Quantifying Uncertainty in Modelled Estimates of Future Extreme Precipitation Events CFCAS Project Progress Report The University of Western Ontario
Outline A - Reanalysis Data for UTRW C Quantifying GCM Uncertainty of Extreme Precipitation Events July 17, 29 2
A - Reanalysis Data for UTRW Objective To evaluate applicability of the NCEP- NCAR Global Reanalysis (NNGR) and Regional Reanalysis (NARR) hydroclimatic data for hydrologic modeling in Southwestern Ontario - Develop hydrologic model based on the reanalysis datasets - Comparison and analysis of the modeled output July 17, 29 3
Daily observed Precipitation and Temperature data from 11 stations for 198 25 STUDY AREA DATA Source: Weather Office, Environment Canada (http://www.climate.weatheroffice.ec.gc.ca/climatedata/c anada_e.html) NNGR Daily Prate and Temp data at 2.5 X 2.5 grid for 198 25 Source: Physical Science Division, NOAA (http://www.cdc.noaa.gov/data/reanalysis/reanalysis.shtml) NARR Daily Prate and Temp data at.3 X.3 grid for 198 25 Source: Data Access Integration (http://loki.ouranos.ca/dai/login e.php) 4 July 17, 29
NNGR NNGR VS. NARR Assimilated dataset using a state-of-the-art analysis/forecast system and past data since 1948 Frequency: 4 times daily at 6 hours interval, 8 climatic variables in several different coordinate systems, NARR A long-term, dynamically consistent, high-resolution, high-frequency, atmospheric and land surface hydrology dataset The grid resolution is 349 277 (.3, 32 km) resolution at the lowest latitude Improvements: (i) direct assimilation of radiances, (ii) use of additional sources of data, (iii) improved data processing, (iv) several Eta model developments in hydrological research, (v) assimilation of precipitation, (vi) land atmosphere coupling, and (vii) improvements to the Noah land surface model (land-model subcomponent of the RR) 5
Methodology Semi-distributed rainfall-runoff model based on the computational engine of HEC-HMS (USACE 26) Model set up: - 32 special units -21 river reaches - 3 reservoirs July 17, 29 6
Continuous Hydrologic Model (Cunderlik & Simonovic, 24) July 17, 29 7
Results 12 Stratford Observed NNGR NARR 14 Wroxeter Observed NNGR NARR 1 12 8 1 6 8 4 2 5 1 15 2 25 3 35 Days 6 4 2 5 1 15 2 25 3 35 Foldens Waterloo Wellington 12 Observed NNGR NARR Observed NNGR NARR 1 1 9 8 6 4 cum prec., mm 8 7 6 5 4 3 2 July 17, 29 2 1 8
Woodstock Observed NNGR NARR 25 2 15 Tmean, C 1 5-5 -1 1 2 3 4 5 6 7 8 9 1 11 12 Month GA Observed NNGR NARR 25. 2. 15. 1. 5.. -5. July 17, 29-1. 1 2 3 4 5 6 7 8 9 1 11 12 Month 9
Hydrologic Model Result NNGR NARR Location RMSE (cumec) r NMSE MAE (cumec) RB (%) RMSE (cumec) r NMSE MAE (cumec) RB (%) Byron 28.97.44 1.3 15.73 31 24.37.65.77 9.95-12 Ingersoll 4.2875.41 1.25 2.62 45 3.44.63.8 1.57-7 St. Marys 1.8.44.97 5.23 26 1.4.59.97 3.72-9 July 17, 29 1
7 25 6 2 5 Summer Prec. Vs Flow Byron Flow Flow 4 3 2 1 5 1 15 2 25 Observed Precipitaiton 7 6 5 4 3 2 1 5 1 15 2 25 NNGR Precipitaiton 7 6 Flow Flow 15 1 5 5 1 15 2 25 Observed Precipitaiton 25 2 15 1 5 5 1 15 2 25 NNGR Precipitaiton 25 2 St. Marys 5 Flow 4 3 Flow 15 1 2 1 5 5 1 15 2 25 NARR Precipitaiton July 17, 29 5 1 15 2 25 NARR Precipitaiton 11
Byron Ingersoll St. Marys July 17, 29 12 Absolute Error in Daily Discharge (May-Oct, 1-5)
Byron Ingersoll St. Marys July 17, 29 13 Variance of Daily Discharge (May-Oct, 1-5)
Conclusion The performance of NARR data set in simulating low flows is satisfactory NNGR data set with its coarse resolution is not appropriate for small watersheds such as the study area The reanalysis data needs careful investigation before their application with hydrologic models Should be used only with a complete explanation of the possible sources of difference between them and the observed data July 17, 29 14
C Quantifying GCM Uncertainty of Extreme Precipitation Events July 17, 29 15
Data Collection July 17, 29 16
Data Collection (Cont d) Weather Data - 25 stations - 1979-25 July 17, 29 17
July 17, 29 GCM Models SRES Scenarios Atmospheric Resolution Variables Lat Long Prec Tmean Tmax Tmean U1m V1m SLP SH CCSR/NIES, 1999 B21 5.6 5.6 Avg CGCM3T47, 25 A1B, B1, A2 3.75 3.75 CGCM3T63, 25 A1B, B1, A2 2.81 2.81 CSIROMK2b, 1997 B11 5.6 3.2 CSIROMK3.5, 21 A1B, A2, B1 1.875 1.875 ECHAM5AOM, 25 A1B, A2, B1 1.875 1.875 ECHO-G, 1999 A1B, A2, B1 3.9 3.9 GFDL-3, 1999 A21 3.75 2.25 GFDLCM2.1, 25 A1B, A2, B1 2. 2.5 GISS-AOM, 24 A1B, B1 3 4 GISS-ER, 24 A1B, A2, B1 4 5 HADCM3, 1997 A1B, A2, B1 2.5 3.75 HADGEM1, 24 A1B, A2 1.9 1.3 MIROC 3.2 HIRES, 24 A1B, B1 1.125 1.125 MIROC 3.2 MEDRES, 24 A1B, A2, B1 2.8 2.8 NCARCCSM3, 25 A1B, A2, B1 1.4 1.4 NCARPCM, 1998 A1B, A2 2.8 2.8
GCM Data Time Slices: - Refined Base period (1979-25) - 22 s (211-24) - 25 s (241-27) - 28 s (271-299) July 17, 29 19
GCM data for 8 variables at grid points for available scenarios years 211-239, 241-7 and 281-299 NCEP/NCAR reanalysis data for 8 variables at grid points for years 1979-25 Standardization & Normalization Normalized NCEP/ NCAR data at grid points Linear interpolation Interpolated GCM data for 8 variables at NARR grid points Standardization & Normalization Normalized GCM data at grid points Observed Precipitation data at station scale Principal component analysis First few principal components of each variable Coefficient Matrix for each variable Multiply by coefficient matrix First few principal components of each variable WG training 1979-1998 WG projection Parameters of WG Projected precipitation for future climate Comparison between the observed and the simulated values WG testing 1999-25 Predicted precipitation for 1999-25 2
Thank You! July 17, 29 21