CLIMATE CHANGE IMPACTS ON RAINFALL INTENSITY- DURATION-FREQUENCY CURVES OF HYDERABAD, INDIA V. Agilan Department of Civil Engineering, National Institute of Technology, Warangal, Telangana, India-506004, Email: agilanvensiv@gmail.com N. V. Umamahesh Department of Civil Engineering, National Institute of Technology, Warangal, Telangana, India-506004, Email: mahesh@nitw.ac.in ABSTRACT The rainfall Intensity-Duration-Frequency (IDF) curves are commonly used in storm water management and other engineering design applications across the world and these curves are developed based on historical rainfall time series data by fitting a theoretical probability distribution to extreme rainfall series. In recent years, it has been widely reorganized that the extreme precipitation events are increasing due to global climate change. In addition, due to population and property concentration in relatively small areas, the flood damage potential in urban areas is high and the extreme rainfall events are the main cause for urban floods. Therefore, it is important to study the climate change impacts on rainfall IDF curves of an urban area. In this study, with the help of five Global Climate (KNN) weather generator based downscaling method, the impacts of climate change on rainfall IDF curves of Hyderabad city, India are studied. Results of this study indicate that the climate change is increasing extreme rainfall events of Hyderabad city. In addition, it is also observed that the return of period of an extreme rainfall of the Hyderabad city is reducing. INTRODUCTION The rainfall Intensity-Duration-Frequency (IDF) curves are commonly used in storm water management and other engineering design applications across the world (Endreny & Imbeah, 2009) and these curves are developed based on historical rainfall time series data by fitting a theoretical probability distribution of annual maximum rainfall series or partial duration series (Cheng & AghaKouchak, 2014). During the last activities (Prodanovic & Simonovic, 2007). This increase in greenhouse gas concentrations is causing largescale variations in atmospheric processes, which can then lead to changes in rainfall and temperature characteristics. Especially, recent studies show that the extreme rainfall events are intensifying due to global climate change (Allen & Ingram, 2002; Trenberth, et al., 2003; Emori & Brown, 2005; Tramblay, et al., 2012; Cavanaugh, et al., 2015; Xu, et al., 2015). Since the flood damage potential in urban areas is high due to population and property concentration in relatively small areas and the extreme rainfall events are the main cause for urban floods, it is essential to study the impacts of climate change on extreme rainfall characteristics of an urban area to reduce the flood damage of that city. Further, the changes in rainfall characteristics can change IDF curves. Anticipating the potential effects of climate change on rainfall IDF curves and adapting to them is one way to reduce vulnerability to adverse impacts (Prodanovic & Simonovic, 2007). In particular, the changes in extreme rainfall events can lead to a revision of standards for urban infrastructures (ex. storm water drainage networks). Current design standards are based on historical climate information. For example, primary storm water drainage network is designed to control a 10-year flood event. But, if the intensity and duration of the 10-year flood event increase in future, it will provide a significantly lower level of protection. Therefore, to prepare for future climate changes, it is crucial to study the impacts of climate change on rainfall IDF curves of a city. Hence, in this study, the impact of climate change on rainfall IDF curves of the Hyderabad city, India is studied. STUDY AREA AND DATA The Hyderabad city is the fourth