Forecasting in a Changing Climate Harold E. Brooks NOAA/National Severe Storms Laboratory (Thanks to Andy Dean, Dave Stensrud, Tara Jensen, J J Gourley, Jerry Brotzge)

Observations Forecasting begins with observations New and experimental radars –Dual-polarization (NEXRAD ) –Phased array Multi-sensor analyses

Radar-estimated rainfall Dual-Pol Reflectivity Overestimate (hail contamination)

R(Gauge) N=1299 R(Z) Non-polarimetric R(K DP ) R(syn) R(Z,Z DR |BC,HCA)R(Z,Z DR |JPOLE) R(Z,Z DR |BC)R(Z,Z DR |JPOLE,HCA) NB: 0.00 RMSE: 0.48 mm CORR: 1.00 NB: RMSE: 6.51 mm CORR: 0.83 NB: RMSE: 5.46 mm CORR: 0.88 NB: RMSE: 4.95 mm CORR: 0.91 NB: RMSE: 5.49 mm CORR: 0.89 NB: RMSE: 5.81 mm CORR: 0.87 NB: RMSE: 5.66 mm CORR: 0.89 NB: RMSE: 5.40 mm CORR: 0.88

Rainfall algorithms evaluated with long-term bias removed R(syn) and R(Z,ZDR|BC,H CA) hit peak All polarimetric algs better than R(Z) Hydrolgic Evaluation – TS Erin (187 mm/3 hr)*

3-Yr Hydrologic Evaluation Open circles show results with long- term bias removed Best Skill Increasing Skill As is Bias Corrected Simple Complex

Oklahoma City Area Tornadoes 5:47 PM 10 May 2010

Questions Questions ???

Forecasting User-based evaluation tools 0-48 hour ensembles –Information content Future ensembles –Very high resolution –Radar input –Run often for a few hours

Example of Object-Oriented Verification Method 12 hour Forecast Valid: 11 May UTC Fcst Object #1 Matched Observed Objects HRRR – Simulated Composite Reflectivity NSSL – Q2 Composite Reflectivity Objects from DTC Model Evaluation Tool Method for Object Based Diagnostic Evaluation Fcst Object #2

Experimental QPF from Ensembles Mean Processed Max Obs

Convective-scale Warn-on-Forecast Vision Probabilistic tornado guidance: Forecast looks on track, storm circulation (hook echo) is tracking along centerline of highest tornadic probabilities Radar and Initial Forecast at 2100 CSTRadar at 2130 CST: Accurate Forecast Most Likely Tornad o Path T=2120 CST T=2150 T=2130 T= % 50% 30% T=2200 CST Developing thunderstorm Most Likely Tornad o Path T=2120 CST T=2150 T=2130 T=2140 T=2200 CST An ensemble of storm-scale NWP models predict the path of a potentially tornadic supercell during the next 1 hour. The ensemble is used to create probabilistic tornado guidance. 70% 50% 30% Stensrud et al (October BAMS)

Thompson and Wicker 29 May 2004 Data Impact Phased Array vs 88D PAR 20 Minutes (20 Volumes) WSR-88D 20 Minutes (5 Volumes)

Generating the ensemble

Challenges Rapid and accurate data quality control –Ordinary vs. extraordinary Model error Sensitivity to errors in environmental conditions Assimilation method to use? Ensemble methods for convective-scale