The Kerala Flood of 2018: Combined Impact of Extreme Rainfall And
Total Page:16
File Type:pdf, Size:1020Kb
Hydrol. Earth Syst. Sci. Discuss., https://doi.org/10.5194/hess-2018-480 Manuscript under review for journal Hydrol. Earth Syst. Sci. Discussion started: 14 September 2018 c Author(s) 2018. CC BY 4.0 License. The Kerala flood of 2018: combined impact of extreme rainfall and reservoir storage Vimal Mishra1*, Saran Aadhar1, Harsh Shah1, Rahul Kumar1, Dushmanta Ranjan Pattanaik2, Amar Deep Tiwari1 5 1Civil Engineering, Indian Institute of Technology Gandhinagar, Gandhinagar, 382355, India 2India Meteorological Department, New Delhi, 110003, India Correspondence to: Vimal Mishra ([email protected]) Abstract. Extreme precipitation events and flooding that cause losses to human lives and infrastructure have increased under the warming climate. In August 2018, the state of Kerala (India) witnessed large-scale flooding, which affected millions of 10 people and caused 400 or more deaths. Here, we examine the return period of extreme rainfall and the potential role of reservoirs in the recent flooding in Kerala. We show that Kerala experienced 53% above normal rainfall during the monsoon season (till August 21st) of 2018. Moreover, 1, 2, and 3-day extreme rainfall in Kerala during August 2018 had return periods of 75, 200, and 100 years. Six out of seven major reservoirs were at more than 90% of their full capacity on August 8, 2018, before extreme rainfall in Kerala. Extreme rainfall at 1-15 days duration in August 2018 in the catchments upstream 15 of the three major reservoirs (Idukki, Kakki, and Periyar) had the return period of more than 500 years. Extreme rainfall and almost full reservoirs resulted in a significant release of water in a short span of time. Therefore, above normal seasonal rainfall (before August 8, 2018), high reservoir storage, and unprecedented extreme rainfall in the catchments where reservoirs are located worsened the flooding in Kerala. Reservoir operations need to be improved using a skilful forecast of extreme rainfall at the longer lead time (4-7 days). 20 1 Introduction Extreme precipitation events, landslides, and floods are the most common natural disasters that affect human society and economy (Coumou and Rahmstorf, 2012; Crozier, 2010; Hirabayashi et al., 2008; Roxy et al., 2017). Frequent extreme precipitation events cause flooding (Fowler et al., 2010), which have become common in India (Mohapatra and Singh, 2003). The frequency of great floods and extreme precipitation events has substantially increased under the warming climate, 25 which is consistent with the observations as well as climate model projections (Ali and Mishra, 2018; Milly et al., 2002). India has witnessed some of the most unprecedented extreme precipitation events that caused flooding and loss of lives in the recent past. For instance, extreme precipitation in Uttarakhand in 2013 resulted in large flooding with the death of more than 6000 people and economic loss of more than 3.8 billion USD ("Rapidly Assessing Flood Damage in Uttarakhand, India,” 2014). Houze et al. (2017) argued that the 2013 extreme precipitation event was different than the historic flood 30 producing events that occurred in the Himalayan region and was not caused by the convective storm. The heavy rain event 1 Hydrol. Earth Syst. Sci. Discuss., https://doi.org/10.5194/hess-2018-480 Manuscript under review for journal Hydrol. Earth Syst. Sci. Discussion started: 14 September 2018 c Author(s) 2018. CC BY 4.0 License. occurred in 2015 caused flooding in Chennai and led to the estimated damage of $3 billion (van Oldenborgh et al., 2016). Similarly, heavy rain in Mumbai in 2005 caused the death of more than 1000 people (Kumar et al., 2008). Extreme precipitation and flooding have become among the costliest natural disasters in India and other regions of the globe (The Human Cost of Natural Disasters 2015: A Global Perspective). Human losses from flooding are projected to increase by 70- 5 80% if the global mean temperature increases above 1.5C from the pre-industrial level (Dottori et al., 2018). Moreover, Dottori et al., (2018) reported that the future flood impacts are likely to have uneven regional distribution, with the highest losses are to occur in Asia. The recent extreme rainfall and widespread flooding in Kerala exemplify the enormity of extreme rainfall and large-scale floods in India. The persistent and extreme rainfall occurred in August 2018 in Kerala affected all the aspects of human lives 10 including socioeconomic conditions, transportation, infrastructure, agriculture, and livelihood. The Kerala flood of 2018 has already attracted attention from the media, scientific community, and policymakers, which is probably the worst flood in a century (The Independent, 16 August 2018). As per the preliminary estimates, the Kerala flood caused the death of more than 440 people (Gulf News, 30th August 2018) and economic damage exceeding $3 billion (News18, 17 August 2018). Despite the state-wide extreme rainfall in Kerala in August 2018, potential causes (heavy rain and reservoir operations) of 15 floods have been greatly debated. Here, we provide the first assessment of the anomalous nature of extreme rainfall and the role of major reservoirs that might have contributed in the flooding. 