biggest city in India and it is the capital of the southern Indian state of Telangana (Figure 1). The Hyderabad city lies between the latitude of 17.25o N and 17.60o N and longitude of 78.20o E and 78.75o E and situated at a height of about 500 m above the mean sea level. The average precipitation of the Hyderabad city is 818 mm/year. The wettest month of the Hyderabad city is August and the normal precipitation of this month is 163 mm (Agilan & Umamahesh, 2016). The Hyderabad city is grouped under semi-arid region and the Köppen-Geiger classification is BSh (Peel, et al., 2007). The hourly 222
observed rainfall data is procured from the India Meteorological Department (IMD) for the period of 01-01- 1972 to 31-12-2013 (42 years). This data is a gauge observation and it is observed at the center of the Hyderabad city i.e. 78.46o E and 17.45o N. From the hourly observations, the 2-Hr, 3-Hr, 6-Hr, 12-Hr, 18-Hr and 24-Hr duration rainfall are calculated. Figure 1: Location map of Hyderabad city (Agilan & Umamahesh, 2015) In addition, five GCM precipitation flux outputs (Table 1) for historical and future time periods are downloaded from CMIP5 website http://www.ipcc-data.org/sim/gcm_monthly/ar5/reference-archive.html (accessed during September and October 2015). Table 1: List of GCMs used and the latitude longitude selected for the study area. Model name Modelling center Latitude Longitude Future climate scenarios CanESM2 Canadian Centre for Climate modeling and Analysis 18.14 N 78.75 E RCP2.6, RCP4.5, RCP8.5 CCSM4 National Center for Atmospheric Research 17.43 N 78.75 E RCP2.6, RCP4.5, RCP6.0, RCP8.5 HadGEM2-ES Met Office Hadley Centre (additional HadGEM2- ES) 17.50 N 78.75 E RCP2.6, RCP4.5, RCP6.0, RCP8.5 MPI-ESM-LR Max Planck Institute for Meteorology 17.72 N 78.75 E RCP2.6, RCP4.5, RCP8.5 NorESM1-M Norwegian Climate Centre 18.00 N 77.50 E RCP2.6, RCP4.5, RCP6.0, RCP8.5 METHODOLOGY The aim of this study is to study the impacts of climate change on rainfall IDF curves of the Hyderabad city, India. The methodology adopted to achieve the aim of this study comprises of following three parts, 1. Developing historical observed rainfall IDF curves using extreme value analysis. 2. Developing future rainfall IDF curves with the help of Global Climate Model (GCM) simulations. 3. Identifying the impacts of climate change on rainfall IDF curves by comparing these two IDF curves in terms of return levels. OBSERVED RAINFALL IDF CURVE The return level of d-hour duration extreme rainfall for the given return period is calculated by fitting appropriate theoretical probability distribution to the d-hour duration extreme rainfall series (i.e. annual maximum series). Suppose the values X= x 1,x 2 x n be the annual maximum rainfall series of n independent and identically distributed (iid) random variable. The annual maximum series converges to Generalized Extreme Value (GEV) distribution and the cumulative distribution function is given by Eq. (1) (Coles, 2001; Katz, et al., 2002; Khaliq, et al., 2006; Nogaj, et al., 2007). 223
F x;,, 1/ x x exp 1, 0, 1 0, 0 x exp exp, 0, 0 (1) where µ, location, scale and shape parameters of the GEV distribution respectively. Once the parameters of the GEV distribution is estimated using the method of maximum likelihood parameter estimation method, the return level of the same d-hour duration rainfall for the given return period is calculated in terms of probability of occurrence and it is given by Eq. (2) (Coles, 2001; Cheng, et al., 2014). z T log 1 p 1, 0 log log 1 p, 0 (2) where p (the probability of occurrence) is the frequency (i.e., how frequently an extreme rainfall of specified intensity and duration is expected to occur). rainfall of specified intensity and duration having a probability of (Cheng & AghaKouchak, 2014). In this way, rainfall