Emerging issues in stormwater management connected to the potential precipitation changes in Italy Questions émergentes pour la maîtrise des eaux pluviales liées au changement possible des précipitations en Italie

Sergio Papiri, Carlo Ciaponi, Sara Todeschini Department of Hydraulic and Environmental Engineering Università degli Studi di Pavia, Via Ferrata, 1, 27100 Pavia, Italy [email protected] – [email protected] – [email protected]

RESUME Cet article présente les premiers résultats d’une recherche sur le comportement des précipitations en Italie et se concentre sur les conséquences possibles du changement des précipitations par rapport à la gestion des eaux de pluie. Des analyses de tendances des précipitations au cours des siècles derniers ont été effectuées sur une base annuelle, saisonnière et quotidienne dans différentes stations de l’Italie (Domodossola, Turin, Milan, Pavia, Genova, Mantoue, Bologne, Florence et Palerme). L’étude inclut l’analyse des précipitations maximum annuelles de différentes durées (1, 3, 6, 12 et 24 heures) pendant environ cinquante ans dans neuf stations (Domodossola, Turin, Voghera, Mantoue, Bologne, Florence, Livourne, Naples et Palerme).

ABSTRACT This paper presents the first results of a broad-based study of precipitation behaviour in Italy and focuses on the possible consequences of precipitation changes on urban stormwater management. Analyses of the yearly, seasonal and daily precipitation over the last few centuries have been carried out at various stations throughout Italy (Domodossola, Turin, Milan, Pavia, Genoa, Mantua, Bologna, Florence and Palermo). The study includes the analysis of annual maximum rainfall of different durations (1, 3, 6, 12 and 24 hours) over a period of about fifty years for nine stations (Domodossola, Turin, Voghera, Mantua, Bologna, Florence, Livorno, Naples and Palermo).

KEYWORDS Annual maximum rainfall of different durations, daily precipitation series, precipitation change, stormwater management

NOVATECH'2007 1 1 INTRODUCTION Analyses of yearly and seasonal precipitation trends over the last few centuries have been carried out on global hemispheric and regional scales. They indicate a small positive global trend in precipitation over land during the 20th century, but also show that this pattern is not homogenous, as large areas are characterized by negative trends (IPCC, 1996 and 2001). As far as Europe is concerned, the data seems to suggest an increase in precipitation in the northern areas and a decrease in southern areas (Schonwiese and Rapp, 1997). Studies based on Italian station databases, confirm the negative trend over Italy, also giving evidence that it is present both over the last decades (Piervitali, 1998) and the last 100-150 years (Palmieri et al., 1991; Siani and Palmieri, 1993; Buffoni et al., 1999; Brunetti et al., 2004). The understanding of total precipitation changes over long periods of time (annual or seasonal) is useful for many applications, but does not reveal important information such as potential changes in daily precipitation distribution, in the number of wet days or in the dry periods over a year or a season. However, the lack of detailed information regarding the behaviour of daily precipitation greatly limits the design and management of adequate and properly functioning stormwater drainage systems, flow control systems (detention basins, flood control reservoirs, etc.) and of stormwater quality control measures (sewer overflows, stormwater storage tanks, etc.). In Italy long and homogeneous daily precipitation series are available for some areas but there are only a few studies (Brunetti et al., 2000 and 2004) on these series. Within this context, the first purpose of this study is to investigate potential tendencies in the available daily precipitation series. A more detailed analysis regarding potential tendencies is obtained using the non- parametric Mann-Kendal test (Sneyers, 1990). The study includes the analysis of series of the annual maximum rainfall of different durations (1, 3, 6, 12 and 24 hours) over a period of about fifty years at various stations. This study is of particular value because understanding extreme rainfall properties is important for the designing of efficient urban drainage and flow control structures. The uncertainty associated with the evaluation of such properties is relevant and has a direct effect on the choice of the extent of urban stormwater control measures required. In the majority of cases, the uncertainty regarding extreme rainfall properties is due to the small number of observations and to the non-uniform spatial distribution of extreme rainfall events (Schilling, 1991; Desbordes et al. 1997; Berne et al., 2004; Mikkelsen et al., 2005; De Toffol et al., 2005). Non-stationarity of extreme rainfall properties can also increase the uncertainty (Bonaccorso et al., 2005; Rauch and De Toffol, 2005). A further purpose of this study, therefore, is to inquire into the existence of possible trends in these properties over the last decades in Italy.

