Clim Dyn (2016) 46:3845–3864 DOI 10.1007/s00382-015-2808-6

Mean state and interannual variability of the Indian summer monsoon simulation by NCEP CFSv2 Ravi P. Shukla1 · Bohua Huang1,2 

Received: 7 March 2015 / Accepted: 21 August 2015 / Published online: 26 August 2015 © Springer-Verlag Berlin Heidelberg 2015

Abstract  The capability of the National Centers for Environmental Prediction climate forecast system version 2 (CFSv2) in simulating the Indian summer monsoon (ISM) is evaluated in the context of the global monsoon in the Indo-Pacific domain and its variability. Although the CFSv2 captures the ISM spatial structure qualitatively, it demonstrates a severe dry bias over the Indian subcontinent. The weaker model monsoon may be related to an excessive surface convergence over the equatorial Indian Ocean, which reduces the moisture transport toward the Indian subcontinent. The excessively low equatorial pressure is in turn a part of a tropical-wise bias with the largest errors in the central and eastern equatorial Pacific associated with the cold sea surface temperature bias and an overly strong inter-tropical convergence zone. In this sense, the model bias in the tropical Pacific influences those in the Indian Ocean-ISM region substantially. The leading mode of the June–September averaged CFSv2 rainfall anomalies covering the ISM and its adjacent oceanic regions is qualitatively similar to that of the observations, characterized by a spatial pattern of strong anomalies over either side of the Indian peninsula as well as center of opposite sign over Myanmar. However, the model fails to reproduce the northward expansion of rainfall anomalies from Myanmar, leading to opposite anomalies over northeast India and Himalayas region. A substantial amount of the

* Ravi P. Shukla [email protected] 1

Center for Ocean‑Land‑Atmosphere Studies (COLA), 270 Research Hall, Mail Stop 6C5, George Mason University, 4400 University Drive, Fairfax, VA 22030, USA

2

Department of Atmospheric, Oceanic, and Earth Sciences, George Mason University, Fairfax, VA, USA





anomalous fluctuation is attributed to the El Niño and the Southern Oscillation (ENSO), although the model variability depends more strongly on ENSO. The active regional influences in the observations may contribute to its baroclinic vertical structure of the geopotential height anomalies in the ISM region, compared with the predominantly barotropic one in CFSv2. Model ENSO deficiencies also affects its ISM simulation significantly. Keywords  Indian summer monsoon · NCEP CFSv2 · Interannual variability · EOF analysis

1 Introduction The Indian summer monsoon (ISM) is one of the most important components of the climate system. The ISM season (June to September, JJAS) is characterized by prolonged and heavy precipitation, contributing 60–90 % to annual rainfall, which is of considerable importance to agriculture, energy, food security and water resource management in this region. Since a strong monsoon can result in floods while a weak monsoon may lead to droughts, both abnormal situations may cause devastating damage to the living condition and economy of the densely populated region. Therefore, the interannual variability of the ISM rainfall (ISMR, Shukla 1987) is of serious economic and social consequences (Webster et al. 1998). At present, the prediction of the ISM interannual variability remains a challenging scientific question. A hierarchy of models has been used to study the ISM predictability and to improve its prediction, ranging from very simple energy balance models to the most complicated state-of-the-art climate models that account for myriads of physical processes and are run on powerful supercomputers. However, dynamical

