Search

Cumulus parameterization (CP) in state‐of‐the‐art global climate models is based on the quasi‐equilibrium assumption (QEA), which views convection as the action of an ensemble of cumulus clouds, in a state of equilibrium with respect to a slowly varying atmospheric state. This view is not compatible with the organization and dynamical interactions across multiple scales of cloud systems in the tropics and progress in this research area was slow over decades despite the widely recognized major shortcomings. Novel ideas on how to represent key physical processes of moist convection‐large‐scale interaction to overcome the QEA have surged recently. The stochastic multicloud model (SMCM) CP in particular mimics the dynamical interactions of multiple cloud types that characterize organized tropical convection. Here, the SMCM is used to modify the Zhang‐McFarlane (ZM) CP by changing the way in which the bulk mass flux and bulk entrainment and detrainment rates are calculated. This is done by introducing a stochastic ensemble of plumes characterized by randomly varying detrainment level distributions based on the cloud area fraction of the SMCM. The SMCM is here extended to include shallow cumulus clouds resulting in a unified shallow‐deep CP. The new stochastic multicloud plume CP is validated against the control ZM scheme in the context of the single column Community Climate Model of the National Center for Atmospheric Research using data from both tropical ocean and midlatitude land convection. Some key features of the SMCM CP such as it capability to represent the tri‐modal nature of organized convection are emphasized. Plain Language Summary: Current climate models perform poorly in the way they represent convection and cloud systems in the tropics. A stochastic multicloud model (SMCM), based on an interacting particles lattice model that tracks the random variations of the cloud area fractions (CAF) of the main cloud types that characterize organized tropical convection, is used here to modify/stochasticize a state‐of‐the‐art cumulus parametrization. The main idea is to use the CAF of various cloud types, as predicted by the SMCM to decide on the distribution of the vertical penetration of convecting air parcels that form the clouds. Accordingly the distribution of the associated cloud top heights changes randomly based on the large scale environment depending on whether the later favors shallow, mid‐level, or deep convection. The new parametrization is tested in the NCAR Single Column Community Climate Model and compared with observations when available. The main features of the new parameterization are discussed here. Based on the results, the stochastic parameterization possess various features that are desirable in order to improve the representation of organized convection in 3D climate models. This is in agreement with the success of the SMCM when used in different climate models, although based on much simplistic approaches. Key Points: The Zhang‐McFarlane cumulus parameterization is modified by using randomly detraining plumes according to a stochastic multicloud model (SMCM)The SMCM tracks the evolution throughout the convection life cycle of the area fractions associated with multiple cloud types that characterize organized convectionThe SMCM framework allows for a natural coupling of shallow and deep convection and reproduces the tri‐modal convection structure [ABSTRACT FROM AUTHOR]

In contrast to the "wet gets wetter and dry gets drier" paradigm, here, using observations and climate model simulations, we show that the mean rainfall over the semi‐arid northwest parts of India and Pakistan has increased by 10%–50% during 1901–2015 and is expected to increase by 50%–200% under moderate greenhouse gas (GHG) scenarios. The GHG forcing primarily drives the westward expansion of the Indian summer monsoon rainfall (ISMR) and is facilitated by a westward expansion of the Indian Ocean (IO) warm pool. Mechanistically, the westward expansion of ISMR is a consequence of the episodic genesis over IO and the northward propagation of an expanded Inter‐Tropical Convergence Zone on a sub‐seasonal time scale. While an adaptation strategy to increased hydrological disasters is a must, harnessing the augmented rainfall would lead to a substantial boost in food productivity, bringing transformative changes in the socio‐economic condition of people in the region. Plain Language Summary: An apparent eastward shift of the Indian monsoon has led to the arid conditions in the west and north‐west regions of India where monsoon was once active, and the Indus Valley civilizations thrived (5,300–3,300‐year BP). A reversal of the process and a westward expansion of the present‐day Indian monsoon would transform the west and north‐west India to a humid "monsoonal" climate and provide food security to the expanding population of the country. The present study shows that the Indian monsoon is indeed expanding to the west with 10% decrease of mean rainfall in the northeast and 25% increase in the west and north‐west during the historical period with a potential of 50%–100% increase in the north‐west under SSP2–4.5. Harvesting the increased rainfall has the potential for significant increase in food productivity bringing in transformative changes in the socio‐economic condition of people of the region. Key Points: Global warming could potentially cause a westward expansion of the Indian monsoonRainfall in northwest India and Pakistan has increased and will continue to rise due to greenhouse gasesIncreased monsoon rainfall can improve food productivity and socio‐economic conditions in the region [ABSTRACT FROM AUTHOR]