intensity is calculated as a function of the return period T, T=1/p (i.e., the average length of time between events of a given intensity and duration). Upon fitting GEV distribution to calculated for different return periods and the rainfall IDF curves are generated. The rainfall IDF curve of the Hyderabad city is developed using IMD observed hourly rainfall data and it is plotted in Figure 2. Figure 2: Rainfall IDF curve of the Hyderabad city from observed rainfall. FUTURE SIMULATED RAINFALL IDF CURVE GCM simulations offer possibilities of what might happen if the future development follows a certain course of action i.e. scenarios (Prodanovic & Simonovic, 2007) and it can be used to develop rainfall IDF curves for future rainfall conditions (Mailhot, et al., 2007; Prodanovic & Simonovic, 2007; Willems, 2013; Rodriguez, et al., 2014; Rupa, et al., 2015). In this study, five GCM outputs of different Representative Concentration Pathways (RCPs) scenario are considered. The list of GCMs and scenarios used in this study are given in Table 1. For each climate model, the grid point which is nearest to Hyderabad city rain gauge station is considered. Rainfall data from GCMs are only available at the low spatial resolution and cannot be applied directly to studies of climate impacts at the smaller scale (Willems, et al., 2012; Bi, et al., 2015). Downscaling of climate variable (in this case rainfall) is a technique that consists of deriving high-resolution 224
climate data from low-resolution GCM data. In this study, in order to obtain future downscaled rainfall time series, the observed for-the-day maximum rainfall series are first altered using change factor method (daily generator (Lall & Sharma, 1996; Prodanovic & Simonovic, 2007). For more information on daily scaling method and resampling using KNN algorithm, the interested reader is referred to Bi et al. (2015) and Prodanovic & Simonovic (2007) respectively. (a) (b) (c) (d) Figure 3: 2015-2056 average change factors for (a) RCP 2.6, (b) RCP 4.5, (c) RCP 6.0 and (d) RCP 8.5 scenario. (a) (b) (c) (d) Figure 4: 2057-2098 average change factors for (a) RCP 2.6, (b) RCP 4.5, (c) RCP 6.0 and (d) RCP 8.5 225
scenario. 2015-2056 average change factors calculated using daily scaling method for RCP 2.6, RCP 4.5, RCP 6.0 and RCP 8.5 scenarios are shown in Figure 3 and Figure 4 shows the same for the 2057-2098 time period. These change factors are then applied to the observed for-the-day maximum rainfall time series of different rainfall durations and the altered time series are further resampled using KNN algorithm. Towards developing future rainfall IDF curves, the annual maximum rainfall series is extracted from the KNN resampled datasets and these annual maximum rainfall series are fitted with GEV distribution (Eq. (1)). Further, return levels are calculated for different return periods with GEV distribution parameters which are obtained from the future extreme rainfall time series of different RCP scenarios. The return levels calculated with future downscaled rainfall data for different durations and return periods are given in Table 2 for both 2015-2056 and 2057-2099 time periods. In addition, the return levels obtained from the observed rainfall series are also given in Table 2 for different return periods and durations. Table 5: Return levels in mm/hr. 2015-2056 2057-2098 Observed RCP2.6 RCP4.5 RCP6.0 RCP8.5 RCP2.6 RCP4.5 RCP6.0 RCP8.5 2-Year 1-Hr 42.85 44.14 41.61 42.04 43.61 42.28 48.37 46.85 48.51 2-Hr 26.73 28.25 26.01 27.54 28.44 26.99 30.89 30.39 31.42 3-Hr 19.13 20.42 18.83 19.46 19.66 19.81 22.42 22.75 22.67 6-Hr 10.79 11.22 10.83 11.34 11.69 11.15 12.60 12.94 12.58 12-Hr 6.26 6.31 6.03 6.67 7.01 6.25 7.14 7.07 7.22 18-Hr 4.57 4.96 4.63 4.80 4.66 4.61 4.98 4.48 5.62 24-Hr 3.65 3.83 3.75 3.69 3.88 3.80 