Dynamical Downscaling of ECMWF Motivation Past work with Seasonal Forecasts: ECMWF cycle 29r2 over East Africa case studies over Eastern and QWeCI southern ECMWF Sys Africa domains 3 Ensemble data and RegCM set up Perfect boundary condition run ECMWF Gulilat Tefera Diro and Adrian Tompkins ENSEMBLE - forcing runs Earth System Physics Section Summary and The Abdus Salam ICTP Future work Thanks to RegCM Team — October 23, 2012 ILRI, Nairobi
Outline Motivation Past work with 1 Motivation ECMWF cycle 29r2 over East Africa ECMWF Sys Past work with ECMWF cycle 29r2 over East Africa 2 3 Ensemble data and RegCM set up Perfect ECMWF Sys 3 Ensemble data and RegCM set up 3 boundary condition run ECMWF ENSEMBLE Perfect boundary condition run 4 forcing runs Summary and Future work ECMWF ENSEMBLE forcing runs 5 Summary and Future work 6
Motivation Motivation Limited area models(eg. MM5, WRF , RegCM,...) have Past work with been used for downscaling short range weather ECMWF cycle 29r2 over East forecasts and/or climate change studies Africa ECMWF Sys 3 Ensemble data and RegCM set up For example, RegCM has been extensively validated on Perfect various domain for its ability to have an "added" value boundary condition run compared to the global climate models ECMWF ENSEMBLE There is a growing demand from impact models for a forcing runs detailed (localized) seasonal forecasts Summary and Future work Do regional climate models improve seasonal forecasts and able to reproduce the year to year fluctuation?
ECMWF and RegCM3 JJAS hindcast climatology Motivation RegCM - ERA-Interim JJAS total rainfall [mm] for 1991-2000 ERA-Interim JJAS total rainfall [mm] for 1991-2000 ECMWF ERAIN RegCM3-ERAIM Past work with 1800 1800 ECMWF cycle 20 1600 20 1600 29r2 over East 1400 1400 15 15 1200 Africa 1200 Latitude Latitude 1000 10 1000 10 800 800 ECMWF Sys 5 5 600 600 3 Ensemble 400 400 0 data and 0 200 200 RegCM set up -5 0 -5 0 25 30 35 40 45 50 55 25 30 35 40 45 50 55 Longitude Longitude Perfect ECMWF Ensemble mean JJAS hindcast (1991-2000) RegCM3 ENSEMBLE-mean JJAS hindcast (1991-2000) boundary ECMWF ensemble mean RegCM3 ensemble mean 1800 1800 20 20 condition run 1600 1600 1400 1400 15 15 ECMWF 1200 1200 Latitude Latitude 10 1000 10 ENSEMBLE 1000 800 800 forcing runs 5 5 600 600 400 400 0 0 Summary and 200 200 Future work -5 0 -5 25 30 35 40 45 50 55 25 30 35 40 45 50 55 Longitude Longitude TRMM JJAS total rainfall [mm] for 1998 − 2007 GPCP JJAS total rainfall [mm] for 1991-2000 1800 25 1800 20 1600 1600 20 GPCP TRMM 1400 1400 15 15 1200 1200 Latitude Latitude 1000 1000 10 10 800 800 5 5 600 600 400 400 0 0 200 200 −5 0 -5 0 25 30 35 40 45 50 55 25 30 35 40 45 50 55 Longitude Longitude
Perfect boundary run: Inter-annual variability Average over all five zones 2.5 Gauge 2 GPCP Motivation ECMWF - ERAIM 1.5 JJAS standard rainfall RegCM - ERAIM Past work with 1 ECMWF cycle 0.5 29r2 over East 0 Africa -0.5 ECMWF Sys -1 3 Ensemble -1.5 data and -2 RegCM set up -2.5 1990 1992 1994 1996 1998 2000 year Perfect boundary Table: Correlation of ERA-Interim and RegCM3 with observations condition run Gauge GPCP ECMWF ECMWF-ERA Interim 0.63 0.77 ENSEMBLE RegCM-ERA Int. forced 0.83 0.88 forcing runs Summary and 20 Future work JJAS total rainfall (mm) 1800 15 14 1600 I 1400 12 1200 IV Latitude 10 10 Latitude IIa 1000 III 800 8 Dry IIb 600 6 400 5 4 200 0 35 40 45 Longitude 0 -5 25 30 35 40 45 50 55 Longitude
ECMWF cycle 29r2 with RegCM: 1998 - 1997 Propagation of error from GCM to RCM Motivation Past work with GPCP JJAS total rainfall [mm] for 1998 -1997 RegCM - ERA-Interim forced JJAS total rainfall [mm] for 1998 - 1997 ECMWF cycle 600 600 29r2 over East 20 20 Africa 400 400 15 15 ECMWF Sys 200 200 3 Ensemble Latitude Latitude 10 10 data and 0 0 RegCM set up 5 -200 5 -200 Perfect boundary -400 -400 0 0 condition run -600 -600 -5 -5 ECMWF 25 30 35 40 45 50 55 25 30 35 40 45 50 55 Longitude Longitude ENSEMBLE forcing runs ECMWF ENSEMBLE mean JJAS total rainfall [mm]: 1998 - 1997 RegCM3 ENSEMBLE-mean JJAS total rainfall [mm]: 1998 - 1997 600 600 Summary and 20 20 Future work 400 400 15 15 200 200 Latitude Latitude 10 10 0 0 5 5 -200 -200 -400 -400 0 0 -600 -600 -5 -5 25 30 35 40 45 50 55 25 30 35 40 45 50 55 Longitude Longitude
