Climate change analysis in Campania

Climatic phenomena are often the product of two or more, simple, interacting non-linear processes. As a result, chaotic processes in the atmosphere are extremely sensitive to small disturbances. Small variations in atmospheric turbulence can result in very different outcomes and then it becomes impossible either to measure the system accurately or to predict its future state (Bryant 1997). The actual atmospheric phenomena taking place over an area are the final stage of a number of different processes occurring on different scales, therefore the estimation of the areal distribution of meteorological parameters from point observations has been, and probably will remain, one of the most difficult issues within geophysics.

The occurrence of intense flooding causing landslides in autumn and winter in Campania depends on small cyclonic areas, the dynamics of which follow the genesis of tropical cyclones (hurricanes), but show a low level of energy (Tranfaglia & Braca 2004). Such meteorological systems, together with convective systems and orographic rainfall, can be intensified by the higher contribution of heat at the sea surface and often cause sudden flooding in coastal regions and in mountain regions exposed to sea winds.

A major challenge to climate researchers is to determine the degree of predictability associated with these and other events. The analysis of historical series of rainfall recorded in four large Italian watersheds shows that more than 50% of their interannual variance is significantly explained by the 22-year harmonics. It is proposed by Mazzarella et al. (2003) that the 22-year solar magnetic activity is able to influence the zonal circulation over the Mediterranean basin and the relative rainfall.

Mazzarella & Tranfaglia (2000) investigated the capability of the historical rain gauge network belonging to the Naples Hydrographic Service to measure the annual rainfall in Campania. They found that the value of the fractal dimension D was equal to 1.84, with a confidence level higher than 99%, within a scaling region enclosed between 8 km and 64 km with a dimensional deficit equal to 0.16 (2-1.84). A 50% lower limit value of the scaling region (4 km) represents the optimal value of the network resolution.

Analysis by deterministic and statistical methods

With a view to verifying the homogeneity of the historical series of annual rainfall data from the stations in Campania, a classic analysis of the double-mass curve was performed. The cumulative values of rainfall depths measured in each station were compared with the corresponding values of rainfall measured at the nearest stations.

For example, analysis between the stations of Naples University and Naples Hydrographic Service and between the stations of Naples Capodi-monte and Naples University, besides evident homogeneity of the historical series, also shows a difference between rainfall values in the order of 5%. From the same analysis it becomes evident that between the stations of Naples Capodimonte and Naples Hydrographic Service there is significant homogeneity (constant slope of the doublemass curve), but also a substantial coincidence between the historical series (the curve of the double-mass is fitted by a straight line with a 45° slope).

On the basis of such considerations, a 182-year historical series of precipitation (from 1821 up to 2003) for the city of Naples was constructed, supplementing the data lacking in the Naples Capodimonte series with those of the Naples Hydrographic Service series (Fig. 3).

In order to verify the influence of extreme events on short-term trends (Burroughs 2001), regression analysis was performed based on the least-squares method. A scatter plot of annual precipitation versus number of rainy days was examined. A best-fit curve was sought to verify whether the data have a significantly linear trend. A good correlation between annual precipitation and number of rainy days was found (R2 = 0.78, Fig. 4). It is important to establish a threshold for the significance of the correlation coefficient

Fig. 3. Annual rainfall (mm/a) measured at Naples Capodimonte from 1821 to 2003.

(Swan & Sandilands 1995). A statistical test was carried out for all available data. The significance of the correlation coefficient was estimated using a i-statistic. The value of r can be tested against a hypothesis H0 (p = 0). The data are taken without bias from a population which is normally distributed with respect to both variables. It was found that t equals 12.86 and critical t = 1.67, with a = 0.05 and degrees of freedom = n — 2 = 47. As the calculated t exceeds the critical t, the null hypothesis is rejected and it can be stated that a significant correlation exists between annual rainfall and number of rainy days in Campania. The observation of the number of rainy days and rainfall data in the last decade shows that the hypothesis proposed by some authors (Alpert et al. 2002; Kitoh 2003) of a Mediterranean-wide decrease in the number of rainy days in respect to the rainfall rate in recent years, does not seem confirmed by the data from Campania region where there has been merely a slight levelling of the slope compared with the curve in Figure 4, and there has been no inversion in the trend.

In order to show the values of individual years that do not follow the signal (Bryant 1997), a curve was also sought to best-fit mean annual time series of precipitation in Campania. Confidence limits were fitted to the trend line to highlight the extreme values.

Fig. 4. Mean annual rainfall versus mean rainy days in Campania (estimated for 49 rain gauge stations from 1951 to 1999) with confidence bands for mean.

Annual rainfalls (mm)

Fig. 4. Mean annual rainfall versus mean rainy days in Campania (estimated for 49 rain gauge stations from 1951 to 1999) with confidence bands for mean.

Was this article helpful?

0 0

Post a comment