2 RESULTS Figure 1 shows the geographical location of the twelve stations where the precipitation series of this database were collected.

2 NOVATECH'2007 47 Domodossola Milan 45 Turin Mantua Voghera Pavia GenoaBologna 43 Livorno Florence

41 Naples

39 Palermo

37

35 6 81210 14 16 18

Figure 1 Geographical location of the twelve stations 2.1 Yearly and seasonal precipitation trends Analyses of yearly and seasonal precipitation have been carried out for various ultrasecular series throughout Italy ([1] Domodossola 1872-1996, [2] Turin 1787- 1994, [3] Milan 1858-2005, [4] Pavia 1812-2002, [5] Genoa 1833-1996, [6] Mantua 1868-1997, [7] Bologna 1813-20031, and [8] Palermo 1866-1999). Table 1 shows the arithmetic mean (m), the standard deviation (s), the percentage variation (v)2 and the probability of trend (P) obtained using the Mann-Kendall test for the precipitation height h [mm] and the number of wet days n [d]. On a yearly basis, a negative trend in precipitation is evident both for the North and the South (very high probability for Milan, Turin and Palermo). A strongly decreasing trend is clear in the annual (and above all autumnal) number of wet days (very high probability for Milan, Turin and Domodossola). The negative trend in the number of wet days is greater and more significant than that of total precipitation, both on a yearly basis and in all seasons. As a consequence, precipitation intensity has a positive trend. 2.2 Daily precipitation trends Analyses of daily precipitation are carried out for five series ([3] Milan 1858-2005, [5] Genoa 1833-1996, [6] Mantua 1868-1997, [7] Bologna 1813-2003 and [8] Palermo 1866-1999). Table 2 shows the arithmetic mean (m), the percentage variation (v) and the probability of trend (P) obtained using the Mann-Kendall test for the yearly and seasonal mean precipitation per day [mm]. Yearly mean precipitation per day presents a strongly increasing tendency (very high probability for Milan, Genoa and Mantua). In particular, mean precipitation per day has a clear positive trend in summer and in autumn. Statistical evaluation of trends in precipitation per rain day has also been carried out dividing the precipitation into classes: C1 daily precipitation (DP) between 0 and 5 mm, C2 DP between 5 and 10 mm, C3 DP between 10 and 20 mm, C4 DP between 20 and 50 mm. Table 3 shows the percentage variation (v) and the probability of trend (P) obtained using the Mann-Kendall test for the annual number of wet days in every

1 series not homogenous (Buffoni et al. 1999). 2 v =(FV-IV)/IVx100; FV and IV respectively final and initial values of the tendency line. NOVATECH'2007 3 class.

