Vol. 125, No. 4 * Pages 521–692 * October - December 2021

A local three-dimensional variational data assimilation (DA) system was implemented operationally in AROME/HU (Application of Research to Operations at Mesoscale) non-hydrostatic mesoscale model at the Hungarian Meteorological Service (OMSZ) in 2013. In the first version, rapid update cycling (RUC) approach was employed with 3-hour frequency in local upper-air DA using conventional observations only. Optimal interpolation method was adopted for the surface data assimilation later in 2016. This paper describes the current developments showing the impact of more conventional and remote-sensing observations assimilated in this system, which reveals the benefit of additional local high-resolution observations. Furthermore, it is shown that an hourly assimilation-forecast cycle outperforms the 3-hourly updated system in our DA. Besides the upper-air assimilation developments, a simplified extended Kalman filter (SEKF) was also tested for surface data assimilation, showing promising performance on both the analyses and the forecasts of AROME/HU system.

Two historical Observing System Experiment (OSE) studies using the ALADIN limited area model and its assimilation system are described. The first study, using an OSE scenario that minimizes the impacts of observations through the lateral boundary conditions, demonstrated the importance of each assimilated terrestrial (radiosonde, aircraft, and wind profiler) observations on the analyses and short-range forecasts of the ALADIN/HU model and proved evidence, that the role of conventional observations cannot be even partly taken over by satellite measurements without degradation of the forecast quality. The second study demonstrated that the assimilation of radiosonde observations remains indispensable even with a progressively increasing amount of aircraft measurements.

The paper presented is dedicated to the evaluation of the influence of various improvements to the numerical weather prediction (NWP) systems exploited at the Slovak Hydrometeorological Institute (SHMÚ). The impact was illustrated in a case study with multicell thunderstorms and the results were confronted with the reference analyses from the INCA nowcasting system, regional radar reflectivity data, and METEOSAT satellite imagery.
The convective cells evolution was diagnosed in non-hydrostatic dynamics experiments to study weak mesoscale vortices and updrafts. The growth of simulated clouds and evolution of the temperature at their top were compared with the brightness temperature analyzed from satellite imagery. The results obtained indicated the potential for modeling and diagnostics of small-scale structures within the convective cloudiness, which could be related to severe weather.
Furthermore, the non-hydrostatic dynamics experiments related to the stability and performance improvement of the time scheme led to the formulation of a new approach to linear operator definition for semi-implicit scheme (in text referred as NHHY). We demonstrate that the execution efficiency has improved by more than 20%.
The exploitation of several high resolution measurement types in data assimilation contributed to more precise position of predicted patterns and precipitation representation in the case study. The non-hydrostatic dynamics provided more detailed structures. On the other hand, the potential of a single deterministic forecast of prefrontal heavy precipitation was not as high as provided by the ensemble system. The prediction of a regional ensemble system A-LAEF (ALARO Limited Area Ensemble Forecast) enhanced the localization of precipitation patterns. Though, this was rather due to the simulation of uncertainty in the initial conditions and also because of the stochastic perturbation of physics tendencies. The various physical parameterization setups of A-LAEF members did not exhibit a systematic effect on precipitation forecast in the evaluated case. Moreover, the ensemble system allowed an estimation of uncertainty in a rapidly developing severe weather case, which was high even at very short range.

In the last decades, wind power became the second largest energy source in the EU covering 16% of its electricity demand. However, due to its volatility, accurate short range wind power predictions are required for successful integration of wind energy into the electrical grid. Accurate predictions of wind power require accurate hub height wind speed forecasts, where the state-of-the-art method is the probabilistic approach based on ensemble forecasts obtained from multiple runs of numerical weather prediction models. Nonetheless, ensemble forecasts are often uncalibrated and might also be biased, thus require some form of post-processing to improve their predictive performance. We propose a novel flexible machine learning approach for calibrating wind speed ensemble forecasts, which results in a truncated normal predictive distribution. In a case study based on 100m wind speed forecasts produced by the operational ensemble prediction system of the Hungarian Meteorological Service, the forecast skill of this method is compared with the predictive performance of three different ensemble model output statistics approaches and the raw ensemble forecasts. We show that compared with the raw ensemble, post-processing always improves the calibration of probabilistic and accuracy of point forecasts, and from the four competing methods, the novel machine learning based approach results in the best overall performance.

