Comparison of different EPS postprocessing methods

Comparison of different EPS postprocessing methods

Pierre Eckert

Ralf Kretschmar

MétéoSuisse

ECMWF Medium range model “does” generate severe weather events: forecast/observed frequency ratio (FBI)

EFI

LEPS

Three „rescaling / downscaling“ techniques for extreme events

Extreme Forecast Index (EFI) rescale with respect to model climate

LEPS downscale ensemble with a LAM

Artificial neural Network (ANN) pattern recognition of extreme situations with respect to a given meteorological parameter (precipitation)

Verification: scores

Event Observed Not Observed

Forecasted A B

Not Forecasted C D

• Forecasted : Index / prob > given threshold

• Hit Rate = A / (A+C)• FAR = B / (A+B)

EFI verification: Switzerland South

Prec ge 50mm/24h

0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1

False Alarm Rate

Hit

Rat

e

T+66

T+90

T+114

T+138

Predictors for the Neural Network

Rescaled DMO (T511)

FFNN + precip DMO

ANN upper air + precip DMORainfall Lugano

Conclusions

• Forecasting of rare events requests special treatment• Downscaling / rescaling• Hit rate is usable• False Alarm Rate must be optimised• DMO, EFI, LEPS, ANN have each their own qualities• The chalenge for the LEPS is to reach the upper left

corner of the hit rate false alarm rate diagram (above the diagonal).

Instability indices

• Produced on all grid points• Should be evalueted• The following figures give an example of one case

Cold front 28.8.2003

Precipitation 27.8 6z – 28.8 6z

V A M P - K A R T E Tageswertdaten vom 28.08.2003TP050 RRSS Niederschlag; Tagessumme (konv.Periode: 0540-0540 Folgetag) (0.1 mm)................................................................................ SHA 73 BAS 95 GUT 67 RUE 79 LAE // TAE 73 FAH 125 KLO 129 STG 65 BUS 133 REH 116 SMA 112 HOE // CHA 70 WYN 148 WAE 78 SAE 57 CDF 146 NAP 63 LUZ 51 VAD 53 NEU 114 BER 65 PIL 24 GLA 79 FRE 212 PAY 123 ALT 39 CHU 147 PLF 208 INT 120 ENG 85 WFJ 186 SCU 134 MLS 49 JUN // GUE 20 DIS 127 DAV 168DOL 303 PUY 235 ABO 100 GRH 18 PIO 304 HIR 515 SAM 213 CGI 342 AIG 164 MVE 66 ULR 11 ROE 311 COM 552 SBE 737 COV 122GVE 240 SIO 104 VIS 17 CIM 900 ROB 113 FEY 70 OTL 1266MAG 833 EVO 62 ZER 70 LUG 103 GSB 72 SBO 1

COSMO LEPS

Faust Index

Surface Lifted Index

SWISS12 Combined Index

Boyden Index