13

3846

monsoon prediction still suffers from many limitations, such as inadequate initialization and imperfect model physics, among others. In particular, the simulation of the ISM mean state and interannual variability needs to be improved significantly. The ISM is a process characterized by strong dynamical-thermodynamic feedbacks and land–ocean–atmosphere interactions. The deep heating source established in boreal summer over the Tibetan Plateau and South Asia determines the central location of the monsoon trough (Wu and Zhang 1998; Goswami and Xavier 2005). The cross-equatorial Somali jet (low-level Jet), originating from Mascrene High and flowing through the equatorial Arabian Sea into the Indian subcontinent (Findlater 1969), transports moisture from the southern Indian Ocean to the core monsoon region (Indian subcontinent and South Asia). As a part of the monsoon Hadley cell, the meridional thermal contrast also drives the upper-level tropical easterly jet stream (TEJ) along 15°S–15°N over southern India and the Indian Ocean (e.g., Krishnamurti and Bhalme 1976). The interannual variability of the ISM system is associated with both regional and global variations. In particular, the El Niño–Southern Oscillation (ENSO) is identified as the leading influence to ISM (Rasmusson and Carpenter 1983; Shukla and Paolin 1983; Webster and Yang 1992; Shukla et al. 2011; Shukla and Kinter 2014). Both observational and modeling studies have demonstrated that the ISMR tends to be below normal (above normal) during El Niño (La Niña) years (Webster and Yang 1992; Kirtman and Shukla 2000; Lau and Nath 2000; Shukla and Kinter 2014) although this reverse relationship between ENSO and the monsoon has weakened during recent decades (Krishna et al. 1999; Kinter et al. 2002). The basin-wide modes of the sea surface temperature (SST) anomalies in the Indian Ocean, such as the Indian Ocean dipole (IOD), also play a major role either in coordination with ENSO or independently (Saha 1970; Shukla 1975; Saji et al. 1999; Ashok et al. 2001; Huang and Kinter 2002; Huang and Shukla 2007a, b; Wu and Kirtman 2003; among others). Furthermore, the Arabian Sea SST anomalies can be another factor (e.g., Shukla and Huang 2015). Previous modeling studies have demonstrated that active ocean–atmosphere coupling is important in simulating both the Asian summer monsoon (ASM) climatology and its variability/predictability. Charney and Shukla (1981) hypothesized that the interannual variability of the tropical climate is mainly determined by the slowly varying surface boundary conditions such as SST, soil moisture and snow cover. Examining the summer monsoon rainfall anomalies simulated by 11 atmospheric general circulation models (AGCMs) that participate in the Climate Variability and Predictability Program (CLIVAR)/Monsoon Inter-comparison Project during 1997–1998 El Niño event, Wang et al.

13

R. P. Shukla, B. Huang

(2004) show a general lack of skill in summer precipitation simulations by these uncoupled models over Southeast Asia and the western North Pacific possibly because the prescribed SST assumes an infinite heat capacity of the ocean, which generates excessive atmospheric response over warm water. Further studies (e.g., Guilyardi et al. 2004; Wang et al. 2005; Wu and Kirtman 2005, 2007) show that thermodynamic air–sea feedback over the “warm ocean” is crucial for maintaining the realistic SST–precipitation relationship. More recently, Zhu and Shukla (2013) further demonstrate the role of the air–sea feedback in maintaining a realistic mean state of precipitation and its variability using coupled and uncoupled seasonal forecast experiments. Although the coupled ocean–atmosphere GCMs (CGCMs) generally simulate a realistic climatology of the global monsoon precipitation, as well as other statistical features, there are still serious systematic bias in its climatology, as well as in the leading patterns of the interannual variability. Some of these biases are clearly related to model resolution, such as those at the windward side of narrow mountains near the western coast of India and the steep slope of the Tibetan Plateau (Kim et al. 2008). Other biases, such as the dry bias over land and the northeastward shift of the intertropical convergence zone (ITCZ) in the tropical North Pacific that are common to many climate simulations, however, are more likely associated with inadequacies in model physics. Levine and Turner (2012) and Levine et al. (2013) also found that a cold SST bias in the Arabian Sea in many CGCMs directly affect the ISM precipitation through moisture transport. Identifying and eliminating the sources of these biases can further improve monsoon simulation. In this paper, we systematically evaluate the ISM simulation by the climate forecast system, version 2 (CFSv2). CFSv2 is the current operational forecast model for subseasonal-to-seasonal predictions at the National Centers for Environmental Prediction (NCEP) (Saha et al. 2014b). Both CFSv2 and its predecessor, CFSv1, have shown considerable skill of seasonal prediction in tropics (e.g., Saha et al. 2010, 2014b). Using long-term simulations and seasonal hindcasts, many previous studies have examined different aspects of the seasonal and interannual variability of the ASM in CFSv1 (e.g., Yang et al. 2008; Achuthavarier et al. 2012; Pokhrel et al. 2012; Chaudhari et al. 2013; Shukla and Kinter 2014) and CFSv2 (e.g., Jiang et al. 2013; Saha et al. 2014a; Zhu and Shukla 2013; Mishra and Li 2014), which have identified some major problems in these simulations and hindcasts. For instance, Chaudhari et al. (2013) reported that CFSv1 shows dry (wet) rainfall bias concomitant with cold (warm) SST bias over east (west) equatorial Indian Ocean. Shukla and Kinter (2014) found that CFSv1 is not able to reproduce the observed ENSO–ISM relationship in rainfall anomalies. Saha et al. (2014a) demonstrated