In the backdrop of overwhelming evidences of associations between North-Atlantic (NA) sea-surface temperature (SST) and the Indian summer Monsoon Rainfall (ISMR), the lack of a quantitative nonlinear causal inference has been a roadblock for advancing ISMR predictability. Here, we advance a hypothesis of teleconnection between the NA-SST and ISMR, and establish the causality between the two using two different nonlinear causal inference techniques. We unravel that the NA-SST and the El Nino and Southern Oscillation (ENSO) are two independent drivers of ISMR with the former contributing as much to ISMR variability as does the latter. Observations and climate model simulations support the NA-SST–ISMR causality through a Rossby wave-train driven by NA-SST that modulates the seasonal mean by forcing long active (break) spells of ISMR. [ABSTRACT FROM AUTHOR]

Systematic biases in simulation of the mean and variability of Indian summer monsoon rainfall (ISMR) and its teleconnection with the El Nino and Southern Oscillation (ENSO) by climate models have proven to be roadblocks for improving the seasonal prediction skill of the ISMR. Yet, the improvements of these biases in simulating the ISMR by coupled ocean–atmosphere general circulation models (CGCMs) have been slow. Here, we investigate the progress made by the latest CMIP6 models in simulating the mean and variability of the ISMR including its multi-decadal variability in comparison to the earlier CMIP3 and CMIP5 models and unravel key common and persisting biases. That the number of models simulating the pattern of climatological mean summer rainfall over the region with fidelity has increased from less than 10% in CMIP3 to more than 50% in CMIP6 is a notable progress. The CMIP6 models also indicate notable progress in simulating the inter-annual, multi-decadal variability and marginal improvement in simulating the ENSO-ISMR correlations. In addition to consolidating these advances in simulating the ISMR in more climate models, future developments must focus on improving the following three persisting biases namely; (1) most models, including those in CMIP6, tend to simulate the mean location of the continental ITCZ during June–September (JJAS) about 10° south of the observed location of ~ 22° N, (2) with the ensemble mean 21-year moving correlation between ISMR and JJAS mean Nino3.4 sea surface temperature for all models being ~ − 0.3 compared to observed ~ − 0.65, almost all models simulate significantly weaker ENSO-monsoon relationship and (3) the number of models simulating the observed variance and periodicity of the multi-decadal variability of ISMR remains small even in CMIP6. We argue that the position bias in simulating the continental ITCZ may be related to the large and significant biases in simulating the JJAS SST over the tropical Indian Ocean, persistent even in the best of CMIP6 CGCMs making the improvement of SST biases (and associated heat transport across the equator) another area where future model developments need to focus. [ABSTRACT FROM AUTHOR]

In the backdrop of a significant improvement in weather prediction with Numerical Weather Prediction models, quantitative prediction of the intensity of heavy rainfall events and associated disasters has remained a challenge. Encouraged by the recent emergence of compelling observational evidence for a significant electrical influence on cloud/rain microphysical processes (Mudiar, Pawar, Gopalakrishnana et al., 2021, https://doi.org/10.1029/2021gl093577), here we propose a hypothesis that the modification of the raindrop size distribution (RDSD) towards larger drop sizes facilitated by cloud electric fields could be one factor responsible for realistic rainfall intensity in weather/climate models. The robustness of the proposed hypothesis is confirmed through a series of simulations of strongly electrified rain events with the Weather Research and Forecasting model incorporating the electrically modified RDSD parameters in the model physics. Our results indicate a possible roadmap for improving hazard prediction associated with extreme rainfall events in weather prediction models. Plain Language Summary: Lightning and heavy precipitation events in the tropics have significant socio‐economic implications in regard to disaster management. The accurate prediction of these meteorological events has remained a challenge despite considerable improvement in the Numerical Weather Prediction model over the years. With convincing observational evidence of a significant impact of the cloud electric field on the raindrop size distribution (RDSD) in the background, here we have attempted to test the model fidelity to simulate rain events with a strong in‐cloud electrical environment using the Weather Research and Forecasting model. A significant underestimation of the observed rain intensity is noted for the rain events that are associated with lightning. A possible roadmap has been discussed to improve the observed dry bias of rainfall by incorporating the electrically modified raindrop RDSD parameters in the model physics. Substantial improvement in the rain intensity has been observed with the incorporation of electrically modified RDSD parameters. Key Points: Cloud electric field substantially modifies the raindrop size distribution in strongly electrified cloudsWeather Research and Forecasting model simulations show a significant underestimation of the observed rain intensity for strongly electrified rain eventsThe simulated rain intensity improved significantly with the incorporation of electrically modified raindrop size distribution parameters [ABSTRACT FROM AUTHOR]

