Keywords: ICON model, verification, weather prediction, forecast accuracy

This study presents a comprehensive verification of the ICON numerical weather prediction model over Ukraine for the year 2024. The evaluation covers key meteorological parameters - air temperature, wind speed, relative humidity, precipitation, and cloud cover - at 24-, 48-, and 72-hour forecast lead times. Both continuous metrics (correlation, mean absolute error, root mean square error, bias) and categorical metrics (POD, FAR, CSI) were applied, along with seasonal and spatial analyses. The model demonstrated high accuracy in forecasting mean temperature, with a correlation coefficient of r = 0.95 at 24 hours, low RMSE (?2.6?°C), and near-zero bias. Cloud cover forecasts also showed excellent performance, with POD > 0.94 and CSI up to 0.86 at a 10% threshold, maintaining stability across regions and seasons. By contrast, wind speed forecasts were less reliable, with lower correlations (r = 0.40 at 24 h), RMSE ~1.75?m/s, and consistent overestimation. Forecasts of relative humidity were moderately accurate (r = 0.88), although a persistent negative bias (~–4.2%) was observed. Precipitation forecasts exhibited the lowest skill, especially at longer lead times and higher thresholds. At a 0.1?mm threshold and 24-hour forecast, POD reached 0.61, but FAR remained high (>0.50), particularly in southern regions with frequent convective activity. Seasonal analysis indicated the best model performance in autumn and winter, with reduced accuracy in summer, especially for humidity and precipitation. Spatial verification at 24-hour lead time revealed regional differences: the lowest RMSE for mean temperature was found in Kherson (2.29?°C), while the highest wind speed error occurred in Donetsk (4.88?m/s). Overall, the ICON model provides robust forecasts for temperature and cloud cover, acceptable performance for humidity, and highlights the need for further refinement in wind and precipitation prediction. These findings offer valuable guidance for improving regional forecast applications and adjusting physical parameterizations under Ukrainian climate and topography conditions.

1. Bastin, S., Bock, O., Chazette, P., et al. (2019). Modeling and observations of lower tropospheric humidity profiles in the Mediterranean. Atmospheric Chemistry and Physics, 19(5), 3155–3178. https://doi.org/10.5194/acp-19-3155-2019

2. Bauer, P., Thorpe, A., & Brunet, G. (2015). The quiet revolution of numerical weather prediction. Nature, 525(7567), 47–55. https://doi.org/10.1038/nature14956

3. Dipankar, A., Schlemmer, L., Gorges, K., et al. (2015). ICON–a novel approach for global modeling of atmospheric dynamics: first results and future perspectives. Geoscientific Model Development, 8(7), 2489–2511. https://doi.org/10.5194/gmd-8-2489-2015

4. Doroshenko, A., Shpyg, V., Budak I., Huda K. (2020). Numerical atmospheric models and their application in different areas of economics [In: Kvasniy L. and Tatomyr I. (eds) Ukraine in the context of global and national modern servisation processes and digital economy]: monograph. Praha: Oktan Print, 155-171. doi: 10.46489/uitcog0909

5. Ebert, E.E., Janowiak, J.E. & Kidd, C. (2016). Comparison of near-real-time precipitation estimates from satellite observations and numerical models over global land areas. Journal of Hydrometeorology, 17(8), 2453–2470. https://doi.org/10.1175/JHM-D-16-0010.1

6. Giorgetta, M. A., Brokopf, R., Crueger, T., Esch, M., Fiedler, S., Helmert, J., et al. (2018). ICON-A, the atmosphere component of the ICON Earth system model: I. Model description. Journal of Advances in Modeling Earth Systems, 10. https://doi.org/10.1029/2017MS001242

7. Hoffmann, P., Wilms, H., Blank, L., & Ludwig, P. (2022). Benchmarking high-resolution regional climate models over complex terrain: A case study for wind simulations in the Alps. Frontiers in Earth Science, 10, 789332. https://doi.org/10.3389/feart.2022.789332

8. Holoborodko, P., Lykhovyd, O., & Fedirko, V. (2020). Regional climatology and atmospheric dynamics of Ukraine: challenges for numerical weather prediction. Ukrainian Hydrometeorological Journal, 12(1), 45–59.

9. Kalnay, E. (2003). Atmospheric modeling, data assimilation and predictability. Cambridge University Press. https://doi.org/10.1017/CBO9780511802270

10. Liu, L., Huang, X., Chen, F., et al. (2021). Evaluation of relative humidity in weather and climate models over complex terrain. Journal of Geophysical Research: Atmospheres, 126(12), e2020JD034123. https://doi.org/10.1029/2020JD034123

11. Shpyg, V., Budak, I. (2015). WRF reflectivity simulation and verification of thunderstorm forecast by radar and surface observation. 16th International Radar Symposium. Symposium Materials (24-26 June 2015, Dresden, Germany). 610-615. doi: 10.1109/irs.2015.7226388

12. Shyian, S., Dudnik, I. & Makarenko, A. (2018). Climatic conditions of Ukraine and their variability: implications for weather prediction. Geographical Review of Ukraine, 11(2), 33–42.

13. Sundqvist, H., Berge, E. & Kristj?nsson, J.E. (1989). Condensation and cloud parame- terization studies with a mesoscale numerical weather prediction model. Monthly Weather Review, 117(8), 1641–1657.

14. Ukrainian Hydrometeorological Center. (2023). Annual report on meteorological observations and challenges. Kyiv: State Hydrometeorological Service of Ukraine.

15. Z?ngl, G., Reinert, D., R?podas, P. & Baldauf, M. (2015). The ICON (ICOsahedral Nonhydrostatic) modelling framework of DWD and MPI-M: description of the nonhydrostatic dynamical core. Quarterly Journal of the Royal Meteorological Society, 141(687), 563–579. https://doi.org/10.1002/qj.2378