Year Spring Summer Autumn Winter h n h n h n h n h n [1] m 1426 97 431 29 337 27 452 25 207 17 s 400 17 200 7 132 6 246 8 128 7 v -2.5 -22 -2 -21 -8 -21 +3.9 -25 -7.6 -23 P 34 100 18 100 67 100 6.8 100 5 100 [2] m 130 82 285 25 224 22 241 20.8 130 14 s 82 25 120 10 98 9 127 8.5 82 6.8 v -13 -41 -9.2 -38 -16 -46 -15 -45 -11.8 -34 P 97 100 59 100 92 100 79 100 63 100 [3] m 978 86 266 25 229 19 297 23 186 19 s 220 14.7 108 7 88 5 126 7 87 7 v -12.4 -19 -23.7 -23 +3.3 -8 -14.4 -26 -9.9 -15 P 93.6 100 97.6 100 33.8 95 86.7 100 41.1 98 [4] m 819 n.a. 220 n.a. 167 n.a. 258 n.a. 174 n.a. s 193 n.a. 89 n.a. 75 n.a. 115 n.a. 79 n.a. v +9.4 n.a. 17 n.a. +13 n.a. +2 n.a. +8 n.a. P 88 n.a. 90 n.a. 69 n.a. 3 n.a. 40 n.a. [5] m 1294 85 293 24 167 13 497 25.5 336 22.3 s 304 13 123 6.6 92 4.4 210 6.9 154 6.8 v -0.3 -16 +0.7 -15 -3.7 -12 -3.3 -24 +5.4 -7.9 P 25.6 100 37.5 100 36.3 099 53 100 39.9 94 [6] m 675 81 174 24 165 17 197 21 138 19 s 141 13 59 6 71 5 78 6 60 6.5 v -5.6 -14 -16.7 -17 -1.7 -19 -7.5 -18 +8.2 +2.7 P 61.9 100 92 100 24 100 34 99 38 29 [8] m 538 70 110 17 28 3.6 191 21 208 29 s 129 12 48 5.8 29 2.6 78 5.5 83 8 v -16 -16 -18 -23 -9 -20 -9.6 -13 -21 -13 P 99 100 91 100 54 99 45 99 99 98 Table 1 Main characteristics of yearly and seasonal precipitation and their trends

Year Spring Summer Autumn Winter m v P m v P m v P m v P m v P [3] 11 +8 86 10 -4 55 12 +15 90 13 +21 94 9 +2 12 [5] 15 +19 100 12 +21 99 13 +20 93 19 +30 89 15 +13 87 [6] 8 +10 99 7 -1 38 9 +18 94 9 +16 89 7 -1 53 [8] 8 -1 24 6 +4 28 7 +17 82 9 +5 46 7 -10 83 Table 2 Yearly and seasonal mean precipitation per day and their trends

4 NOVATECH'2007

C1 C2 C3 C4 v P v P v P v P [3] -16 100 -25 100 -30 100 -5 87 [5] -24 100 -18 100 -7 97 -9 98 [6] -22 100 -9 96 +0.4 65 -5 93 [8] -9 93 -25 100 -25 100 -22 100 Table 3 Trends in annual number of wet days in every precipitation class

There is a strongly decreasing trend in annual number of wet days of every precipitation class, especially for classes C1 and C2 (very high probability for Milan, Genoa and Palermo). 2.3 Trends in annual maximum rainfall of different durations Analyses of annual maximum rainfall series of different durations (1, 3, 6, 12 and 24 hours) have been carried out for nine stations ([A] Domodossola 1930-1998, [B] Turin 1928-1994, [C] Mantua 1933-1993, [D] Voghera 1935-2002, [E] Bologna 1935-1989, [F] Florence 1950-1996, [G] Livorno 1933-1997, [H] Naples 1950-1996, [I] Palermo 1953-2000). Table 4 presents the percentage variation (v) and the probability of trend (P) obtained using the Mann-Kendall test in annual maximum rainfall [mm] of different durations. An increasing trend is present in annual maximum rainfall of different durations (1, 3, 6, 12 and 24 hours) for almost all the stations. In particular, the positive tendency is greater for the durations of 6, 12 and 24 hours. Figure 2 shows the relationships between Depth and Duration (with a 10 year Return Period, T) obtained by dividing the available data for each series into two sets. Depth and duration points with a 10 year return period are greater in the last period than in the first in northern areas whilst a trend is not evident in the centre and in the south of Italy where moreover the analysed stations are scarce. The study also considers depth duration points with return periods of 2, 5 and 20 years. These points (omitted for reasons of space) confirm the same behaviour.

