Unraveling the global teleconnections of Indian summer monsoon clouds: expedition from CMIP5 to CMIP6

Highlights

• The total cloud fraction (TCF) over the ISM region is strongly associated with global predictors.

• The seasonal mean biases of TCF and rainfall have seen improvement from CMIP5-MME to CMIP6-MME.

• Improvement from CMIP5 to CMIP6 models may be attributed to the better depiction of the observed global teleconnections.

Abstract

We have analyzed the teleconnection of total cloud fraction (TCF) with global sea surface temperature (SST) in multi-model ensembles (MME) of the fifth and sixth Coupled Model Intercomparison Projects (CMIP5 and CMIP6). CMIP6-MME has a more robust and realistic teleconnection (TCF and global SST) pattern over the extra-tropics (R ~ 0.43) and North Atlantic (R ~ 0.39) region, which in turn resulted in an improvement of rainfall bias over the Asian summer monsoon (ASM) region. CMIP6-MME can better reproduce mean TCF and have reduced dry (wet) rainfall bias on land (ocean) over the ASM region. Model bias with respect to seasonal mean rainfall, TCF, and outgoing longwave radiation (OLR) in CMIP6-MME are improved over the Indian Summer Monsoon (ISM) region by ~40%, ~45%, and ~ 31%, respectively, as compared to CMIP5-MME. Further, CMIP6-MME demonstrates a better spatial correlation with observation/reanalysis. The present study has also shown the lag correlations in the teleconnection analysis, i.e., the correlation of June–September (JJAS) mean of rainfall/TCF with October–December (OND) SST from observation/reanalysis, CMIP5-MME, and CMIP6-MME. The CMIP6-MME performs better than CMIP5-MME as compared to observation/reanalysis. The results establish the credibility of the CMIP6 models and provide a scientific basis for improving the seasonal prediction of ISM.

Introduction

Over the past few decades, progress in climate modeling has provided new insight regarding ASM. The inter-annual and intraseasonal variability of ASM in general and Indian summer monsoon (ISM) rainfall (ISMR) in particular controls the livelihood and health of over two billion people and regulates the country's economy through food production (Gadgil, 2003; Gadgil and Rupa Kumar, 2006; Gupta et al., 2019; Krishna Kumar et al., 2004). Hence, predicting this variability well in advance can help farmers in crop management and the government to be prepared for natural calamities (e.g., flood and drought).

The continuous efforts to develop model resolution and physics have shown advancement in ASM rainfall simulation in the CMIP6 (Eyring et al., 2016) models relative to the CMIP5 (Taylor et al., 2012) and CMIP Phase 3 (CMIP3, Meehl et al., 2007) models. Sperber et al. (2013) and Seo et al. (2013) have compared CMIP5 and CMIP3 model ensembles and pointed out that CMIP5 displayed better capabilities over ASM in many aspects of monsoon diagnostics. Gusain et al. (2020), in a preliminary study, compared a few CMIP5 and CMIP6 models regarding their spatiotemporal variation over the ISM region. Recent studies have compared CMIP5 and CMIP6 in simulating mean rainfall over the East Asian Summer Monsoon (Xin et al., 2020) and east Africa (Ayugi et al., 2021) region. Zhu and Yang (2021) also assessed the CMIP5 and CMIP6 in terms of simulated inter-decadal and inter-annual variation of global precipitation. Vignesh et al. (2020) filled the gap of similar assessment studies in global cloud distribution. They have compared CMIP5 and CMIP6 models based on the seasonal and regional variations of cloud fractions. Other researchers (Chen et al., 2020; Tokarska et al., 2020; Zelinka et al., 2020; Nie et al., 2020) have conducted similar assessment studies for different variables (e.g., temperature, circulation, etc.). These studies generally highlight the improvement of the mean structure of CMIP6 participating models as compared to CMIP5 models with certain exceptions in various regions across the globe.

