3. Precipitation Changes in India

The paleoclimate proxy data of monsoonal record of the past 640,000 years suggest that the millennial-scale variability arose by the solar insolation changes which are caused by precession and obliquity (Cheng et al. 2016). In the last 11,000 years (The Holocene age), summer monsoon is declining with variability at the multi-decadal scale to centennial scales (Chao and Chen 2001). The decline of the summer monsoon is linked to, among other factors, the southward migration of the ITCZ as a result of the decrease in solar insolation (Fleitmann et al. 2007). Additionally, prolonged wet/drought periods of multi-decadal and century-scales have occurred during the last 4000 years (Sinha et al. 2011; Prasad et al. 2014) with notable century scale long declining trends 1550–2200 years BP (Roman Warm Period; RWP) and 100–550 years BP (Little Ice Age; LIA) and an increasing trend during 650–1050 year BP (Medieval warm period; MWP) in summer monsoon (Trends marked by arrows; Fig. 3.2). Abrupt changes in monsoon around 2800 years BP and 2350 years BP have been attributed to solar variability (Sinha et al. 2018). Tree ring-based studies from the Himalayan region reveal a declining trend in the summer monsoon over the last 200 years with the possible linkages to large-scale greenhouse warming, and anthropogenic aerosol emissions (Xu et al. 2013; Shi et al. 2017). Speleothem (cave deposits)-based 4000-year-long monsoon reconstructions from central, peninsular and northeast India indicate that the monsoon has undergone multi-decadal changes of larger magnitude in the Holocene than in the last 200 years alone (Fig. 3.2). Therefore, as per the available paleo records, changes in the monsoon due to the regional forcing such as anthropogenic aerosol emissions are difficult to detect against the large multi-decadal natural variability.

The mean summer monsoon rainfall (JJAS) over India from 1979 to 2005 from multiple observational datasets is shown in Fig. 3.3. It is noted that, in general, all datasets show a high rainfall zone over east-central India and low rainfall zones over northwest India, northern parts of Kashmir, and the rain shadow area of southeast India. The discrepancy in observations occurs mainly over northern parts of India, Himalayan region (e.g., Prakash et al. 2015) and northeast India (Bidyabati et al. 2017).

The annual rainfall averaged over Indian landmass does not show any trend over the period 1901–2015. However, in the recent period 1951–2015 as well as 1986–2015 the annual rainfall series shows decreasing trend (though not statistically significant, or evident in all the datasets). The summer monsoon rainfall series averaged over India landmass does not show any long-term trend on a century-scale where it has been found that the contribution from increasing heavy rain events has been offset by decreasing moderate rain events (Goswami et al. 2006). However, a downward trend of rainfall over the Indian subcontinent has been observed in the period 1951–2004 (Kulkarni 2012). The decreasing tendency of summer precipitation is found to have accelerated during 1971–2002 (Kothawale et al. 2008) and is evident in multiple datasets (Annamalai et al. 2013). This secular decline in mean rainfall is attributed to wakening monsoon Hadley circulation. A decreasing trend in rainfall may also be a result of multi-decadal epochal variability associated with an east-west shift in monsoon rainfall due to anomalous warming of the Indo-Pacific warm pool (Annamalai et al. 2013; Guhathakurta et al. 2015) or weakening of the land-ocean temperature gradient (Kulkarni 2012). The drying trend may be attributed to a number of factors such as: (a) increasing anthropogenic aerosol concentration over northern hemisphere which may cool northern hemisphere and slowing of the tropical meridional overturning circulation (e.g. Ramanathan et al. 2005; Chung and Ramanathan 2006; Bollasina et al. 2011, 2014); (b) to increasing trend of the Pacific Decadal Oscillation (Salzmann and Cherian 2015); (c) to significant weakening of monsoon low-level southwesterly winds, the upper-tropospheric tropical easterlies from the out-flow aloft, the large-scale monsoon meridional overturning circulations (Rao et al. 2010; Joseph and Simon 2005; Sathiyamoorthy 2005; Fan et al. 2010; Krishnan et al. 2013); (d) to strong Atlantic Multidecadal Oscillation (AMO) which weakens the meridional temperature gradient resulting in early withdrawal of monsoon over India and thus reducing the mean rainfall (Goswami et al. 2006); (e) to significant increase in the duration and frequency of ‘monsoon- breaks’ (dry spells) over India since the 1970s (e.g. Ramesh Kumar et al. 2009; Turner and Hannachi 2010);(f) to rapid warming of western Indian ocean which reduces the meridional temperature gradient dampening the monsoon circulation (Gnanaseelan et al. 2017; Roxy 2015; Roxy et al. 2015) and (g) to changes in land use/land cover (Niyogi et al. 2010; Pathak et al. 2014; Paul et al. 2016; Krishnan et al. 2016). However, enhanced warming over Indian subcontinent and comparatively slower rate of warming over India Ocean has favoured the land-ocean temperature gradient in the recent decade (2002–2014) and helped a possible short-term revival of the monsoon over India at the rate of 1.34 mm/day/decade (Jin and Wang 2017). Thus, over the recent three decades (1986–2015) all-India summer monsoon shows a decreasing tendency, but the decline is not statistically significant.