With the recognition that the circulation is strongly coupled with the Indian summer monsoon rainfall (ISMR), a number of large-scale circulation indices have been proposed to represent the Indian summer monsoon to aid diagnosis of teleconnections with ISMR and evaluation of simulations by climate models. The circulation indices, however, correlate poorly with ISMR raising doubts on the coupling between the two. In this study, we identify reasons behind the poor correlations and estimate a potential limit on this correlation between circulation indices and ISMR. One mechanistic reason for poor correlations between circulation indices and ISMR is that the period and variance explained by the oscillatory modes of circulation indices differ significantly from those of ISMR leading to their phases also differing between modes of circulation indices and corresponding modes of ISMR. Further, the wind anomaly patterns associated with wind indices project strongly on the climatological monsoon circulation unlike that associated with the ISMR, another reason for their poor correlation with ISMR. As a result, all circulation indices underestimate the negative correlation between June–September sea-surface temperature (JJAS SST) over the Niño3.4 and ISMR and fail to simulate the lead–lag relationship between the two. A conservative estimate of the interannual variance explained by coupling of the JJAS rainfall over continental India andwinds in the neighbourhood is strong, explaining about 55% variance distributed in three leading coupled modes. Therefore, the effectiveness of any circulation indices in explaining the ISMR variance is potentially limited to that associated with the dominant mode (∼30%). The local response of winds to deep convection over the waters of the Indian Ocean and western Pacific warm pool leads to a strong increasing trend in some of the circulation indices and adds “noise” modes contributing to the de-correlation between circulation indices and ISMR. [ABSTRACT FROM AUTHOR]

Reliable projections of Indian Summer Monsoon Rainfall (ISMR) to the next century by climate models is critical for policy toward sustainable development goals but depends on sensitivity of the models simulating ISMR as global mean temperature changes. Here, using observed ISMR and historical global temperature (1850–2005), we find that the mean ISMR scales as (3.95 ± 1.13%)/K, lower than the Clausius‐Clapeyron (CC) scaling (∼7%/K). However, the sensitivity of the intensity of daily rainfall extremes is (29.5 ± 3.4)%/K, four times as large as the CC scaling. We also find that the sensitivities of simulated mean ISMR by 36 CMIP5 and CMIP6 models each at 13.36%/K and 11.82%/K, respectively, are 2–3 times larger than the observed sensitivity and independent of the chosen period. Larger than simulated CC scaling of ISMR is consistent with simulated strengthening of Indian monsoon circulation in contrast to observations. Models' sensitivity appears to be due to their stronger than observed response to forcing over land resulting in stronger north‐south heating gradient and acceleration of monsoon circulation. Projected sensitivity of ISMR in models with increased greenhouse gas forcing weakens compared to the historical period approaching the CC scaling. Required to delineate the contributions of larger than observed intrinsic sensitivity of the models from projected sensitivity, our findings provide a way to correct the bias potentially improving the reliability of ensemble mean projections of ISMR. Plain Language Summary: Reliability of projections of ISMR to the future by climate models remaining poor due to large intermodel variability making decisions on adaptation and mitigation of climate change difficult. Further, the changes in ISMR in projections could be biased by natural multidecadal variability. Here, making unbiased estimates of changes of ISMR and global mean temperature, we show that the large intermodel variability of projections of ISMR is partly due to the fact that the models are far too sensitive in simulating ISMR changes to changes in global mean temperature compared to observations. As the model biases are systematic leading to the large model sensitivity, our findings also provide a way to correct them and constrain the intermodel variability and improve reliability of the ensemble mean. Key Point: Hydrological sensitivity of Indian Summer Monsoon Rainfall [ABSTRACT FROM AUTHOR]

