolomak, Professor.
The simple linear regression model and parameter estimation
Why Model? Make predictions or forecasts where we don’t have data.
Regression and Correlation
Using teleconnections from the Pacific and Indian oceans for short-
Tushar Sinha Assistant Professor
Correlation and Simple Linear Regression
Basic Estimation Techniques
Change in Flood Risk across Canada under Changing Climate
A project within the EC 5th Framework Programme EVK2-CT
Correlation and Simple Linear Regression
Basic Estimation Techniques
Paul D. Sampson Peter Guttorp
Correlation and Simple Linear Regression
Stochastic Simulation and Frequency Analysis of the Concurrent Occurrences of Multi-site Extreme Rainfalls Prof. Ke-Sheng Cheng Department of Bioenvironmental.
Load forecasting Prepared by N.CHATHRU.
Linear Regression and Correlation
Presentation transcript:

Downscaling precipitation extremes Rob Wilby* & Chris Dawson * Climate Change Unit, Environment Agency Department of Computer Science, Loughborough

Motivation One of the most important – and yet least well-understood – consequences of future changes in climate may be alterations in regional hydrologic cycles and subsequent changes in the quantity and quality of regional water resources. Gleick (1987: 137)

A hierarchy of precipitation extremes Sub-daily - flash floods, urban drainage and water quality Daily - riverine flooding and tidal surges Multi-day - extensive floodplain inundation Single-season - surface water dominated systems Multi-season - groundwater dominated resource zones Annual - strategic water supply

Why consider multi-site/ multi-day precipitation totals….? ….winter 2000/01! ….recent trends

Experimental development of SDSM multi-site functionality and extremes Two approaches to daily precipitation extremes: –Compositing predictors associated with the largest daily precipitation totals across SEE and NWE. –A conditional re-sampling method for multi-site, multi-day precipitation downscaling. Demonstrated using multiple stations in Eastern England (EE) and the Scottish Borders (SB). Concluding remarks.

Variations in predictor strength Correlation between daily wet–day amounts at Eskdalemuir (55º 19 N, 3º 12 W) and mean sea level pressure (MSLP), and near surface specific humidity (QSUR) over NWE, 1961–1990.

Compositing daily extremes in SEE

Compositing daily extremes in NWE

Conditional re-sampling method Inverse normal transformation of area-average wet-day amounts across EE and SB. Obtain coefficients and standard error of model residuals from linear regression of transformed amounts versus regional predictor variables. Downscale area-average amounts and map to nearest neighbour wet-day amount/date in training set. Resample single site amounts contributing to the area average on the chosen date(s).

Many conditional variables (such as nonzero precipitation amounts and sunshine hours) are highly skewed. Therefore, a range of transformations for r t are available in SDSM, including exponential, fourth root, and inverse normal (version 2.3 only). Illustration of the inverse normal transformation Inverse normal transformation

Conditional variables, including nonzero precipitation amounts r t are simulated by where Z is a K 1 vector of standard Gaussian (i.e., normally distributed, with zero mean and unit variance) explanatory variables, is the coefficient matrix, and is an error term which is modelled stochastically (by assuming zero mean and variance equal to model standard error). Conditional variables

Downscaled (red) daily precipitation totals using NCEP predictors for compared with observations (blue). Example results for Kew Gardens, London

The location of the climate model grid boxes and stations used to evaluate the muliti-site downscaling of precipitation extremes using SDSM. Multi-site modelling

Frequency of predictor variable selection for individual stations included in EE area model; * included in SB area model

N -day annual max winter precipitation in EE Solid lines represent observations; other symbols are model syntheses (triangles = VAR ; squares = RND ; circles = DET model).

N -day annual max winter precipitation in SB

Solid lines represent observations; other symbols are model syntheses (triangles = VAR ; squares = RND ; circles = DET model). Annual maximum winter precipitation (20- and 60-day totals) at selected stations

Pairwise correlations of station daily precipitation series Inter-station correlations for all pairs of stations in EE and SB EE SB

Solid lines represent exponential decay functions fitted to the observations (filled circles); open symbols are model syntheses (triangles = VAR ; squares = RND ; circles = DET model). Correlation decay lengths for all pairs of stations in EE

Concluding remarks Compositing could isolate key predictors for extremes Regional and seasonal dependency of predictor set Practical advantages of re-sampling via area-averages Fully deterministic re-sampling was least successful How best to stratify data for re-sampling? More work needed on spatial aspects of extremes