Why spatial verification? SAL Ensemble SAL MesoVICT data Results Conclusions Using the SAL technique for ensemble forecast verification Sabine Radanovics Until March 2017: LSCE/ESTIMR, now: CNRM/GMME/MICADO May 10, 2017 1 / 19
Why spatial verification? SAL Ensemble SAL MesoVICT data Results Conclusions Why was spatial verification invented? 2 / 19
Why spatial verification? SAL Ensemble SAL MesoVICT data Results Conclusions Why was spatial verification invented? COSMO2 VERA CLEPS 50 48 prec 40 2007062100 30 20 46 10 0 44 5 10 15 5 10 15 5 10 15 Forecaster Paul 2 / 19
Why spatial verification? SAL Ensemble SAL MesoVICT data Results Conclusions Why was spatial verification invented? COSMO2 VERA CLEPS 50 48 prec 40 2007062100 30 20 46 10 0 44 5 10 15 5 10 15 5 10 15 Forecaster Paul 2 / 19
Why spatial verification? SAL Ensemble SAL MesoVICT data Results Conclusions Why was spatial verification invented? COSMO2 VERA CLEPS 50 48 prec 40 2007062100 30 20 46 10 0 44 5 10 15 5 10 15 5 10 15 Model developer Cate 3 / 19
Why spatial verification? SAL Ensemble SAL MesoVICT data Results Conclusions Why was spatial verification invented? COSMO2 VERA CLEPS 50 48 prec 40 2007062100 30 20 46 10 0 44 5 10 15 5 10 15 5 10 15 RMSE = 452.02 RMSE = 352.03 Model developer Cate 3 / 19
Why spatial verification? SAL Ensemble SAL MesoVICT data Results Conclusions Why was spatial verification invented? COSMO2 VERA CLEPS 50 48 prec 40 2007062100 30 20 46 10 0 44 5 10 15 5 10 15 5 10 15 RMSE = 452.02 RMSE = 352.03 → Need to develop verification that gives credit Model developer to the attributes Paul appreciates Cate 3 / 19
Why spatial verification? SAL Ensemble SAL MesoVICT data Results Conclusions SAL - Structure, Amplitude, Location (Wernli et al., 2008) Amplitude component A S = 0 A = 0.667 L = 0 obs fc 16 precip 12 6 5 4 y 3 8 2 1 0 4 4 8 12 16 4 8 12 16 x Doubled precipitation → relative Amplitude error A 4 / 19
Why spatial verification? SAL Ensemble SAL MesoVICT data Results Conclusions SAL - Structure, Amplitude, Location Location component L S = 0 A = 0 L = 0.442 obs fc 16 precip 12 3 2 y 8 1 0 4 4 8 12 16 4 8 12 16 x Shifted feature → location error of center of mass relative to domain size L 5 / 19
Why spatial verification? SAL Ensemble SAL MesoVICT data Results Conclusions SAL - Structure, Amplitude, Location Structure component S S = 0.4 A = 0 L = 0 obs fc 16 precip 12 3 2 y 8 1 0 4 4 8 12 16 4 8 12 16 x Flatten feature → structure error in terms of scaled feature volumes S 6 / 19
Recommend
More recommend