Environmental Change Activity Store (CCAF) Call for recommendations on "Environmental Change; Variability and Extremes" .

Uploaded on:
Category: Animals / Pets
Basic Validation step: test of the winter's precision/summer ... what's more, perceptions in the north area for pre-winter. Precipitation are reproduced less precisely for summer and ...
Slide 1

Environmental Change Action Fund (CCAF) Call for proposition on "Environmental Change; Variability and Extremes" A first assessment of the quality and shortcomings of measurable downscaling techniques for reproducing extremes over different locales of eastern Canada Georges-É. Desrochers, Hydro-Québec Elaine Barrow & Philippe Gachon, CCIS Victoria Slonosky, Ouranos Taha Ouarda, INRS-ETE Tan-Danh Nguyen, McGill Diane Chaumont, Ouranos Marie-Claude Simard, Ouranos Massoud Hessami, INRS-ETE Mohammed Abul Kashem, INRS-ETE Alain Bourque, Ouranos René Roy, Hydro-Québec Guenther Pacher, Hydro-Québec Charles Lin, McGill Van TV Nguyen, McGill André St-Hilaire, INRS-ETE Bernard Bobée, INRS-ETE Jennifer Milton, Environment Canada Jeanna Goldstein, Environment Canada

Slide 2

Datasets Calibration Validation Tests to assess model execution (clarified fluctuation, RMSE, RRMSE, aptitude scores, extremes files) Method to recreate atmosphere situations: Use of the Empirical Statistical Downscaling Models Datasets: crude, institutionalized by means and standard deviation (NCEP, GCMs) Validation techniques: basic, cross, bootstrap Treatment of «unexplained» a portion of difference: expansion, randomization

Slide 3

SDSM - relapse based downscaling model with stochastic climate generator LARS-WG - stochastic climate generator occasional definitions the decision of change capacities ( fourth root, characteristic log, reverse ordinary ) the estimation of the contingent model parameters ( change swelling, inclination revision ) the picked timeframe and its length the nearby learning to characterize blend of indicators Empirical Statistical Downscaling (depends on exact connections between neighborhood scale predictands and provincial scale indicators; course sorts; amazing quality examination and so on ) SENSITIVITY TO:

Slide 4

Calibration step: SDSM structure. Distinctive variations of the exchange capacity variables ( different relapses, direct and non-straight, consolidated with stochastic climate generator) Seasonal definition: Monthly (*) Calibration period: 1961-1975 Threshold for Precipitation: 1mm/day (*) indicator variables should be precisely reproduced by GCMs (standardization lessens methodical predispositions in the mean and difference of GCMs indicators)

Slide 5

Quebec (Canada) Regions of Statistical Downscaling Robustness Study 1 2 4 3 6 5

Slide 6

Candidate indicator variables to shape ideal indicator set (Fourth root is picked as change capacity) Free air parameters, huge scale surface course parameters, dampness are prescribed for factual downscaling (Beckmann and Buishand, 2002; Hewitson, 2001; Huth, 1999; Huth et al., 2001; Huth, 2002; Trigo and Palutikof, 1999; Wilby et al., 2001; Wilby and Wigley, 2000).

Slide 7

Inflation parameter change for SDSM precipitation recreation Montreal-Dorval area 1976-1990 Autumn %tile-%tile plot of SDSM –WG downscaled precipitation versus perceptions Simple Validation step Inflation parameter = 3 Bias relationship parameter = 0.85 25 till 90%tile Average Inflation parameter = 12 Bias connection parameter = 0.85 25 5

Slide 8

Uncertainty connected with the utilization of GCM information Simple Validation venture till 90 %-tile Autumn %tile-%tile plots for Montreal-Dorval locale 1976-1990 of mimicked precipitation versus perceptions SDSM-Generator: CGCM1 information CGCM1 GHG+A1 SDSM-WG: NCEP information Estimation measurement SDSM WG/Gen GCM inf. 3 inf. 7 inf. 9 inf. 12 inf. 15 predisposition - 3.6 - 1.0/ - 1.3 - 0.8/ - 1.2 - 0.8/ - 1.1 - 0.6/ - 1.0 - 0.52/ - 0.8 RMSE 8.7 6.8/7.8 7.1/8.0 7.2/8.2 7.5/8.4 7.7/8 RMSE %til. 4.9 6.4/5.5 5.0/4.3/3.5 3.1/2.8 .2/1.2

