1. Introduction
Indian Summer Monsoon (ISM) has enormous impacts not only on the Indian economy, but also on the global economy. It provides the major share of the total annual rainfall of the country, and being one of the most populated countries in the world, its variations influence the global economy. Moreover, the ISM system also has an important contribution to the atmospheric circulation, by dominating the northern summer Hadley circulation [
1].
The boreal summer (June–July–August–September (JJAS)) climatological precipitation has a widespread maximum over Central Northeast (CNE) India, which includes the Indo-Gangetic Plain. The ISM represents a large-scale heat source around the CNE India, which could be related to both the regional Hadley and Walker cell, following the linear theory [
2]. Hence, additional attention was given to that region. A number of observational works has investigated the trend of monsoon rainfall over India during the climate change period of the last half of the 20th century [
3,
4]. Considering a region (76°–87° E, 20°–28° N) around the CNE India, a study [
5] detected a decreasing trend for ISM precipitation. They analysed the period of 1940–2005 and used the NOAA GFDL CM 3 model for all forcing (natural and anthropogenic) conditions. The Climate Research Unit (CRU) observational data also showed a marked reduction from the 1950s to the end of the 20th century (statistically significant at the 95% confidence level). That observation is compared with the model output in that study [
5]. The results from CRU are also broadly consistent with previous observational studies [
4,
6,
7]. Another study [
8], using a slightly different region around the similar location of central-northern India (74.5° E–86.5° E and 16.5°N–26.5° N), also found a decreasing trend of ISM rainfall. Though the observational results indicated a drying trend, it is interesting to explore using various model outputs whether there is any consensus among model results. The initial focus here is on an analysis region of a location and size similar to that used by the earlier mentioned study [
5]. Later, it also included a region around CNE as considered by [
8]. Those two regions are marked by CI and CII respectively in
Figure 1.
A recent study [
9] discussed how the response of precipitation changes varies regionally with respect to the global warming scenario. For the tropical ocean, two viewpoints exist. One predicts ‘wet-gets-wetter’, which means more rainfall in precipitation-prone regions [
10,
11]. The other suggests ‘warmer-gets-wetter’, meaning increased precipitation would occur with a rise in sea surface temperature. Further analyses [
12,
13,
14], however, suggested that the two mechanisms are complementary and not contradictory. Those mentioned that the variability of the annual mean precipitation over the tropical ocean follows the ‘warmer-gets-wetter’ mechanism, but the seasonal mean precipitation suggests the ‘wet-gets-wetter’ rule. The precipitation in the high to mid-latitudes has enhanced by 0.5–1% per decade, which is consistent. There is an exception though over East Asia [
15], and the reason for such a deviation in East Asia is still poorly understood [
16,
17]. The drying trend around the CNE region of India is consistent with such a deviation.
Studies have identified clear connections between the ISM and the El Niño Southern Oscillation (ENSO) [
18]. Such associations were also recently discussed [
19], which put emphasis on the climate change period of the last half of 20th Century. It also discussed the role of the North Atlantic Oscillation (NAO) and the Indian Ocean Dipole (IOD). A research [
20] analysed ISM-ENSO teleconnection during the latter half of the last century using the reanalysis product. They proposed a route, whereby the ISM could have a remote influence via the modification of Eurasian temperature. In examining ISM-ENSO connections, studies even identified modulating roles of atmosphere-ocean coupling systems originated in the Northern Hemisphere (via (Pacific Decadal Oscillation) PDO [
21] and the North Atlantic [
22]). On the other hand, using wavelet techniques on observational data, [
23] suggested an inter-decadal variation in the monsoon-ENSO behaviour. They showed that it is true, irrespective of the analysis method and different datasets. The ISM is such a complicated system that understanding and predicting its varied behaviour is always a challenge. Noticeable improvements in forecasting are realized through the inclusion of more detailed physics of the climate and higher resolution in models. A major step forward is the incorporation of coupled ocean-atmosphere models, those including air-sea interaction.
However, most of the current generation climate models are still not capable of simulating realistic ENSOs [
24,
25]. The overall skill of ENSO prediction in retrospective forecasts made with ten different coupled GCMs was also investigated [
26]. It analysed seasonal output from the APCC/CliPAS (Asian-Pacific Economic Cooperation Climate Center/Climate Prediction and its Application to Society) and DEMETER (Development of a European Multi-model Ensemble system for seasonal to inTER-annual prediction) projects during the common 22 years from 1980 to 2001. They indicated that the overall prediction skill is in need of improvement. Recent modelling studies [
26,
27] have also confirmed this. Given the crucial role played by the Pacific in the global climate, advancement of our knowledge relating to the ENSO is obviously important.