3.80 3.89 4.09 5-Year 1-Hr 52.27 53.87 50.55 52.33 54.31 52.59 60.76 63.48 59.64 2-Hr 32.89 34.86 31.45 32.99 34.81 33.51 39.15 40.65 39.31 3-Hr 24.29 25.59 23.70 24.89 24.88 24.94 28.57 30.15 28.63 6-Hr 14.03 14.72 14.00 14.51 15.10 14.60 16.76 17.62 16.59 12-Hr 8.44 8.72 8.16 8.65 9.01 8.33 9.98 9.93 10.02 18-Hr 6.38 6.84 6.23 6.55 6.57 6.42 7.23 6.85 7.59 24-Hr 5.21 5.34 5.12 5.26 5.41 5.38 5.74 6.04 5.90 25-Year 1-Hr 56.25 58.48 54.63 57.50 59.68 57.62 68.53 73.86 65.10 2-Hr 35.97 38.27 34.05 35.44 37.99 36.95 43.76 47.29 43.18 3-Hr 27.60 29.01 26.76 28.20 28.36 28.14 32.45 35.02 32.25 6-Hr 16.32 17.40 16.16 16.56 17.37 17.05 19.61 20.83 19.35 12-Hr 10.01 10.65 9.82 9.78 10.16 9.98 12.17 12.25 12.03 18-Hr 7.72 8.11 7.30 7.68 8.05 7.79 9.06 9.17 8.80 24-Hr 6.45 6.46 6.08 6.49 6.49 6.59 7.50 8.21 7.33 25-Year 1-Hr 59.64 62.83 58.34 62.60 64.96 62.47 77.88 86.30 70.39 2-Hr 39.01 41.74 36.51 37.64 41.12 40.50 48.76 55.50 46.94 3-Hr 31.68 33.35 30.46 32.10 32.79 31.96 37.15 41.14 36.49 6-Hr 19.37 20.30 18.95 19.12 20.22 20.37 23.33 25.03 22.98 12-Hr 12.16 12.55 12.27 11.03 11.43 12.41 15.35 15.77 14.78 18-Hr 9.59 9.74 8.67 9.06 10.19 9.75 11.88 13.38 10.24 24-Hr 8.30 8.00 7.36 8.30 7.94 8.34 10.48 12.26 9.46 226
COMPARISON Overall, from the Table 2, it is observed that the return of period of an extreme rainfall of the Hyderabad city is reducing. In particular, during 2015-2056 and under RCP 8.5 scenario, the 10 year return period future extreme rainfall of 1-hour duration is 59.68 mm/hr and it is nearly equal (59.64 mm/hr) to the 25 year return period observed extreme rainfall. Similarly, during 2057-2098 and under RCP 8.5 scenario, the 5 year return period future extreme rainfall of 1-hour duration is 59.64 mm/hr and it is equal (59.64 mm/hr) to the 25 year return period observed extreme rainfall. Therefore, from the results, it is observed that the return of period of an extreme rainfall of the Hyderabad city is reducing and the Hyderabad city drainage networks will fail more frequently than its actual design. SUMMARY In order to study the impacts of climate change on rainfall IDF curves of the Hyderabad city, rainfall IDF curves of two future time periods are developed with the help of five GCM (Table 1) simulations and KNN weather generator based downscaling method. For this study. The KNN weather generator is coded in the R programming language. Further, these IDF curves are compared with the IDF curves which are developed with observed rainfall of the Hyderabad city. Since the study area is an urban catchment and most of the urban infrastructure is designed for return level of 25 year or less, the observed and future IDF curves are compared in terms of 2, 5, 10 and 25 year return levels. From the comparison results, it is observed that the return period of an extreme rainfall of the Hyderabad city is reducing. In other words, if the IDF curve developed from the observed data is used for an infrastructure design of the Hyderabad city, the drainage networks will fail more frequently than its actual design. ACKNOWLEDGEMENTS This work is supported by the Information Technology Research Academy (ITRA), Government of India under, ITRA-water grant ITRA/15(68)/water/IUFM/01. We also thank the India Meteorological Department in this study. REFERENCES 1. Agilan, V. & Umamahesh, N. V., 2015. Detection and attribution of non-stationarity in intensity and frequency of daily and 4-h extreme rainfall of Hyderabad, India. Journal of Hydrology, Volume 530, p. 677 697. 2. Agilan, V. & Umamahesh, N. V., 2016. Modelling nonlinear trend for developing non-stationary rainfall intensity duration frequency curve. International Journal of Climatology. 3. Allen, M. R. & Ingram, W. J., 2002. Constraints on future changes in climate and the hydrologic cycle. Nature, Volume 419, pp. 224-232. 