Large socioeconomic impact of the Indian summer monsoon (ISM) extremes motivated numerous attempts at its long range prediction over the past century. However, a rather low potential predictability (PP) of the seasonal ISM, contributed significantly by "internal," interannual variability was considered insurmountable. Here we show that the internal variability contributed by the ISM subseasonal (synoptic + intraseasonal) fluctuations, so far considered chaotic, is partly predictable as found to be tied to slowly varying forcing (e.g., El Niño and Southern Oscillation). This provides a scientific basis for predictability of the ISM rainfall beyond the conventional estimates of PP. We establish a much higher actual limit of PP (r∼0.82) through an extensive reforecast experiment (1,920 years of simulation) by improving two major physics in a global coupled climate model, which raises a hope for a very reliable dynamical seasonal ISM forecasting in the near future. Key Points: The observed link between synoptic variability and predictable modes suggest a high predictability of the ISMRCFSv2 with improved physics shows ISMR prediction skill higher than current estimate of potential predictabilityModel shows ∼70% of interannual variability of ISMR is predictable, which is much higher than earlier estimates (∼45%) [ABSTRACT FROM AUTHOR]

An outstanding problem of climate models is the persistent dry bias in simulating precipitation over the south Asian summer monsoon region. Guided by observations, it is hypothesized that the dry-bias in simulating precipitation by the models is related to underestimation of high pass variance by most models. An analysis of the simulated mean and variance in precipitation by 36 coupled models show that the dry bias in simulating the mean precipitation by the models is indeed proportional to the underestimation of the variance. Models also indicate that the underestimation of the high-pass variance arise due to the underestimation of the intense rainfall events by models. Further, it is found that the higher resolution models simulate increasingly reduced dry bias by simulating high-frequency variance better through better simulation probability of intense rainfall events. The robustness of our findings over different regions and during both boreal summer and winter seasons indicates the universality of the hypothesis. [ABSTRACT FROM AUTHOR]

Interactions between midlatitude disturbances and the monsoonal circulation are significant for the Indian summer monsoon (ISM) rainfall. This paper presents examples of monsoon-midlatitude linkage through anticyclonic Rossby wave breaking (RWB) over West Asia during June, July and August of the years 1998-2010. RWB events over West Asia are identified by the inversion of the potential vorticity air mass at three different isentropic levels (340, 350, and 360 K) using daily NCEP-NCAR reanalysis. It is observed that RWB took place over West Asia before/during breaks in the ISM. Further, these events occur on the anticyclonic shear side of the subtropical jet, where the gradient of the zonal wind is found to be high. RWB is responsible for the southward movement of high potential vorticity air from the westerly jet, leading to the formation of a blocking high over the Arabian region. In turn, this blocking high advects and causes the descent of upper tropospheric cold and dry air towards Central India. Such an air mass with low moist static energy inhibits deep monsoonal convection and thereby leads to a dry spell. In fact, we find that RWB induced blocking over West Asia to be one of the major causes of dry spell/break episodes in ISM. Additionally, the presence of cold air over Central India reduces the north-south thermal contrast over the monsoon region thereby modifying the local Hadley circulation over the region. [ABSTRACT FROM AUTHOR]

Future projections of the Indian summer monsoon rainfall (ISMR) and its large-scale thermodynamic driver are studied by using CMIP5 model outputs. While all models project an increasing precipitation in the future warming scenario, most of them project a weakening large-scale thermodynamic driver arising from a weakening of the upper tropospheric temperature (UTT) gradient over south Asian summer monsoon region. The weakening of the UTT gradient under global warming scenarios is related to the increase in sea surface temperature (SST) over the equatorial Indian Ocean (EIO) leading to a stronger increase of UTT over the EIO region relative to the northern Indian region, a hypothesis supported by a series of Atmospheric General Circulation Model (AGCM) experiments forced by projected SSTs. To diagnose the inconsistency between the model projections of precipitation and the large-scale thermodynamic driver, we have examined the rate of total precipitation explained by convective and stratiform precipitations in observations and in CMIP5 models. It is found that most models produce too much (little) convective (stratiform) precipitation compared to observations. In addition, we also find stronger precipitable water-precipitation relationship in most CMIP5 models as compared to observations. Hence, the atmospheric moisture content produced by the model immediately gets converted to precipitation even though the large-scale thermodynamics in models weaken. Therefore, under global warming scenarios, due to increased temperature and resultant increased atmospheric moisture supply, these models tend to produce unrealistic local convective precipitation often not in tune with other large-scale variables. Our results questions the reliability of the ISMR projections in CMIP5 models and highlight the need to improve the convective parameterization schemes in coupled models for the reliable projections of the ISMR. [ABSTRACT FROM AUTHOR]