A recent collective initiative among different modelling communities around the world conducted similar experiments that comprised the Coupled Model Inter-comparison Project, Phase 5 (CMIP5) (link:
http://cmip-pcmdi.llnl.gov/cmip5/experiment_design.html). The details are all described clearly [
28]. Changes in ENSO variability during 2050–2100 are studied using CMIP3 (third phase of Coupled Model Inter-comparison Project) experiments [
25]. In terms of the amplitude of ENSO, they showed that models are still unable to reach consensus for predictions in the future. However, the coupled CMIP5 models are more capable of simulating ENSO-like interannual variability in the central and eastern equatorial Pacific [
29]. Compared to the CMIP3 group of models, more CMIP5 models show a realistic range of ENSO frequencies in the band of 2–7 years in the equatorial Pacific. Nearly half of the CMIP5 models show Sea Surface Temperature (SST) anomalies peaking from November–January, as seen in observations. It is of considerable importance to determine how ENSO responds under rising amounts of greenhouse gases [
30]. Numerous studies also discussed the evolution of ENSO in the Representative Concentration Pathway (RCP) scenario runs [
31,
32], though uniform consensus is yet to be reached.
Using CMIP3 experiments, a study [
33] examined ISM precipitation averaged over land regions (60–90° E, 7–27° N). They also elaborately discussed various issues as to why models fail to match observations and indicated that advanced understanding and model improvements are essential for improving the skill of ISM prediction. In terms of mechanisms relating to disagreement among model results and observations, various studies addressed this from different angles (those include sea-surface temperatures in the Indo-Pacific [
34]; circulation-based changes [
34]; and aerosol-based changes [
5]). Using the model, it was noted [
35] that for observed fluctuations in the ENSO-ISM correlation, sampling variability can also be a responsible factor.
A recent work [
36] discussed ISM precipitation in detail and compared CMIP3 with CMIP5 models. It elaborately discussed its linkage with ENSO. The performance of CMIP5 is shown to be improved. Relating to the ENSO-ISM teleconnection, recent research [
37,
38] also considered various CMIP5 model outputs. In those studies, they elaborately discussed the teleconnection focusing on two different types of ENSO: Eastern Pacific (EP)-type ENSO (and often known as Canonical ENSO) and Central Pacific (CP)-type ENSO (often known as Modoki ENSO). This is because various studies suggested that there are differences in global and local influences between ENSO Modoki and Canonical ENSO ([
39], among others). Focusing on ISM, it was indicated [
38] that more than 80% of the CMIP5 models capture ENSO-ISM regional teleconnections around the CNE region, irrespective of EP or CP ENSO category.
A few of the principal aims of the CMIP5 project are assessing the mechanisms responsible for model differences and determining why similarly forced models are producing a range of responses, among others [
28]. The proper phasing and teleconnections of ISM-ENSO on various time scales could also be an important aspect to explore. The current study tries to explore those areas relating to ISM precipitation and Niño temperature, using CMIP5 model outputs.
The structure of this study is as follows.
Section 2 discusses the methodology and data. In the Results section, first, the temporal and spatial pattern of ISM around regions in CNE using CMIP5 models (
Section 3.1) is the focus. This is followed by some characteristics of Niño 3.4 in models (
Section 3.2). Later, ISM and Niño3.4 correlation (
Section 3.3), while
Section 3.4 compares ISM with global precipitation in future scenarios. Conclusions are presented in
Section 4. Overall, this study identifies a few areas where the CMIP5 models show consistencies/disagreements with observations. It also presents an overview of future scenarios.
4. Conclusions
Some general features of ISM precipitation were explored. It was found that the trend pattern around the CNE region of India varies from model to model. The trend of decreasing rainfall around CNE India, as detected in the observations, though noticed in some models, is not true for every case. Various ENSO features also differ in the models; those include variability, trend, phasing, etc. This is also true for historical, as well as RCP scenarios. Interestingly, unlike other models, the model FGOALS-g2 does not show any trend in Niño 3.4 temperature for either the historical or RCP 8.5 scenario. This observation could be useful for model evaluation purposes. ISM and ENSO correlation was also studied in the historical and RCP 8.5 scenarios, for all of India, as well as two specific regions of India (CI and CII). This suggests a negative correlation for almost all models. Precipitation during JJAS in land regions of the globe shows a clear rising trend in all CMIP5 model outputs for the RCP 8.5 scenarios, but the same for the Indian subcontinent fails to indicate anything clearly. The various future scenarios suggest a much larger uncertainty for ISM rainfall (CI region) in comparison to global precipitation, and the model ensemble does not indicate any decrease in precipitation for India.
In this study, we not only identified some areas where the CMIP5 models show disagreements, but also discussed some aspects where most models agree. Such analyses could be beneficial for improving models and gaining a better understanding of the process representation in models. Additional observations relating to FGOALS-g2 could be used for model evaluation purposes. Finally, this study also provides an indication for the longer term trend of future precipitation.