However, despite the improvement in the simulation of the mean structure of ISMR, the skill of its prediction is still below the potential limit for the models (Kumar et al., 1999; Rajeevan et al., 2012). Tiwari et al. (2014) found that these low skill scores may originate from the sparse representation of observed teleconnection of ISMR with Pacific SST, i.e., El-Nino and Southern Oscillation (ENSO), revealed in numerous studies (Charney and Shukla, 1981; Krishnamurthy and Goswami, 2000; Ropelewski and Halpert, 1989; Sikka, 1980; Mooley and Parthasarathy, 1984; Pradhan et al., 2016; Goswami and Xavier, 2005; Goswami and Jayavelu, 2001; Pokhrel et al., 2012; Dwivedi et al., 2015; Saha et al., 2019, Saha et al., 2020). Zhou et al. (2009) have found that the south Asian monsoon is better reproduced in 20th century models, and it is primarily driven by tropical Pacific forcing, i.e., ENSO. Roy et al. (2019) and Mahendra et al. (2021) have tested the fidelity of this ENSO-ISMR relationship from CMIP5 and CMIP6 models, respectively but with a different approach. Kucharski et al. (2009) have also shown that the decadal variability of ISMR is not only governed by tropical SSTs, e.g., ENSO, but also by extratropical SSTs like Atlantic Multidecadal Oscillations (AMO). However, the climatic relationship or teleconnection of ISMR with these potential sources of predictability, such as North Atlantic Oscillations (NAO), Atlantic Multidecadal Oscillations (AMO), and Extra Tropics (ET), seems to gain less attention comparatively (Burns et al., 2003; Srivastava et al., 2002; Chang et al., 2001; Chattopadhyay et al., 2015; Sankar et al., 2016; Borah et al., 2020). Recent studies (Huang et al., 2020a, Huang et al., 2020b) have identified inter-decadal Pacific Oscillation (IPO) as an important tropical forcing factor in modulating the long-term ISM rainfall changes. They reveal that IPO influences ISMR (Huang et al., 2020a) and is also responsible for the uncertainties in ISMR simulation in models (Huang et al., 2020b). Scaife et al. (2009) have diagnosed a set of atmospheric climate models in the “CLIVAR C20C project” and found that all the models fail to simulate the observed increase in the NAO. Several studies have discussed the structure and variability of these modes (viz. NAO, AMO, ENSO, etc.) from CMIP3 or CMIP5 models (Medhaug and Furevik, 2011, Ting et al., 2011; Zhang and Wang, 2013; Wang et al., 2017, etc.) and Fasullo (2020) have assessed their performances with CMIP6 models. However, the teleconnection studies of ISMR with these sources (e.g., AMO) from CMIP models are found in limited studies (Luo et al., 2018; Joshi and Ha, 2019).

On the other hand, the different types of teleconnection studies regarding ISM clouds are not focused on detail from both observational and modeling aspects. This is despite the fact that clouds play a seminal role in governing rainfall variability (Chaudhari et al., 2016) through the modulation of heating (Hazra et al., 2017a, Hazra et al., 2017b; Hong et al., 2016; Baker, 1997) and induced circulation (Kumar et al., 2014). The individual cloud complexes are also revealed by satellite-derived cloud data (Wang and Rui, 1990). Zhang et al. (2020) used multisource satellite data sets to show the different cloud characteristics (e.g., the seasonal cycle of cloud fraction, cloud top height and cloud radiation forcing) in different monsoon regions across the globe. They found that high-level clouds occur the most in all monsoon regions. Chaudhari et al. (2016) have also found that high-level clouds dominate over the ISM region during the summer monsoon. However, low-level clouds are also found over the Asian monsoon region in other seasons (Zhang et al., 2020; Luo et al., 2009). Nakazawa (1988) showed that the intraseasonal variability (ISV) is related to the large-scale cloud complexes. Therefore, earlier studies (Stephens et al., 2002; Hazra et al., 2015, Hazra et al., 2016, Hazra et al., 2017a, Hazra et al., 2017b, Hazra et al., 2020; Waliser et al., 2009) highlighted that clouds regulate radiative energy and maintain water cycle balances. Bony et al. (2015) reported that clouds are essential for climate sensitivity studies in climate models. However, the misrepresentation of clouds, precipitation, and circulation is still persistent in many new-generation models (Morrison et al., 2020). De et al. (2019) emphasized that the interaction between clouds and large-scale circulation remains a ‘grey area of climate science’. Guo et al., 2015, Guo et al., 2021 have demonstrated the seminal role of cloud radiative heating in driving the monsoon circulation. Cloud microphysical processes also play a significant role in modulating the ISV of ISM (Bony et al., 2015; Kumar et al., 2017; Hazra et al., 2017a, Hazra et al., 2020; Dutta et al., 2020, Dutta et al., 2021). Therefore, the role of clouds in general and cloud condensates like cloud ice, in particular, are essential for the simulation of sub-seasonal disturbances by AOGCMs (Atmosphere-Ocean General Circulation Models) with higher fidelity (Dutta et al., 2020; Hazra et al., 2017a, Hazra et al., 2017b, Hazra et al., 2020). This is also a critical requirement for the seasonal prediction of south Asian monsoon rainfall in particular and tropical rainfall in general.

Therefore, it is essential to explore whether the inter-annual variability of ISM clouds is also teleconnected with the slowly varying predictable component like SST around the globe. Assessment of the simulated relationship from the recent two generations of CMIP will also provide new insight for the scientific community in model development. Recently, Kim et al. (2020) evaluated the performance of CMIP5 and CMIP6 models based on the observed teleconnection of heatwaves over Korea. This kind of assessment study from CMIP5 and CMIP6 models has been overlooked for the teleconnection of ISM clouds (and rainfall) with global predictors along with its mean structure. For this important reason, the aim of this study is to evaluate the performance of CMIP6 ensembles compared to CMIP5 in simulating the ASM in general and ISM clouds in particular. In this context, we discuss the following points:

i) Assessment of the mean structure of ASM and ISM with respect to rainfall, clouds, and convection from ensembles of CMIP5 and CMIP6, considering more models (30 for each) than previous studies (Gusain et al., 2020; Vignesh et al., 2020).

ii) Exploring the vital role of cloud fractions for ASM (ISM in particular) through teleconnection with global predictors (ENSO, NAO, AMO, and ET) and comparative evaluation of this teleconnection from CMIP5 and CMIP6 models and their respective MME.

iii) Similar assessment of simulated teleconnection of ISM rainfall from CMIP5 and CMIP6.