Bulgaria flood case during 02 09 2022

During 01/09/22 a Central Mediterranean cyclone moved eastwards through the southern part of the Balkan Peninsula and then northwards towards the Black Sea. In the northwestern part of the Bulgaria the wind was directed to northwest, while in other parts of the country to east-northeast. A powerful convection developed in Central Bulgaria and especially in the valley of city of Karlovo. Precipitation began in the late afternoon of 1.IX and at midnight it became more intensive, until about 6 UTC on 2.IX. In several settlements in the municipality of Karlovo, as the village of Karavelovo, as a result of torrential rain, buildings and industrial enterprises were flooded, roads were broken, bridges were destroyed, over 60 cars were dragged away by the water element and dikes of Stryama River were destroyed. The operational models at NIMH, ALADIN-BG (that is run operationally with a horizontal resolution of 5 km, 105 vertical levels, 72 hours forecast length, using lateral and boundary conditions - LBCs from ARPEGE) and AROME-BG (that is run operationally with a horizontal resolution of 2.5 km, 90 vertical levels, 48 hours forecast length, using LBCs from ALADIN-BG), as well IFS from ECMWF did not forecast such a heavy rain. However, 3 days before, on 30 August, the 18 UTC run of ALADIN (which is forced by ARPEGE) predicted relatively correctly in place and time the 6h accumulated precipitation, which unfortunately was not confirmed by the next model runs. High resolution tests with AROME performed (with a resolution of 500 m - AROME-HR) using deode_prototype on ATOS did not show any improvement of the performance of the forecast, as the LBCs from IFS did not provide initial conditions for thunderstorm formation. In Figure 1a is shown the measured at the automatic stations accumulated precipitation for the last 6h at 06 UTC on 02/09/22. The green points indicate the detected by ATDnet lightning for the same period, indicating the strong thunderstorm formed in this region. In Figure 1b is shown the low amount of 6h precipitation forecasted AROME-HR for this period, due to the missing of unstable atmospheric conditions in the LBCs provided by IFS. Figures 1c,d,e show the 6h accumulated precipitation forecasted by ALADIN-BG for the same period on 30/08/22 (which is the best forecast for the heavy precipitation that occurred), 31/08/22 (which already underestimated significantly the expected precipitation) and 01/09/22 (which did not predict any precipitation) respectively. In all graphs, the bold black point indicate the location of the village of Karavelovo. Our results confirm the importance of the LBCs and the fact that if they do not provide any conditions for precipitation for example, the high resolution runs cannot improve the performance of the forecast of a severe weather event. Tests will be performed to force AROME-HR with ALADIN-BG from 30/08/22 (18 UTC run) to evaluate the impact of the hectometric resolution on forecast performance (in time and especially in place) as well of its forecast range increase (as it will require a forecast range higher than 48 hours, which is usually believed not to be reliable for a non-hydrostatic model).

Figure 1. 02/09/2022 – 06UTC: Total precipitation for the last 6h: from top to bottom a) measured at the automatic stations (green points indicate the location of the detected lightning by ATDnet); b) forecasted by AROME-HR; and forecasted by ALADIN-BG during the consecutive runs on c) 30/08/22; d) 31/08/22; e) 01/09/22 – Bold black point indicate the location of Karavelovo, where damaged flooding occurred due to the precipitation