2 Data and Methods We obtained daily observed rainfall data from India Meteorological Department (IMD) at 0.25° for the period 1901-2018 (till 21st August). IMD rainfall dataset is developed using more than 6000 observing gauge stations across India, and a 20 substantial number of stations are located in Kerala. Station based rainfall observations were interpolated using Inverse Distance Weighting interpolation (Pai et al., 2014; Shepard, 1968). The gridded dataset captures climatological, orographic, and other features associated with the Indian summer monsoon rainfall over India (Mishra et al., 2014; Shah and Mishra, 2015). Moreover, heavy rainfall variability over the Western Ghats and foothill of Himalaya region is well represented (Pai et al., 2014). Gridded daily rainfall from IMD has been widely used in hydro-climatic studies as well as for analysis of 25 extreme rainfall over India (Ali et al., 2014; Ali et al., 2018; Kumar et al., 2013; Mishra et al., 2016; Shah & Mishra, 2015). Using the daily rainfall gridded data, we estimated mass curve (the relationship between cumulative rainfall and time) and depth-duration- frequency (DDF) curve to evaluate the severity and intensity of extreme rainfall during August 2018 in Kerala. DDF curves are essentially similar to the intensity-duration-frequency (IDF) curves but with rainfall depth instead of intensity. The mass curve and DDF curves are widely used to extract the information of a storm event (Borga et al., 2005; 30 Overeem et al., 2009). One of the important questions that the Kerala (Fig. S1) flood event poses is if the extreme rain event was once in a century (with the probability of exceedance 1% or less) event or not. Therefore, we estimated return period of heavy rain occurred in 2 Hydrol. Earth Syst. Sci. Discuss., https://doi.org/10.5194/hess-2018-480 Manuscript under review for journal Hydrol. Earth Syst. Sci. Discussion started: 14 September 2018 c Author(s) 2018. CC BY 4.0 License. August 2018 for 1-15 days durations. We used Generalized Extreme Value (GEV) distribution (Katz et al., 2002, 2005), which has been widely applied to analyze extreme hydro-climatic events (Ali et al., 2014; Min et al., 2011; Mishra et al., 2016). We fit the GEV distribution to annual maximum rainfall for the selected durations. The GEV distribution has three parameters (location, scale, and shape) that were estimated using the Maximum Likelihood method. Using the GEV 5 distribution, we developed DDF curves for 1-15 days extreme rainfall duration for 20, 50, 100, 200, and 500 years return period. We tested the goodness of fit of GEV distribution using QQ plots (Fig. S2) and Chi-Square test, and we found that the GEV distribution passed the goodness-of-fit test for the most of the rainfall durations (Table S1). Furthermore, we also estimated the anomaly of cumulative rainfall during the flood to quantify the rainfall departure from its mean estimated for the reference period of 1901-2017. 10 To analyze the reservoir condition before the flood event, we obtained reservoirs storage information for the seven major reservoirs (Idukki, Idamalyar, Kakki, Kallada, Malampuzha, Parambikulam, and Periyar) from India Water Resources Information System (WRIS) (IWRIS, www.india-wris.nrsc.gov.in) for the period 2007-2018. Using the IWRIS data, we analyzed the abnormality in the reservoir storage before the flood in Kerala. There is a number of small and medium size reservoirs in Kerala; however, as their storage information is not available, we mainly focus on the major reservoirs. 15 3 Results First, we analyze the mass curve of rainfall between May 1 and April 30 for the period of 1901-2017 (Fig. 1a). We considered the period from May 1 as the onset of the summer monsoon happens mostly in May in Kerala. We find that the long-term (1901-2017) average annual rainfall in Kerala is about 2400 mm with a standard deviation of 400 mm. The 117 years observed record of rainfall in Kerala shows that the two wettest years occurred in 1924 and 1961 with the annual 20 rainfall of about 3600 mm (Fig. 1a). Rainfall during August 2018 in Kerala departs significantly from the averaged mass- curve. However, the 2018 rainfall (till August) in Kerala is lower than the observed rainfall in 1924 and 1961. We estimated DDF curves for extreme rainfall averaged over the entire state of Kerala for the 1-15 days duration and 50-500 year return period using GEV distribution fitted on annual maximum rainfall for the 1901-2017 period (Fig 1b, Table S2, and S3). The 117 years record of observed rainfall provides us a basis to estimate the return period of extreme rainfall during 25 August 2018 (Table S3 and S4).