## Results and Discussion

The estimation of the optimal parameter set for the stochastic rainfall model, NSRP, was first conducted. Then, the NSRP model was used to simulate ten rainfall realizations, each one 100 years long. More simulation runs were considered to take account the uncertainty in the rainfall series linked to the stochastic nature of the model. The capability in preserving the statistical properties of rainfall time series and the good agreement between the Depth-Duration-Frequency (DDF) curves extracted from observed and synthetic rainfall record demonstrated the goodness of the model in reproducing the observed data (not shown here for sake of brevity).

Following the calibration procedure proposed by Montanari et al. [4], the FARIMA model was implemented for generating ten temperature series, each one 100 years long. The obtained series of rainfall and temperature were employed as input for discharge computation through the MISDc, previously calibrated on a number of significant observed flood events. For instance, Fig. 11.2 shows the comparison between observed and simulated discharge through MISDc for the three most significant flood events occurred for the Niccone and Genna catchments. As it can be seen, a fairly good agreement was detected.

Finally, once the ten discharge time series were obtained by MISDc, the maximum annual discharge was extracted, thus having ten flood frequency curves. The mean flood frequency curve was then used as reference for the two scenarios A2 and B2. Then, for each scenario, such flood frequency curve was compared with the one inferred by the actual observations. The comparison was first based on the changing of temperature and then adding also the rainfall. The results showed that the effects of the only temperature change on flood frequency were less evident than the ones obtained by perturbing both rainfall and temperature time series (see also Table 11.1). Through the comparison of two scenarios (see Table 11.1), it is evident that A2 was more critical for the short term (2020) with an increase of the maximum discharge up to 78%; whereas for the long term (2080) the differences in magnitude were not exceeding 15% for all sub-catchments, except for the Caina, for which (see Table 11.1) the discharge reduction drops below 30%. On the contrary, for the scenario B2, the forecast at 2080 was such that the change in maximum discharge was much higher than the other ones. It is worth noting that, for the two neighbouring catchments, Genna and Caina, which are characterized by similar rainfall regimes, the climate change effects on flood frequency were different, as shown in Fig. 11.3.

In particular, for the Genna catchment, the forecast based on the scenario A2 was characterized by a high increase of discharge at short term which tends to dampen

Fig. 11.2 Comparison between observed and simulated discharge through the MISDc model for several flood event occurred in the study period for: (a-c) Niccone, and (d-f) Genna catchment

Fig. 11.2 Comparison between observed and simulated discharge through the MISDc model for several flood event occurred in the study period for: (a-c) Niccone, and (d-f) Genna catchment

Table 11.1 Percentage differences, for different return periods, between the peak discharges estimated considering the emission scenarios A2 and B2 and the ones inferred by the actual observations

Scenario A2 Scenario B2 Scenario A2 Scenario B2

Forecasting Return period (years) Return period (years)

Table 11.1 Percentage differences, for different return periods, between the peak discharges estimated considering the emission scenarios A2 and B2 and the ones inferred by the actual observations

Scenario A2 Scenario B2 Scenario A2 Scenario B2

Forecasting Return period (years) Return period (years)

period |
20 |
50 |
100 |
20 |
50 |
100 |
20 |
50 |
100 |
20 |
50 |
100 |

Caina catchment |
Genna catchment | |||||||||||

2020 |
26 |
24 |
31 |
3 |
2 |
6 |
78 |
63 |
40 |
40 |
39 |
39 |

2050 |
-12 |
-12 |
-12 |
-15 |
-13 |
-11 |
33 |
15 |
-4 |
28 |
24 |
12 |

2080 |
-31 |
-31 |
-26 |
28 |
23 |
23 |
12 |
-3 |
-15 |
54 |
43 |
21 |

Cerfone catchment |
Niccone catchment | |||||||||||

2020 |
34 |
37 |
35 |
22 |
31 |
29 |
56 |
57 |
53 |
34 |
38 |
38 |

2050 |
9 |
14 |
24 |
4 |
11 |
9 |
37 |
39 |
28 |
7 |
14 |
22 |

2080 |
8 |
14 |
12 |
30 |
29 |
25 |
1 |
8 |
8 |
66 |
69 |
57 |

at long periods. For the Caina catchment, the increase of maximum discharges at short term was much more limited than that of Genna catchment; whereas for the other two periods a decrease is estimated. These differences might be due to different permeability of two sub-catchments which is greater for the Caina catchment. Therefore the geomorphological and land use characteristics might have a fundamental role in the impact of climate change. This aspect is important also for the consequence of climate change in terms of territorial forward planning.

CAINA catchment

GENNA catchment

300 250

2oo l5o loo 5o o

300 250

2oo l5o loo 5o o

300 250

100 50 0

300 250

100 50 0

10 100 return period (years)

300 250

2oo l5o loo 5o o

10 100 return period (years)

10 100 return period (years)

▲ observed data

Fig. 11.3 Flood frequency curves: comparison between the observed and the simulated peak discharge inferred by the actual observations (1989-2007) and for the emission scenarios A2 and B2 with different forecasting periods (2020, 2050, 2080)

## Post a comment