There is considerable spatial variability in precipitation changes. As compared to the period 1901–1975, rainfall has reduced by 1–5 mm/day during 1976–2015 over central parts of India (the core monsoon zone), Kerala and extreme northeastern parts and increased over the Jammu and Kashmir region as well as in parts of western India (Kulkarni et al. 2017). Regional anthropogenic forcings such as from aerosols and land-use change from urbanization and agricultural intensification could be dominant contributors to this recent spatial variability (Paul et al. 2018).

Trends in Indian rainfall records have been extensively studied, but the subject remains complicated by the high spatiotemporal variability of rainfall arising from complex atmospheric dynamics and, to some extent, differences that emerge from the methods used in creating the datasets. Monsoon rainfall has shown moderate increasing trends in 27 (out of 36) subdivisions across India (Guhathakurta and Rajeevan 2008). The linear trend in annual as well as seasonal rainfall shows a statistically significant decreasing trend over Jharkhand, Chhattisgarh, and Kerala, and eight subdivisions, viz. Gangetic WB, West UP, Jammu and Kashmir, Konkan and Goa, Madhya Maharashtra, Rayalaseema, Coastal AP and North Interior Karnataka show increasing trends (Guhathakurta and Rajeevan 2008). Based on high-resolution gridded data for 1901–2015, there are statistically significant decreasing trends in annual as well as seasonal rainfall over Kerala, Western Ghats and some parts of central India including Uttar Pradesh, Madhya Pradesh, and Chhattisgarh as well as some parts of the northeastern states. Whereas rainfall over Gujarat, Konkan coast, Goa, Jammu and Kashmir and east coast shows a significant increasing trend (Fig. 3.4).

Climate change is not just affecting the southwest monsoon but is also driving changes in the northeastern monsoon. The variability of northeast monsoon rainfall has increased in the period 1959–2016. Seasonal rainfall has increased over Tamil Nadu, Rayalaseema, as well as south peninsular India because of an increase in the number of high-intensity rainfall events in the recent period compared to 1901–1958 (Nageswararao et al. 2019). Table 3.1 gives summary statistics for rainfall over India based on 1951-2015.

ENSO and IOD typically exert an offsetting impact on Indian summer monsoon rainfall (ISMR), with an El Niño event tending to lower, whereas a positive IOD tending to increase ISM rainfall (Ashok and Saji 2007). In a recent study, Li et al. (2017) showed that CMIP5 models simulate an unrealistic present-day IOD-ISMR correlation due to an overly strong control by ENSO. As such, a positive IOD is associated with an ISM rainfall reduction in the simulated present-day climate. They further highlight that the uncertainties in ISM rainfall projection can be in part due to the present-day simulation of ENSO, the IOD, their relationship, and their rainfall correlations. Thus, the natural variability also plays a dominant role in the diverse ENSO-monsoon relationship during the twentieth century (Li and Ting 2015). From CMIP5 models, the analysis by Li et al. (2017) further showed that in future, the enhance SST warming could likely lead to a weak ENSO-monsoon relation as well.

Multi-model average changes considered in the different model sources, in general, suggest wetter future conditions. CMIP5 models project an increase of 6% (RCP4.5) and 8% (RCP8.5) in the near future over the central Indian region (core monsoon zone defined by Rajeevan et al. 2008). Projected changes in rainfall by the end of the twenty-first century are 10% (RCP4.5) and 14% (RCP8.5), respectively. The vast majority of the CMIP models shows enhanced monsoon precipitation due to global warming (e.g. Kitoh et al. 1997; Douville et al. 2000; Ueda et al. 2006; Cherchi et al. 2011; Rajendran et al. 2012; Krishnan et al. 2013); however, they indicate a likely weakening of large-scale monsoonal circulation (Krishnan et al. 2016). Many studies noted that the poor skill in simulating monsoon amplifies the ambiguities in understanding the future changes in projected monsoon rainfall (e.g. Chaturvedi et al. 2012; Saha et al. 2014; Sharmila et al. 2015; Krishnan et al. 2016). Studies have highlighted the wide inter-model spread in the simulated precipitation changes over South Asia, which therefore makes the assessment of regional hydroclimatic response a bit ambiguous in reality (e.g. Kripalani et al. 2007; Annamalai et al. 2007; Turner and Slingo 2009; Sabade et al. 2011; Fan et al. 2010; Hasson et al. 2013; Saha et al. 2014).