1 hour 3 hours 6 hours 12 hours 24 hours v P v P v P v P v P [A] +6 91 -5 91 +8 92 +10 98 +15 99 [B] +3 42 +17 27 +27 39 +32 41 +35 44 [C] +8 73 +12 67 +0.6 40 +2 66 +13 76 [D] -7 26 +11 18 +14 32 +19 21 +18 5 [E] -8 5 +42 91 +37 92 +39 94 +25 84 [F] +26 33 +25 70 +19 86 +19 80 +10 76 [G] -3 92 -6 90 -6 90 -18 97 -18 98 [H] +28 71 +16 33 +26 63 +41 76 +48 85 [I] -16 37 -24 28 -21 13 -6 47 +14 84 Table 4 Trends in the annual maximum rainfall of different durations

NOVATECH'2007 5 200 1930-1963 120 1928-1960 1964-1998 1961-1994 150 80 100 h [mm] h [mm] 40 50

0 0 01020300102030 [A] t [hour] [B] t [hour] 120 1933-1962 160 1935-1968 1963-1993 1969-2002 90 120

60 80 h [mm] h h [mm] 30 40

0 0 0 1020300102030 t [hour] [C] t [hour] [D] 120 1935-1961 120 1950-1972 1962-1989 1973-1996 90 80 60 h [mm] h [mm] 40 30

0 0 0 1020300102030 [E] t [hour] [F] t [hour] 180 1933-1964 160 1950-1972 1965-1997 1973-1996 135 120

90 80 h [mm] h [mm] 45 40

0 0 0 102030 0102030 [G] t [hour] [H] t [hour] 120 1953-1976 1977-2000 90

60 h [mm] 30

0 0102030 [I] t [hour] Figure 2 Depth-Duration points for different sets of data (T = 10 years)

6 NOVATECH'2007

3 DISCUSSION 3.1 Effects on flow control systems Up until today stormwater drainage system design guidelines are frequently based on extreme rainfall properties. The non-stationarity of these properties can have a direct effect on the suitable dimension of sewer pipes. This analysis only considers the annual maximum rainfall series of the minimum duration of 1 hour. However, seeing that critical durations for sewers are often shorter than 1 hour, possible trends in annual maximum rainfall series of durations shorter than 1 hour must be investigated in order to design adequate and properly functioning stormwater drainage systems. The tendency for the annual maximum rainfall values of short durations to increase causes higher peak flows, so the existing detention facilities (i.e. detention basins and flood control reservoirs) could result insufficient for smoothing out peak discharges in the receiving water bodies and preventing floods. 3.2 Effects on stormwater quality control measures The highly significant negative trend of the annual number of wet days together with the small negative trend in annual precipitation indicates more intense rainfall events. This is a critical factor to be taken into account when planning stormwater quality control measures. In fact, as the rainfall events are weaker, control sewer overflows (CSOs) are better at reducing number of discharges, volume and pollutant mass to the receiving water bodies. Moreover, the ability of stormwater storage tanks with fixed capacity to capture rainfall volume and pollutant mass decreases if the precipitation height of events increases. A negative tendency in the number of wet days also implies a longer time between two events and so more deposits are formed in combined sewers. The presence of sediment deposits in combined sewer pipes causes both hydraulic and environmental problems, such as the reduction of flow capacity and stronger first flush pollution discharges through CSOs into water bodies.

4 CONCLUSION This paper reports the main characteristics and the time tendencies of different (yearly, seasonal, daily and annual maximum of 1, 3, 6, 12 and 24 hours) precipitation series for various stations throughout Italy. The observed precipitation series exhibit trends over time and these detected tendencies directly affect the design and management of properly functioning stormwater drainage systems, flow control systems (detention basins, flood control reservoirs, etc.) and of stormwater quality control measures (sewer overflows, stormwater storage tanks, etc.). These results show that the consolidated engineering practice that assumes the stationarity for hydrological processes should be revised. Traditional design concepts, such as return period, should be adapted to non- stationary data and it is also necessary to gauge the extent to which a trend that has been observed in the past can be assumed to continue into the future. For this reason it is also necessary to improve statistical methods to identify the main characteristics and the nature of detected trends more clearly.