This study investigates the influence of Simplified Arakawa Schubert (SAS) and Relax Arakawa Schubert (RAS) cumulus parameterization schemes on coupled Climate Forecast System version.1 (CFS-1, T62L64) retrospective forecasts over Indian monsoon region from an extended range forecast perspective. The forecast data sets comprise 45 days of model integrations based on 31 different initial conditions at pentad intervals starting from 1 May to 28 September for the years 2001 to 2007. It is found that mean climatological features of Indian summer monsoon months (JJAS) are reasonably simulated by both the versions (i.e. SAS and RAS) of the model; however strong cross equatorial flow and excess stratiform rainfall are noted in RAS compared to SAS. Both the versions of the model overestimated apparent heat source and moisture sink compared to NCEP/NCAR reanalysis. The prognosis evaluation of daily forecast climatology reveals robust systematic warming (moistening) in RAS and cooling (drying) biases in SAS particularly at the middle and upper troposphere of the model respectively. Using error energy/variance and root mean square error methodology it is also established that major contribution to the model total error is coming from the systematic component of the model error. It is also found that the forecast error growth of temperature in RAS is less than that of SAS; however, the scenario is reversed for moisture errors, although the difference of moisture errors between these two forecasts is not very large compared to that of temperature errors. Broadly, it is found that both the versions of the model are underestimating (overestimating) the rainfall area and amount over the Indian land region (and neighborhood oceanic region). The rainfall forecast results at pentad interval exhibited that, SAS and RAS have good prediction skills over the Indian monsoon core zone and Arabian Sea. There is less excess rainfall particularly over oceanic region in RAS up to 30 days of forecast duration compared to SAS. It is also evident that systematic errors in the coverage area of excess rainfall over the eastern foothills of the Himalayas remains unchanged irrespective of cumulus parameterization and initial conditions. It is revealed that due to stronger moisture transport in RAS there is a robust amplification of moist static energy facilitating intense convective instability within the model and boosting the moisture supply from surface to the upper levels through convergence. Concurrently, moisture detrainment from cloud to environment at multiple levels from the spectrum of clouds in the RAS, leads to a large accumulation of moisture in the middle and upper troposphere of the model. This abundant moisture leads to large scale condensational heating through a simple cloud microphysics scheme. This intense upper level heating contributes to the warm bias and considerably increases in stratiform rainfall in RAS compared to SAS. In a nutshell, concerted and sustained support of moisture supply from the bottom as well as from the top in RAS is the crucial factor for having a warm temperature bias in RAS. [ABSTRACT FROM AUTHOR]

Reliable age models are fundamental for any palaeoclimate reconstruction. Available interpolation procedures between age control points are often inadequately reported, and very few translate age uncertainties to proxy uncertainties. Most available modeling algorithms do not allow incorporation of layer counted intervals to improve the confidence limits of the age model in question. We present a framework that allows detection and interactive handling of age reversals and hiatuses, depth-age modeling, and proxy-record reconstruction. Monte Carlo simulation and a translation procedure are used to assign a precise time scale to climate proxies and to translate dating uncertainties to uncertainties in the proxy values. The presented framework allows integration of incremental relative dating information to improve the final age model. The free software package COPRA1.0 facilitates easy interactive usage. [ABSTRACT FROM AUTHOR]

Reasonably realistic climatology of atmospheric and oceanic parameters over the Asian monsoon region is a pre-requisite for models used for monsoon studies. The biases in representing these features lead to problems in representing the strength and variability of Indian summer monsoon (ISM). This study attempts to unravel the ability of a state-of-the-art coupled model, SINTEX-F2, in simulating these characteristics of ISM. The coupled model reproduces the precipitation and circulation climatology reasonably well. However, the mean ISM is weaker than observed, as evident from various monsoon indices. A wavenumber-frequency spectrum analysis reveals that the model intraseasonal oscillations are also weaker-than-observed. One possible reason for the weaker-than-observed ISM arises from the warm bias, over the tropical oceans, especially over the equatorial western Indian Ocean, inherent in the model. This warm bias is not only confined to the surface layers, but also extends through most of the troposphere. As a result of this warm bias, the coupled model has too weak meridional tropospheric temperature gradient to drive a realistic monsoon circulation. This in turn leads to a weakening of the moisture gradient as well as the vertical shear of easterlies required for sustained northward propagation of rain band, resulting in weak monsoon circulation. It is also noted that the recently documented interaction between the interannual and intraseasonal variabilities of ISM through very long breaks (VLBs) is poor in the model. This seems to be related to the inability of the model in simulating the eastward propagating Madden-Julian oscillation during VLBs. [ABSTRACT FROM AUTHOR]