An operational air quality forecasting model system has been developed and provides daily forecasts of ozone, nitrogen oxides, and particulate matter for the area of Hungary and three big cites of the country (Budapest, Miskolc, and Pécs). The core of the model system is the CHIMERE off-line chemical transport model. The AROME numerical weather prediction model provides the gridded meteorological inputs for the chemical model calculations. The horizontal resolution of the AROME meteorological fields is consistent with the CHIMERE horizontal resolution. The individual forecasted concentrations for the following 2 days are displayed on a public website of the Hungarian Meteorological Service. It is essential to have a quantitative understanding of the uncertainty in model output arising from uncertainties in the input meteorological fields. The main aim of this research is to probe the response of an air quality model to its uncertain meteorological inputs. Ensembles are one method to explore how uncertainty in meteorology affects air pollution concentrations. During the past decades, meteorological ensemble modeling has received extensive research and operational interest because of its ability to better characterize forecast uncertainty. One such ensemble forecast system is the one of the AROME model, which has an 11-member ensemble where each member is perturbed by initial and lateral boundary conditions. In this work we focus on wintertime particulate matter concentrations, since this pollutant is extremely sensitive to near-surface mixing processes. Selecting a number of extreme air pollution situations we will show what the impact of the meteorological uncertainty is on the simulated concentration fields using AROME ensemble members.

This study is focusing on the past and, in particular, the present of the ALADIN-Climate model used at the Hungarian Meteorological Service. The currently applied model version is 5.2 (HMS-ALADIN52). In the recent experiments, the CNRM-CM5 global model outputs were downscaled in two steps to 10 km horizontal resolution over Central and Southeast Europe using RCP4.5 and RCP8.5 scenarios. Temperature and precipitation projections are analyzed for 2021-2050 and 2071–2100 with respect to the reference period of 1971–2000 with focus on Hungary. The results are evaluated in comparison to 26 simulations selected from the 12 km horizontal resolution Euro-CORDEX projection ensemble (including two additional versions of ALADIN-Climate: CNRM-ALADIN53 and CNRM-ALADIN63) to get more information about the projection uncertainties over Hungary and to assess the representativeness of HMS-ALADIN52.
The HMS-ALADIN52 simulations project a clear warming trend in Central and Southeast Europe, which is more remarkable in case of greater radiative forcing change (RCP8.5). From the 2040s, the Euro-CORDEX simulations start to diverge using different scenarios. The total range of the annual change over Hungary is 1.3–3.3 °C with RCP4.5 and 3.2–5.7 °C with RCP8.5 by the end of the 21st century. HMS-ALADIN52 results are approximately near to the median: 2.9 °C with RCP4.5 and 4 °C with RCP8.5. CNRM-ALADIN53 shows generally similar results to HMS-ALADIN52, but simulations with CNRM-ALADIN63 indicate higher changes compared to both. In terms of seasonal mean precipitation change, the HMS-ALADIN52 simulations assume an increase between 9% and 33% (less in spring, more in autumn) over Hungary in both periods and with both scenarios. Most of the selected Euro-CORDEX simulations show a precipitation increase, apart from summer, when growth and reduction can be equally expected in 2021–2050, and the drying tendency continues towards the end of the century. Increase projected by HMS-ALADIN52 is mostly confirmed by CNRM-ALADIN53, while CNRM-ALADIN63 predicts precipitation decrease in summer. Precipitation results do not show a significantly striking difference between the scenarios, likely due to the fact that internal variability and model uncertainty are more relevant sources of uncertainty in precipitation projections over our region.

Cities, due to their warmer and dryer local climate in addition to their dense population, are subjected to large future climate change risks. Land surface models, with detailed urban parameterization schemes, serve as an adequate tool to refine the rough regional climate projections over the cities. In this study, the future temperature conditions in Budapest are studied with the SURFEX land surface model (LSM), driven by the HMS-ALADIN5.2 regional climate model (RCM) and considering the high-emission RCP8.5 scenario. Special attention is dedicated to explore the differences between the RCM and LSM in terms of the results, their interpretation, and further use in impact models. According to the investigated model combination, the winter season may warm the most, with 1.9 °C in 2021–2050 and 4.3 °C in 2071–2100, although the magnitude of this change is smaller in SURFEX than in ALADIN. Besides the mean changes, four climate indices, based on high and low temperature thresholds, were studied, and it was found that the low temperature indices (frost days and very cold days) may relatively decrease more in SURFEX compared to ALADIN over Budapest, and in the city center compared to the suburbs and rural areas. In addition, the urban heat island (UHI) intensity is projected to decrease in SURFEX mainly in spring and summer (by 2071–2100 with 0.35 °C and 0.32 °C, respectively). Finally, a simple method is provided to correct the SURFEX temperature fields, using the ALADIN model, with eliminated systematic biases and the simulated UHI field.