Slide 9

Simple Validation step: test of the precision of the winter/summer greatest temperature reenacted arrangement for 1976-1990. Estimation of vulnerability connected with the utilization of GCMs Winter/Summer SDSM-WG SDSM-GEN CGCM1 GHG+A1 Bias (deg C) Montreal-Dorval - 0.5/1.1 3.8/ - 0.6 3.5/ - 1.9 Kuujjuarapic - 0.6/0.3 4.8/ - 4.3 8.2/2.1 Moosonee - 0.5/1.0 5.5/ - 3.1 7.3/0.3 Percentiles Bias (deg C) Montreal-Dorval - 0.5/1.1 3.8/ - 0.6 3.4/ - 1.9 Kuujjuarapic - 0.6/0.3 4.8/ - 4.3 8.2/2.0 Moosonee - 0.3/1.0 5.5/ - 3.1 7.2/0.3 Winter/Summer SDSM-WG SDSM-GEN CGCM1 GHG+A1 RMSE (deg C) Montreal-Dorval 2.9/2.4 9.8/5.9 8.0/4.9 Kuujjuarapic 3.7/4.5 10.6/9.9 11.8/7.4 Moosonee 3.5/3.7 11.4/8.4 11.3/6.3 Percentiles RMSE (deg C) Montreal-Dorval 0.8/1.2 3.9/1.3 6.1/2.1 Kuujjuarapic 0.8/1.4 5.1/4.4 8.8/4.1 Moosonee 0.4/1.3 5.8/3.2 8.4/2.6 Spring %tile-%tile plot of SDS models and GCM Tmax versus perceptions for Montreal area 1976-1990

Slide 10

Relevant files to the field of client interest (got from downscaled arrangement and contrasted and watched) Software STARDEX ( STatistical and Regional dynamical Downscaling of Extremes for European areas) Diagnostic Extremes Indices diagram: Agronomical important files for Spain (Winkler et al., 1997): the Julian date of first and last ice the principal occurance of Tmax > 25 deg C the recurrence of days with Tmax > 35deg C Water assets applicable files ( Goldstein and Milton, 2003): Max number of back to back dry days Max number of continuous wet days 90th percent. of rainday sums Greatest 5-day all out precipitation 90th Tmax percent http://www.cru.uea.ac.uk/cru/ventures/stardex/

Slide 11

Results, Recommendations and Conclusions: The progression of the SDSM acceptance should be executed with the distinctive arrangement of indicators and settings parameters with check via seasons or months SDSM-WG recreates sufficiently Tmax for all seasons. Nearby atmosphere (Tmax reproduction) is spoken to with higher precision for winter by SDSM-GEN than by CGCM1 GHG+A1 for the north of Quebec Estimation measurement reports less disparity values between Tmax downscaled mimicked information (SDSM-GEN) and perceptions in the north locale for harvest time Precipitation are recreated less precisely for summer and pre-winter SDS models might utilize yield of the diverse GCMs which constrained by various kind of the nursery gasses qualities to treat vulnerabilities SDSM reenacted situations should be dealt with separately. It is not conceivable to normal mimicked situations day by day STARDEX programming should be utilized to characterize extremes files - a measure of closeness amongst watched and reenacted time arrangement The main form of the Ouranos SDSM approval apparatus is made

Slide 12

Future Plans Definition of the exchange capacities variations for various Quebec districts and examination of their similitude Use of a stepwise different direct relapse system Use of the CGCM2 - SRES «A2», «B2» yield Further check of the capacity of the Statistical DownScaling models to catch extremes occasions Use of STARDEX programming to characterize e xtremes records Thank you to CCAF

View more...