Applying the Clausius–Clapeyron equation, the water vapour holding capacity of the atmosphere is expected to increase by about 7% per degree of warming. The enhanced availability of moisture can naturally lead to more precipitation over different parts of the globe (Trenberth 1998; Meehl et al. 2005). Indeed, there is a considerable multiscale feedback through large-scale circulation and various thermodynamic and dynamic conditions locally that also contribute to rainfall occurrence. In general, climate simulations hint that the global warming is expected to enhance the Indian summer monsoon precipitation by 5–10%, albeit some climate models suggest less. In the Indian monsoon region, the seasonal movement of the ITCZ brings rainfall over land, and the convection is strongly coupled with large-scale circulation (Goswami 2006; Joseph and Sabin 2008; Gadgil 2018). However, this convective-coupling is relatively weak in most of the CMIP5 models and mostly simulating an increase in precipitation irrespective of the anomalous decrease in monsoon circulation (Krishnan et al. 2016). Sabeerali et al. (2015) postulated that the increase in future ISMR simulated in CMIP5 models is a result of unrealistic local convective precipitation enhancement that is not related to large-scale monsoon dynamics, which is mainly due to the unrealistic representation of stratiform and convective cloud ratio in the coupled model. Using CMIP5 outputs, Sabeerali and Ajaymohan (2018) showed that there is a possibility of a shorter rainy season (defined using tropospheric temperature gradient as outlined by Goswami and Xavier (2005)) by the end of the twenty-first century in the RCP8.5 scenario.

Even though there is ambiguity in changes in projected seasonal mean precipitation, future projections from CMIP5 models show a significant increase in interannual variation during the summer monsoon season (Menon et al. 2013). Studies such as Sarita and Rajeevan (2016) have suggested that the periodicity of El Ninos is likely to shift to a shorter period (2.5–3 year) compared to present day (3–5 year) which will have serious implication on monsoon in interannual timescale. The increase in summer monsoon precipitation over South Asian region projected by CMIP5 models is mostly sustained by an increase in moisture supply due to enhanced warming (Mei et al. 2015). By analysing 14 CMIP models, they further showed that towards the end of the century the precipitable water over Indian landmass may increase by 8–16 mm/day, the evapotranspiration by around 0.6 mm/day while the change in moisture convergence around 2.4 mm/day or more under RCP8.5 scenario. Figure 3.10 shows the change in vertically integrated moisture convergence towards the end of the century under RCP4.5 and RCP8.5.

Interannual variations in the northeast monsoon are much larger compared to the southwest monsoons (Fig. 3.7). Projected changes for the near future and far future from CORDEX, NEX, and CMIP5 show a moderate increase in rainfall over the Indian landmass. Note that significant change is seen only over the region where northeast monsoon is not much significant (Figs. 3.8 and 3.9). Over major areas in Tamil Nadu and coastal Andhra, only modest changes are projected for both RCP 4.5 and RCP 8.5. Parvathi et al. (2017) reported that CMIP5 models project a robust reduction of the wind intensity during the northeast monsoon season especially over the Arabian Sea by the end of the twenty-first century (a reduction of 3.5% for RCP4.5 and 6.5% for RCP8.5, on an average). Despite a decrease in the winter monsoon winds, they noted an increased rainfall (10 ± 2%) in the winter monsoon rain zones in the equatorial Indian Ocean. Over Indian landmass, the precipitation enhancement is minimal in the near future in the RCP4.5 scenario; however, both projections (RCP4.5 and RCP8.5) shows an increase in precipitation [CMIP5 (10–20%), CORDEX (10–25%) and NEX (15–35%)] by the end of the twenty-first century.