NOVATECH'2007 7 LIST OF REFERENCES Berne A., Delrieu G., Creutin J.-D. and Obled C. (2004) Temporal and spatial resolution of rainfall measurements required for urban hydrology. Journal of Hydrology, 299: 166-179. Bonaccorso B., Cancelliere A., Rossi G. (2005). Detecting trends of estreme rainfall series in Sicily. Advances in Geosciences 2: 7-11. Brunetti M., Buffoni L., Maugeri M., Nanni T. (2000). Precipitation intensity trends in Northern Italy. International Journal of Climatology 20: 1017-1031. Brunetti M., Buffoni L., Mangianti F., Maugeri M., Nanni T. (2004). Temperature, precipitation and extreme events during the last century in Italy. Global and Planetary Change 40: 141-149. Buffoni L., Maugeri M., Nanni T. (1999) Precipitation in Italy from 1833 to 1996. Theoretical and Applied Climatology 63: 33-40. Chocat B., Ashley R., Marsalek J., Matos M.R., Rauch W., Shilling W., Urbonas B. (2004) Urban Drainage – Out-of-sight-out-of-mind? Closing Conference Proceedings of 5th Int. Conference on Sustainable Techniques and Strategies in Urban Water Management, Lyon, France, 6-10 June 2004. De Toffol S., Bürstrom De Simon Y. and Rauch W. (2005) On the effect of spatial variances of historical rainfall time series to CSO performance evaluation (submitted). 10th International Conference on Urban Drainage, 21-26 August 2005, Copenhagen /Denmark. Desbordes M., Masson J.M., Neppel L. (1997) Spatial extension of extreme rainfall events: return period of isohyets area and influence of rain gauges network evolution. Atmosferic Research 45: 183-199. IPCC (1996) Climate change. The IPCC Second Assessment Report [Houghton J.T., Meira Filho L.G., Callander B.A., Harris N., Kattenberg A., Maskell K.] Cambridge University Press: New York, NY. I.P.C.C. (2001) Climate Change 2001. The Scientific Basis. Contribution of Working group I to the Third Assessment Report of the Intergovernmental Panel on Climate Change. [Dai, X.; Ding, Y.; Griggs, D.J.; Houghton, Johnson, C.A.; J.T.; Maskell, K.; Noguer, M.; Van Der Linden, P.J.]. Cambridge University Press, Cambridge, United Kingdom and New York, NY, USA. Mikkelsen P. S., Madsen H., Arnbjerg-Nielsen K., Rosbjerg D. and Harremoes P. (2005) Selection of Regional Historical Rainfall Time Series as Input to Urban Drainage Simulations. Atmospheric Research 77: 4-17. Palmieri S., Siani A.M., D’Agostino A. (1991) Climate fluctuations and trends in Italy within the last hundred years. Annales Geophysicae 9: 769-776. Piervitali E., Colacino M., Conte M. (1998) Rainfall over the central-western Mediterranean basin in the period 1951-1995. Part I: precipitation trends. Il Nuovo Cimento 21C: 331-344. Rauch W., De Toffol S., (2005) On the issue of trend and noise in the estimation of extreme rainfall properties. 10th Conference on Urban Drainage, 21-26 August 2005, Copenhagen/Denmark. Schonwiese C.D., Rapp J. (1997) Climate Trend Atlas of Europe Based on Observations 1891- 1990. Kluwer Academic Publishers: Dordrecht. Shilling W. (1991) Rainfall data for urban hydrology: what do we need? Atmospheric Research 27:5-21. Siani A.M., Palmieri S. (1993) Climate scenarios derived from historical rainfall time series in Italy. Bollettino Geofisico 1:765-776. Sneyers R. (1990) On the statistical analysis of the series of observation. WMO, Technical Note N. 143, Geneva.

5 ACKNOWLEDGEMENTS This research has been financed by MIUR 2005. Special thanks are given to Ferdinando Brambilla for his contribution to this research while working on his thesis.

8 NOVATECH'2007