Statistical characterisation of winter precipitation in the Abruzzo region (Italy) in relation to the North Atlantic Oscillation (NAO)

• Published in: Atmospheric Research
• Main Authors: VERGNI, LORENZO, Di Lena, Bruno, Chiaudani, Alessandro
• Other Authors: Vergni, Lorenzo
• Format: Article in Journal/Newspaper
• Language: English
• Published: 2016
• Subjects: Drought event; L-moment; Probability distribution; Spatial variability; North Atlantic; North Atlantic oscillation
• Online Access: http://hdl.handle.net/11391/1377007; https://doi.org/10.1016/j.atmosres.2016.03.028; http://www.sciencedirect.com/science/article/pii/S0169809516300849

This paper analyses the relationship between the North Atlantic Oscillation (NAO) and winter precipitation at 75 stations in the Abruzzo region (Central Italy) during the period 1951-2009. The analysis was carried out for individual months (December, January, February, March) as well as for 3-month accumulation periods, December to February (DJF) and January to March (JFM). A composite analysis, based on the positive and negative values of the North Atlantic Oscillation Index (NAOI) was used to compare the characteristics of precipitation associated with NAOI positive (NAO+) and negative (NAO-) conditions. Firstly, the NAO+ and NAO- precipitation subsets were compared by using the Brown-Forsythe and Mann-Whitney tests to assess the presence of significant differences in the variance and mean, respectively. Then the L-moment ratio diagrams were used to identify, for each period and NAO phase, the best-fitting two-parameter distribution. The NAO+ and NAO- conditions were also compared on the basis of the first (D1) and ninth (D9) precipitation deciles. The results show that the NAO mainly affects the mean precipitation amounts that are in most cases lower (higher) during the NAO+ (NAO-) phase, particularly for JFM. This influence is characterised by a relevant spatial variability that can be attributed to local factors (orography and distance from the sea). No significant differences were found in the precipitation variance, but this means that NAO+ precipitation, due to lower mean values, is relatively more variable than NAO- precipitation. The L-moment ratio diagrams reveal that the Weibull distribution always provides the best overall fitting for NAO- conditions (with the exception of DJF for which the log-normal is the best). For NAO+ conditions the best-fitting distributions become the generalized Pareto (January and February), the Weibull (DJF) and the Gamma (JFM), while no change in the distribution type is observed in December and March. In most cases, both D1 and D9 are lower during NAO+ phases. However, ...