Detecting changes in the characteristics of extreme rain events is an important issue in view of their large impacts on human society (Ghosh et al. 2016). It is difficult to attribute a specific extreme event during the monsoon is owing to anthropogenic climate change—like the Uttarakhand surges of 2015, or recent flood in Kerala during 2018 monsoon—yet it is robustly anticipated that a warming atmosphere will result in more severe weather. It has been observed that from 1950 onwards there has been a significant rising trend in the frequency and intensity of extreme heavy rain events over central India, along with a decreasing trend in the moderate rain events (Goswami et al. 2006; Dash et al. 2009; Kulkarni et al. 2017; Krishnan et al. 2016; Roxy et al. 2017) (Fig. 3.11a). Consecutive dry days with minimum spell length of 5 days show significant increase of about 4 days in the period 1951–2015, while consecutive wet days show decrease of about 10 days in this period. Prolonged break spells appear to be more frequent in 1951–2015. Roxy et al. (2016) showed that the widespread changes in extreme rain events are mainly dominated by dynamic response of the atmosphere rather than thermodynamic factor alone. Krishnan et al. (2016) showed that, the enhancement of such deep localized convection, leading to heavy rainfall events, are more likely to happen in an atmosphere with weak vertical shear (Romatschke and Houze 2011). Increased variability of low-level monsoon westerlies (Mishra et al. 2018; Roxy et al. 2017) and warming of north Arabian sea lead to increased moisture supply and thus enhance such events (Roxy et al. 2017). By examining the changes in the distribution of moderate and heavy monsoon precipitation in Historical and GHG, only simulations Krishnan et al. (2016)have shown that along with increase of atmospheric moisture the decrease of easterly vertical shear of the SAM circulation is also pivotal for favouring localized heavy rainfall over the Indian region (Fig. 3.11b). In a recent study, Singh et al. (2014) found statistically significant increase in the intensity and frequency of extreme wet and dry spells during the ISM during the 1951–2011 period.

The IPCC Special Report on Extremes (SREX; Intergovernmental Panel on Climate Change 2012) appraisal reported that extreme precipitation events globally are certain to rise in the future. From 1950 onwards, the number of extreme precipitation events over Indian landmass has also become more significant than it before (Sillmann et al. 2013; Goswami et al. 2006; Rao et al. 2014). A recent study by Mukherjee et al. (2017) showed that 1–5-day precipitation maxima at 5–500 year return period will increase (10–30%) with anthropogenic warming in RCP8.5 scenario. They further showed that the frequency of precipitation extremes is projected to rise more prominently in the RCP8.5 scenario over southern and central India by the middle and end of the twenty-first century. The analyses of select precipitation-based indices, defined by the Expert Team on Climate Change Detection and Indices (ETCCDI), computed with a consistent methodology for climate change simulations for different emission scenarios are discussed below.

Relative changes in the contribution of very wet days to total wet day precipitation (R95PTOT), the daily intensity index (SDII) and maximum 5-day precipitation (RX5day) with respect to 1976–2005 reference period are shown in Fig. 3.12. Similar to most of the global tropics, over Indian landmass as well, extreme precipitations are projected to increase throughout the twenty-first century. In RCP8.5, R95PTOT and SDII are projected to rise by 15 and 21%, by the end of the twenty-first century, whereas RX5day is projected to rise by 38%. The spread among ensemble members (shading in Fig. 3.12) is more in RCP8.5 comparing to RCP4.5 scenario all through the twenty-first century.

The spatial pattern of the projected multi-model ensemble means of the precipitation extremes identifies moderately higher increase in the contribution of very wet days to total wet day precipitation (R95PTOT; Fig. 3.13a), the daily intensity (SDII; Fig. 3.13b), and in the maximum 5-day precipitation (RX5day; Fig. 3.13c) are visible along the west coast, central and northern Indian states. Both the scenarios (RCP4.5 and RCP8.5) showed consistent results in the projected changes for both the near and far future. Even though the number of consecutive dry days (CDD; Fig. 3.13d) is increasing over various parts of India, the experiments provide a consensus only over the Indian peninsular region throughout the twenty-first century in the RCP8.5 scenario. The simultaneous increase in both CDD and RX5day indicates an increase of both dry and wet epochs along the west coast and the peninsular region of India. This analysis is in agreement with the study by Mukherjee et al. (2017), who noted the frequency of extreme precipitation from CMIP5-GCM shows an increasing trend prominent over southern India (Fig. 3.14). This result enhances our confidence in assessing a likelihood of an increase in future precipitation extremes over the Indian peninsula throughout the twenty-first century, under the propensity of global warming signals.