Multi-scale assessment of regional high-resolution reanalyses precipitation fields over Italy

Meteorological reanalyses are used for multiple applications, from understanding past climate to assessing extreme weather events. In this study, an extensive validation of high-resolution regional reanalyses was performed, evaluating their capability to reproduce precipitation fields over Italy, which is characterized by complex orography and land-sea interactions.

In this study, we carried out an inter-comparison of nine different reanalysis products, using the ECMWF ERA5 product as the global reanalysis reference. Some of the reanalyses considered cover Europe (BOLAM, COSMO-REA6, CERRA), while others are specifically designed for Italy (MERIDA, MERIDA-HRES, MOLOCH, SPHERA, VHR-REA_IT), using a variety of different atmospheric models and parametrizations.

The precipitation fields inter-comparison employed wavelet decomposition techniques and frequency distribution analysis. Then, a validation against independent observations involved both observation products interpolated onto regular grids and punctual data from stations’ series.

The wavelet decomposition permitted assessing the effective information of every reanalysis at each spatial scale, clustering the products into global, regional, and convection-permitting ones. On the other hand, the frequency distribution of daily rainfall amounts allowed proving the capability of higher-resolution reanalyses to depict the frequency of occurrence of extreme precipitation events better than the ERA5 global product.

Moreover, the comparison of the reanalyses fields with observations was performed by calculating the relative bias to assess climatology reliability and the Stable Equitable Error in Probability Space (SEEPS) to evaluate the ability to discriminate among dry, light rain, and heavy rain days. The results indicated wet biases over the Alps, especially during spring and summer, and dry biases over Liguria, Tuscany, the Apennines, and Southern Italy during autumn and winter. Significant differences were found among reanalysis products, with MOLOCH and MERIDA-HRES showing the best performances.

Finally, the trend analysis revealed a long-term signal in the reanalysis precipitation bias, with a significant increase in the annual amount in most reanalyses (with the only exception being VHR-REA_IT), suggesting caution when using these products for climate change studies in this area.