Files containing data in Figures 1-3, and 5-8 in the manuscript Korhonen N. et al. "Adding value to Extended-range Forecasts in Northern Europe by Statistical Post-processing Using Stratospheric Observations" Files and their contents: Korhonen_et_al_Figure1Data.nc Data of Fig. 1: annual mean CRPSS and its p-value for forecast weeks 1-6 for the raw (RAW) and the mean bias-corrected (BC) surface mean temperature forecast. Korhonen_et_al_Revised_Fig2Data_columns.txt Data of Fig. 2: observed mean daily AO indes 1-2, 3-4, and 5-6 weeks after different thresholds of the zonal mean zonal wind at 60 degrees N and 10 hPa. Korhonen_et_al_Revised_Fig3Data_columns.txt Data of Fig. 3: observed mean daily AO indes 1-2, 3-4, and 5-6 weeks after different stratospheric situations. Korhonen_et_al_Revised_Figure5a_fData.nc Data of Fig. 5a-f: ERA-Interim observed mean t2m anomalies during November-February in 1981-2016, 1-2, 3-4, and 5-6 weeks after ZMZW weaker than 3.8m/s or ZMZW stronger than 41m/s. p-value that the means of these surface temperature anomalies differ 1-2, 3-4, and 5-6 weeks after ZMZW weaker than 3.8m/s or ZMZW stronger than 41m/s. Korhonen_et_al_Revised_Figure5g_lData.nc Data of Fig. 5g-l: ERA-Interim observed mean t2m anomalies during November-February in 1997-2016, 1-2, 3-4, and 5-6 weeks after ZMZW weaker than 3.8m/s or ZMZW stronger than 41m/s. p-value that the means of these surface temperature anomalies differ 1-2, 3-4, and 5-6 weeks after ZMZW weaker than 3.8m/s or ZMZW stronger than 41m/s. Korhonen_et_al_Revised_Figure5m_rData.nc Data of Fig. 5m-r: ECMWF reforecasted mean t2m anomalies for November-February in 1997-2016, 1-2, 3-4, and 5-6 weeks after ZMZW weaker than 3.8m/s or ZMZW stronger than 41m/s. p-value that the means of these reforecasted surface temperature anomalies differ 1-2, 3-4, and 5-6 weeks after ZMZW weaker than 3.8m/s or ZMZW stronger than 41m/s. Korhonen_et_al_Revised_Figure6a_fData.nc Data of Fig. 6a-f: ERA-Interim observed mean t2m anomalies during November-February in 1981-2016, 1-2, 3-4, and 5-6 weeks after SWIneg or SWIplain. p-value that the means of these surface temperature anomalies differ 1-2, 3-4, and 5-6 weeks after SWIneg or SWIplain. Korhonen_et_al_Revised_Figure6g_lData.nc Data of Fig. 6g-l: ERA-Interim observed mean t2m anomalies during November-February in 1997-2016, 1-2, 3-4, and 5-6 weeks after SWIneg or SWIplain. p-value that the means of these surface temperature anomalies differ 1-2, 3-4, and 5-6 weeks after SWIneg or SWIplain. Korhonen_et_al_Revised_Figure6m_rData.nc Data of Fig. 6m-r: ECMWF reforecasted mean t2m anomalies for November-February in 1997-2016, 1-2, 3-4, and 5-6 weeks after SWIneg or SWIplain. p-value that the means of these reforecasted surface temperature anomalies differ 1-2, 3-4, and 5-6 weeks after SWIneg or SWIplain. Korhonen_et_al_Revised_Figure7Data.nc Data of Fig. 7: The CRPSSs of the ECMWF surface temperature reforecasts in forecast weeks 3–4 and 5–6, when the kSWI was ranging from 0.0 to 1.0. Korhonen_et_al_Revised_Figure8Data.nc Data of Fig. 8: CRPSS of forecast weeks 3–4 and 5–6 of the ECMWF’s mean temperature reforecasts for November–February 1997–2016 after mean bias-correction (BiasCorrected) in cases the ZMZW at 60 N and 10 hPa was a)-b) below 3.8m/s, c)-d) above 41 m/s, e)-f) between 3.8 m/s and 41 m/s. at the start of the forecast. And g)-h)after both mean bias-correction and the SWI based post-processing (PostProcessed) in cases the ZMZW at 60 N and 10 hPa was between 3.8 m/s and 41 m/s. p-value that the CRPSS is above zero.

2020