Even though multi-model prediction systems may have better skill in predicting the interannual variability (IAV) of Indian summer monsoon (ISM), the overall performance of the system is limited by the skill of individual models (single model ensembles). The DEMETER project aimed at seasonal-to-interannual prediction is not an exception to this case. The reasons for the poor skill of the DEMETER individual models in predicting the IAV of monsoon is examined in the context of the influence of external and internal components and the interaction between intraseasonal variability (ISV) and IAV. Recently it has been shown that the ISV influences the IAV through very long breaks (VLBs; breaks with duration of more than 10 days) by generating droughts. Further, all VLBs are associated with an eastward propagating Madden–Julian Oscillation (MJO) in the equatorial region, facilitated by air–sea interaction on intraseasonal timescales. This VLB-drought–MJO relationship is analyzed here in detail in the DEMETER models. Analyses indicate that the VLB-drought relationship is poorly captured by almost all the models. VLBs in observations are generated through air–sea interaction on intraseasonal time scale and the models’ inability to simulate VLB-drought relationship is shown to be linked to the models’ inability to represent the air–sea interaction on intraseasonal time scale. Identification of this particular deficiency of the models provides a direction for improvement of the model for monsoon prediction. [ABSTRACT FROM AUTHOR]

Factors responsible for limited predictability of the Asian summer monsoon (ASM) are investigated within a conceptual framework for predictability. Predictability of the seasonal mean depends on the interannual variability (IAV) of the monsoon annual cycle (MAC) and is determined by relative contribution of the predictable “external” component of IAV compared to the unpredictable “internal” IAV. Contributions of slow processes such as those involving air–sea interactions associated with the El Niño–Southern Oscillation (ENSO) or local warm ocean–atmosphere interactions in generating IAV of the MAC are reviewed. Empirical evidence that these air–sea interactions modulate the MAC is presented. Estimates of internal IAV have been made from observations as well as atmospheric model simulations. In contrast to a large part of the Tropics where the summer climate is predictable, with the internal variability being much smaller than the external one, the limited predictability of the Asian monsoon appears to be due to the fact that the contribution from the external IAV over the region is relatively weak and comparable to that from internal IAV. Cause for large internal IAV over the ASM region is investigated, and it is proposed that the internal IAV of the MAC is primarily due to interaction between the MAC and the summer intraseasonal oscillations (ISOs). Two mechanisms through which ISOs lead to internal IAV of the MAC are unraveled. The seasonal bias of the ISO anomalies can influence the seasonal mean if the spatial structure of the ISO has significant projection on that of the seasonal mean and if frequency of occurrence of positive and negative phases is unequal. Evidence supporting this is presented. In addition, it is demonstrated that the chaotic summer ISOs modulated by the annually varying forcing associated with the “slow annual cycle” can lead to IAV of the seasonal mean. Empirical evidence that IAV of ISO activity is related to IAV of the seasonal mean or MAC is also presented. Thus, the Asian monsoon would remain a difficult system to predict. To exploit the predictable signal, however, it is imperative that systematic bias of the models is improved and the space–time structure of the summer ISOs is simulated accurately. [ABSTRACT FROM AUTHOR]

The El Nino and Southern Oscillation (ENSO) 'diversity' has been considered as a major factor limiting its predictability, a critical need for disaster mitigation associated with the trademark climatic swings of the ENSO. Improving climate models for ENSO forecasts relies on deeper understanding of the ENSO diversity but currently at a nascent stage. Here, we show that the ENSO diversity thought previously as 'complex,' arises largely as varied contributions from three leading modes of the ENSO to a given event. The ENSO 'slow manifold' can be fully described by three leading predictable modes, a quasi-quadrennial mode (QQD), a quasi-biennial (QB) mode and a decadal modulation of the quasi-biennial (DQB). The modal description of ENSO provides a framework for understanding the predictability of and global teleconnections with the ENSO. We further demonstrate it to be a useful framework for understanding biases of climate models in simulating and predicting the ENSO. Therefore, skillful prediction of all shades of ENSO depends critically on the coupled models' ability to simulate the three modes with fidelity, providing basis for optimism for future of ENSO forecasts. [ABSTRACT FROM AUTHOR]