V01 GEFS2P5/20 12 2015073000 GFSANL G2 HIST TMP H2 11.15 3.62 3.04 2.73 2.80 2.84 2.84 3.22 3.21 3.84 3.21 3.55 3.73 3.78 3.73 3.51 3.60 4.12 4.23 5.47 21.77 V01 GEFS2P5/20 12 2015073000 GFSANL G2 RELP TMP H2 6.87 5.83 5.36 4.91 4.39 4.45 3.99 3.94 4.51 4.11 3.76 4.00 4.03 4.16 4.53 5.03 5.10 6.12 7.02 7.89 V01 GEFS2P5/20 12 2015073000 GFSANL G2 RMSE TMP H2 0.67359 1.15754 -0.20714 0.63651 0.95791 V01 GEFS2P5/20 12 2015073000 GFSANL G2 RELI TMP H2 1222. 373. 392. 417. 365. 358. 344. 351. 337. 285. 314. 262. 277. 257. 222. 227. 253. 240. 320. 437. 3257. 77556. 5293. 3290. 2559. 1901. 1556. 1309. 1192. 975. 792. 715. 619. 574. 490. 422. 429. 425. 392. 432. 592. 3606. V01 GEFS2P5/20 12 2015073000 GFSANL G2 HTFR TMP H2 1.000 0.884 0.848 0.811 0.771 0.737 0.703 0.670 0.636 0.604 0.577 0.547 0.522 0.496 0.472 0.450 0.429 0.405 0.382 0.351 0.310 1.000 0.193 0.141 0.111 0.088 0.072 0.059 0.049 0.040 0.033 0.028 0.024 0.020 0.017 0.014 0.012 0.010 0.008 0.007 0.005 0.004 V01 GEFS2P5/20 12 2015073000 GFSANL G2 RPS TMP H2 0.94665 0.76644 0.77157 0.51910 1.94059 0.73250 0.25844 0.41729 0.67795 V01 GEFS2P5/20 12 2015073000 GFSANL G2 BSS TMP H2 0.05179 0.09000 0.42455 0.00201 0.04022 0.09000 0.16881 0.55784 0.80602 V01 GEFS2P5/20 12 2015073000 GFSANL G2 ECON TMP H2 -0.35 -0.02 0.18 0.26 0.35 0.46 0.58 0.67 0.71 0.59 0.47 0.30 0.04 -0.34 -0.99 -1.89 -3.70 -5.64 (18) Header: GEFS2P5/20 12 2015073000 GFSANL G2 HIST TMP H2 GEFS2P5/20: Ensemble system/ensemble size (this case, 20 member 2.5degree GEFS) 12: forecast hours 2015073000: validation time (so running time = 2015073000 -12 = 2015072912) GFSANL: Name of analysis data as truth) G2: grid#2 (i.e. 2.5 degree grid, other example G2/NH, North Hemisphere subdomain in grid#2 ) HIST: score type (other example: RELP, RMSE, RELI, HTFR, RPS, BSS, ECON) TMP: field name H2: 2 m height (other example, H10, P1000, P850, etc) HIST: Hostogram, = ensemble size + 1 RELP: Relitive position (closest position to members), = ensemble size RELI: Reliability plot, first 21 (ensemble size + 1) are observed numbers, next are 21 (ensemble size + 1) forecast numbers accordingly How to get reliability plot: (1) X-axis (forecast frequency) has 21 (ensemble size +1) ticks, starting from 0. (2) Devide observed numbers (first 21) by forecast frequency numbers (next 21), pair by pair (first devided by 22th, second by 23th, ...., 21 by 42) (3) Put the 21 results on the plots. In this case, 21 ticks on X-axias are: 0, 0.05, 0.10, 0.15, ...., 0.95, 1.00, so following 21 (x,y) pairs are (0, 1222./77556.=0.0158) (1/20=0.05, 373./5293.=0.0704) (2/20=0.1, 392./3290.=0.1191) ...... (20/20=1.0, 3257./3606.=0.9032) Using these 21 pairs fo data to get reliability plot HTFR: ROC plot, first 21 (ensemble size + 1) are hit rates, next 21 (ensemble size + 1) are 21 (ensemble size + 1) false alarm rates accordingly How to get ROC plot: 21 pairs of (hit rates, false alarm rates), hit rates on x-axis, false alarm rates on y-axis ECON: Economic value: fixed 18 numbers How to get plots: X-axis is C/L (Cost/Loss), 18 ticks are 1/18, 2/18, ...., 18/18 Y-axis, directly use 18 numbers as y values RMSE: 6 Error related scores Spread, RMSE, Medium-err, Absolute-err, Pattern-anomaly-correlation(PAC) RPS: 6 Ranked probability scores (RPS) and Continuous ranked probability scores (CRPS), last 3 are not used Forecast-RPS, Climatology-RPS, Ranked-probability-skill_scores, Forecast-CRPS, Climatology-CRPS, CRPSS BSS: 9 Brier scores (BS), Brier Skill Score (BSS) and BS's decomposition components Forecast-BS, Climatology-BS, BSS, BS-reliability-component, BS-resolution-component, BS-uncertainty, last 3 are not used