4. Bi, E. G., Gachon, P., Vrac, M. & Monette, F., 2015. Which downscaled rainfall data for climate change impact studies in urban areas? Review of current approaches and trends. Theoretical and Applied Climatology. 5. Cavanaugh, N. R., Gershunov, A., Panorska, A. K. & Kozubowski, T. J., 2015. The probability distribution of intense daily precipitation. Geophysical Research Letters, Volume 42. 6. Cheng, L. & AghaKouchak, A., 2014. Nonstationary Precipitation Intensity-Duration-Frequency Curves for Infrastructure Design in a Changing Climate. Nature: Scientific Reports, Volume 4, p. 7093. 7. Cheng, L., Kouchak, A. A., Gilleland, E. & Katz, R. W., 2014. Non-stationary extreme value analysis in a changing climate. Climatic Change, Volume 127, p. 353 369. 8. Coles, S., 2001. An introduction to Statistical Modelling of Extreme Values. London: Springer. 9. Emori, S. & Brown, S. J., 2005. Dynamic and thermodynamic changes in mean and extreme precipitation under changed climate. Geophys Res Lett, Volume 35, p. L17706. 10. Endreny, T. A. & Imbeah, N., 2009. Generating robust rainfall intensity duration frequency estimates with short-record satellite data. Journal of Hydrology, Volume 371, p. 182 191. 11. Katz, R. W., Parlange, M. B. & Naveau, P., 2002. Statistics of extremes in hydrology. Advances in Water Resources, 25(8), pp. 1287-1304. 12. Khaliq, M. N. et al., 2006. Frequency analysis of a sequence of dependent and/or non-stationary hydrometeorological observations: A review. Journal of Hydrology, Volume 329, pp. 534-552. 13. Lall, U. & Sharma, A., 1996. A nearest neighbor bootstrap for time series resampling. Water Resources Research, 32(3), p. 679 693. 227
14. Mailhot, A., Duchesne, S., Caya, D. & Talbot, G., 2007. Assessment of future change in intensity duration frequency (IDF) curves for Southern Quebec using the Canadian Regional Climate Model (CRCM). Journal of Hydrology, Volume 347, p. 197 210. 15. Nogaj, M., Parey, S. & Castelle, D. D., 2007. Non-stationary extreme models and a climatic application. Nonlin. Processes Geophys., Volume 14, p. 305 316. 16. Peel, M. C., Finlayson, B. L. & McMahon, T. A., 2007. Updated world map of the Koppen-Geiger climate classification. Hydrol. Earth Syst. Sci., Volume 11, p. 1633 1644. 17. Prodanovic, P. & Simonovic, S. P., 2007. Development of rainfall intensity duration frequency curves for the City of London under the changing climate. Water Resources Research Report-Report no. 058, Ontario, Canada: The University of Western Ontario. 18. Rodriguez, R. et al., 2014. Influence of climate change on IDF curves for the metropolitan area of Barcelona (Spain). International Journal of Climatology, 34(3), p. 643 654. 19. Rupa, C., Saha, U. & Mujumdar, P. P., 2015. Model and parameter uncertainty in IDF relationships under climate change. Advances in Water Resources, Volume 79, p. 127 139. 20. Tramblay, Y., Neppel, L., Carreau, J. & Sanchez-Gomez, E., 2012. Extreme value modelling of daily areal rainfall over Mediterranean catchments in a changing climate. Hydrological Processes, 26(25), p. 3934 3944. 21. Trenberth, K. E., Dai, A., Rasmussen, R. M. & Parsons, D. B., 2003. The changing character of precipitation. Bull. Amer. Meteor. Soc., 84(9), p. 1205 1217. 22. Willems, P., 2013. Revision of urban drainage design rules after assessment of climate change impacts on precipitation extremes at Uccle, Belgium. Journal of Hydrology, Volume 496, p. 166 177. 23. Willems, P., Arnbjerg-Nielsen, K., Olsson, J. & Nguyen, V., 2012. Climate change impact assessment on urban rainfall extremes and urban drainage: Methods and shortcomings. Atmospheric Research, Volume 103, p. 106 118. 24. Xu, L. et al., 2015. Precipitation trends and variability from 1950 to 2000 in arid lands of Central Asia. Journal of Arid Land, 7(4), pp. 514-526. 228