Propagation of error: ENSO vs western Indian ocean SST teleconnection Motivation Past work with ERA−Interim JJAS Skin temperature ECMWF cycle 5 [1998 − 1997] 29r2 over East 60 ° N Africa ECMWF Sys 30 ° N Latitude 3 Ensemble data and 0 ° RegCM set up 0 Perfect 30 ° S boundary condition run 60 ° S ECMWF 150 ° W 100 ° W 50 ° W 0 ° 50 ° E 100 ° E 150 ° E ENSEMBLE −5 forcing runs Longitude Summary and Future work ECMWF Ensemble mean hindcast for JJAS 5 skin temperature [1998 − 1997] 60 ° N 30 ° N Latitude 0 ° 0 30 ° S 60 ° S 150 ° W 100 ° W 50 ° W 0 ° 50 ° E 100 ° E 150 ° E
Probabilistic verification: ROCS ROC Curve Area under ROC curve (ROCA) 1 Motivation Compares against a random Hit Rate Past work with ECMWF cycle forecast 29r2 over East Africa A skillful forecast → ROCA > 0 . 5 ECMWF Sys 3 Ensemble 0 or → 2* (ROC area -0.5) > 0 data and 0 1 False Alarm Rate RegCM set up Perfect Table: ROC Area (Area under the roc curve) over Ethiopia boundary condition run ROCs Dry Normal Wet ECMWF Gauge GPCP Gauge GPCP Gauge GPCP ENSEMBLE forcing runs ECMWF-ENSEMBLE 0.67 0.86 0.54 0.75 0.55 0.90 RegCM3 0.86 0.64 0.69 0.38 0.71 0.86 Summary and Future work RegCM ROCS for dry category of JJAS ECMWF ROCS for dry category of JJAS 1 1 20 20 0.8 0.8 0.6 0.6 15 15 0.4 0.4 Latitude 0.2 Latitude 0.2 10 10 0 0 -0.2 -0.2 5 5 -0.4 -0.4 -0.6 -0.6 0 0 -0.8 -0.8 -1 -1 -5 -5 25 30 35 40 45 50 55 25 30 35 40 45 50 55 Longitude Longitude
ECMWF cycle 29r2 with RegCM: RPSS � K 1 1 1 i = 1 ( CDF forc , i − CDF obs , i ) 2 ] Motivation RPS = K − 1 [ Obs Past work with ECMWF cycle RPS forc RPSS − D = 1 − 29r2 over East CDF ( RPS clim + RPS clim ∗ 1 M ) Africa Forc A skillful forecast → RPSS > 0 ECMWF Sys 3 Ensemble 0 0 data and Mean debiased RPSS with respect to climatology for ECMWF Mean debiased RPSS with respect to climatology for RegCM3 RegCM set up 1 1 20 20 Perfect 0.8 0.8 boundary 0.6 0.6 15 15 condition run 0.4 0.4 0.2 0.2 Latitude Latitude 10 10 ECMWF 0 0 -0.2 -0.2 ENSEMBLE 5 5 -0.4 -0.4 forcing runs -0.6 -0.6 0 0 -0.8 -0.8 Summary and -1 -1 Future work -5 -5 25 30 35 40 45 50 55 25 30 35 40 45 50 55 Longitude Longitude 1.0 RegCM3_wrt_GPCP ECMWF_wrt_GPCP RegCM3_wrt_gage ECMWF_wrt_gage 0.5 RPSS 0.0
ECMWF ensemble hindcasts from 33r1 Resolution : 1.125 0 × 1.125 0 in horizontal and 62L in Motivation Past work with vertical ECMWF cycle 29r2 over East Africa Hindcast period: 1991-2001 ECMWF Sys 3 Ensemble data and RegCM set up 9 member ensembles Perfect boundary condition run addressing forecast uncertainty ECMWF ENSEMBLE forcing runs uncertainty in initial condition: Perturbed initial conditions Summary and Future work model error: Perturbed physics Two start dates (May and November): here we used the Nov 1 st start for Malawi domain 6 month hindcasts starting from November 1 st
Model setup and experimental design Resolution : 25km (288x200) in horizontal and 18 levels Motivation in the vertical Past work with ECMWF cycle Convection scheme: Grell over land and Emanuel over 29r2 over East Africa the ocean ECMWF Sys Three experiments were carried out: 3 Ensemble data and ’Perfect’ boundary condition run RegCM set up with climate mode: continuous run Perfect boundary with seasonal forecast mode: initialize every year from condition run Nov 1 st ECMWF ENSEMBLE seasonal hindcasts boundary condition forcing runs Summary and For the ’perfect’ boundary simulation: Future work ERA-Interim re-analysis,OI-weekly SST simulation period: Jan 1990 to May 2002 For hindcasts simulation The 9 ECMWF ensemble members from 33r1 cycle are downscaled independently simulation period: Nov 1 st to May 1 st every year between 1991 and 2002
Model domain and Topography Motivation RegCM ECMWF Past work with ECMWF cycle 3000 5 29r2 over East 5 3000 Africa 0 0 ECMWF Sys 2500 2500 3 Ensemble -5 −5 data and RegCM set up 2000 2000 −10 -10 Perfect Latitude Latitude boundary -15 −15 condition run 1500 1500 -20 −20 ECMWF ENSEMBLE 1000 1000 forcing runs -25 −25 Summary and -30 −30 Future work 500 500 -35 −35 0 0 0 10 20 30 40 50 60 0 10 20 30 40 50 60 Longitude Longitude RegCM has a better realistic surface features due to its high resolution.
Recommend
More recommend
