Czech case (26 30 Oct 2017)
Extra materials:
During late October 2017, Europe experienced a persistent frontal zone caused by a large high-pressure system over the British Isles and a low-pressure trough stretching from northern Scandinavia to the southern Balkans. In the strong north-westerly flow, the secondary pressure low triggered storm cyclone Herwart, which began to intensify rapidly. This intensification was linked to the influx of cold Arctic air along the Scandinavian coast and the leeward effects of the Norwegian mountains. In the Czech Republic, a significant runoff situation associated with the Herwart pressure low also occurred in the Krkonoše and Jizera Mountains. Rainfall totals were around 60mm/24h in places and were intensified by powerful winds that reached hurricane force on the mountain ridges.
ALARO On-Demand DT prototype 0.4.0, 500m resolution (2000x1600), 87L, coupled to 4-5km DT i8hy
- OD-DT prototype A initialized with “HIGHRES” ALARO analysis + DFI
- OD-DT prototype B downscaled + DFI
- OD-DT prototype C downscaled without DFI
- “Old” ALARO CSC operational at a time based on CY38T1 with local improvements namely in radiation (ACRANEB2) and turbulence (TOUCANS) schemes (analysis obtained by surface assimilation, synchronous DFI blending with ARPEGE NWP assimilation followed by 3DVAR run in 4.7 km horizontal resolution, coupled to ARPEGE)
- “HIGHRES” ALARO CSC with recent improvements of physics parameterizations (analysis obtained by surface assimilation, synchronous DFI blending with ARPEGE NWP assimilation followed by 3DVAR run in 2.3 km horizontal resolution, downscaled to 500m and balanced with 1h DFI, coupled to ARPEGE)
Comparison of reference experiments with resolutions of 4.7km and 500 m demonstrates the benefits of increased model resolution. The reference experiment “HIGHRES” best captures bias values of hourly rainfall totals and their temporal evolution, accurately depicting the onset and the course of the flood event. On-Demand DT ALARO experiments with 500 m resolution coupled with DT showed higher precipitation bias over the Krkonoše and Jizera Mountains, also the timing is not aligned. OD-DT experiments differ not only by initial and boundary conditions, another factor is the NWP model configuration settings. The “HIGHRES” reference contains a modified model physics setup including prognostic graupels, which are not yet in the OD-DT prototype. The initial condition sensitivity experiments were performed using the “HIGHRES” reference analysis and a short digital filter (DFI) on top of the downscaled DT. Very little effect of the DFI on the precipitation bias was observed. The initial condition with data assimilation showed potential to improve the precipitation prediction in the first hours.
Drawing conclusions from a single case study which has a large scale character may be challenging. Nonetheless, hectometric simulation showed added value to the prediction of precipitation and the impact of initial conditions is noticeable. Next phase of the project should focus on determining the optimal setup of the NWP model configuration.
Figure 1. BIAS of 1h accumulated precipitation over Krkonoše a Jizera Mountains for 28 Oct 2017 12UTC 24h-forecast of “OLD” ALARO reference (blue), “HIGHRES” ALARO reference (green), OD-DT prototype A (orange), OD-DT prototype B (violet), OD-DT prototype C (light blue).
Figure 2. The 3h accumulated precipitation for 28 Oct 2017 12UTC zoomed over the region of Krkonoše a Jizera Mountains. From left to right: the radar and rain gauges based quantitative precipitation estimate, “HIGHRES” ALARO reference, OD-DT prototype A, OD-DT prototype B, OD-DT prototype C.
Figure 3. The 3h accumulated precipitation for 28 Oct 2017 12UTC over the Czech Republic. From left to right: the radar and rain gauges based quantitative precipitation estimate,“HIGHRES” ALARO reference, OD-DT prototype A, OD-DT prototype B, OD-DT prototype C.