Sub-daily temporal reconstruction of historical extreme precipitation events using NWP model simulations Vojtěch Bližňák1 (bliznak@ufa.cas.cz), Miloslav Müller1,2, Petr Zacharov1, Marek Kašpar1, Róbert Kvak1,2 1 Institute of Atmospheric Physics, CAS, Prague, Czech Republic 2 Charles University in Prague, Faculty of Science, Prague, Czech Republic Motivation Data NWP model COSMO Horizontal resolution: 2.8 km Frequency of forecasts: 10 min Purpose Enables to evaluate historical extreme precipitation events (EPEs) in terms of their sub-daily extremity Enables a quantitative assessment of the relationship among properties of EPEs and causal atmospheric (thermo) dynamic conditions Current status: Observations of historical EPEs in a high temporal step with a sufficient level of accuracy are missing Possible solution: Combination of rain gauge and NWP model data using an appropriate adjustment method Temporal reconstruction Rain gauge data 650-780 rain gauges Temporal resolution: 24 hours Radar reflectivity data 2 Czech C-band Doppler radars - Brdy, Skalky Constant Altitude Plan Position Indicator (CAPPI 2 km) Temporal resolution: 10 minutes (up to May 2009) 5 minutes (since June 2009 onwards) Horizontal resolution: 1 km Verification Temporal reconstruction Selected EPEs NWP MODEL 10-min precipitation totals 12.08. 2002 Stratiform precipitation (large area) Prevailing N-NW flow High precipitation totals in the west part of the Czech Republic Precipitation caused immense floods on Vltava and Labe rivers Long-lasting precipitation (several days) 24.06. 2009 Convective precipitation (local) Prevailing NE-E flow High precipitation totals in the east part of the Czech Republic Precipitation caused flash floods on Jičínka river Time of the highest precipitation totals: 17-18 UTC Correlation of the fields with a shift between -70 and +70 pixels in a direction [xi, yi] Direction [x, y] with highest correlation NWP MODEL 10-min shifted precipitation totals NWP MODEL 24-h precipitation sum RAIN GAUGE RECORDS 24-h precipitation sum 24-h shifted precipitation sum Ratio 24-h adjusted precipitation sum Adjustment (Sokol, 2003) Results Observation 12.08.2002 Non adjusted model In general good forecast Adjusted model Minor adjustment in respect to good original forecast Observation 24.06.2009 Non adjusted model Mismatch of the forecast and observation Adjusted model Very good match with observation Verification Averaged Fraction Skill Score (FSS) of 10-min precipitation totals from raw and adjusted model 12.08.2002 24.06.2009 Improved FSS values for adjusted model Comparison of 24-h precipitation sums Comparison of 24-h precipitation sums Improved accuracy of 24-h sums Significantly improved accuracy of 24-h sums FSS for 10-min precipitation totals from raw and adjusted model Similar FSS values for raw and adjusted model Similar correlations of 10-min re-forecasts Significantly improved FSS values for adjusted model in the time of the storm occurrence Significantly higher correlations of 10-min re-forecasts in the time of the storm occurrence Conclusions and outlook NWP model COSMO produces 10-min re-forecasts of stratiform precipitation with a good level of accuracy => small differences in the accuracy of raw and adjusted model NWP model COSMO adjusted by 24-h rain gauge observations including a shift based on the highest correlation between modelled and observed precipitation sums significantly improved the accuracy of 10-min re-forecasts of convective precipitation in space and time Suggested adjustment method will be then applied to more historical EPEs to demonstrate its sub-daily extremity Each EPE will be also evaluated by return periods of respective maximum precipitation totals utilizing return levels of given time intervals. Acknowledgement The research was supported by the Czech Science Foundation under the 17-23773S project, and by the Czech Ministry of Agriculture under the QJ1520265 project. The acknowledgement belongs to the Czech Hydrometeorological Institute for provided data and Z. Sokol for provision of adjustment procedure. References [1] Sokol, Z., 2003. The use of radar and gauge measurements to estimate areal precipitation for several Czech river basins. Stud. Geophys. Geod., 47, 587–604. [2] Ebert, E.E., 2009. Neighborhood verification: a strategy for rewarding close forecasts. Weather Forecast., 24, 1498–1510. 9th EMS European Conference on Severe Storms – ECSS 2017, 18 - 22 September 2017, Pula, Croatia