Mean state and interannual variability of the Indian summer monsoon simulation by NCEP CFSv2

that CFSv2 produces better spatial patterns of seasonal mean rainfall and circulation over extended Indian domain than CFsv1 does but the central Indian dry bias and cold SST bias in the Indian Ocean still persist in CFSv2. Shin and Huang (2015) further noticed that, in comparison with observations, CFSv2 generally simulates an earlier monsoon onset and stronger and longer lasting dry breaks and wet episodes that are phased-locked to the seasonal cycle in the ISM region. Therefore, its “fast annual cycle” significantly deviates from that of the observations (LinHo and Wang 2002). These systematic errors may affect the model’s capability to predict ISM. In fact, the correlation skill of the seasonal precipitation from the CFSv2 hindcasts is generally not high in the ISM and Indian Ocean domain (Zhu and Shukla 2013) although certain anomalous patterns seem to be predictable with the lead of one season (Zuo et al. 2013). In the light of these previous studies, we evaluate the ISM climatology and interannual variability simulated by CFSv2 in two new aspects. First, we analyze the ISM mean state as a component of the global summer monsoon regimes (Wang and Ding 2006) in the ASM and the Indo-Pacific domain. The broader view of the tropical climatology provides further insight on the potential sources of the ISM bias. Second, we examine the leading mode of the ISM interannual variability derived from a regional perspective in observation and model. In the latter aspect, we concentrate on the ISM connections with the ENSO variability within the ASM system. Our results demonstrate that some of the ISM systematic bias may originate from the basin-wide cold bias in the whole tropics, which may be amplified in the upper atmosphere. We also found that the phase relationship between the ISM and the northwestern Pacific summer monsoon (NWPSM) is weaker in the CFSv2 associated with leading mode of rainfall to that in observations. Further analysis shows that this model ENSO deficiency also influence its ISM simulation. The rest of paper is organized as follows. Section 2 describes the CFSv2, the experimental design, the verification dataset and the analysis method. In Sect. 3, we have discussed the ISM climatology in the tropical climate. Section  4 describes the characteristic of the major ISM modes in CFSv2 and it connections with ENSO. The summary and discussion are given in Sect. 5.

2 Model and data The CFSv2 is composed of interacting atmospheric, oceanic, sea ice and land component models. The atmospheric model is a lower-resolution version of the NCEP Global Forecast System at a T126 horizontal resolution (105-km grid spacing) and 64 vertical levels in a hybrid

3847

sigma-pressure coordinate. The ocean model is the Geophysical Fluid Dynamics Laboratory (GFDL) Modular Ocean Model (MOM) versions 4.0, which is configured for the global ocean with a horizontal grid of 0.5° × 0.5° poleward of 30°S and 30°N and a meridional resolution increasing gradually to 0.25° between 10°S and 10°N. The vertical coordinate is geopotential (z-) with 40 levels (27 of them in the upper 400 m). The maximum depth is approximately 4.5 km. The oceanic and atmospheric components exchange surface momentum, heat and freshwater fluxes, as well as SST, every 30 min. More details about CFSv2 can be found in Saha et al. (2014b). We have conducted a 30-year simulation using an updated version of the CFSv2 at the Center for Ocean–Land–Atmosphere Studies (COLA). In this experiment, we have included time varying GHG in the CFSv2 model. For this model version, an inconsistency of the model original code at the air–sea interface is eliminated, which leads to a significant improvement of the model simulation in northern higher latitudes, especially during boreal summer (Huang et al. 2015). The simulation is initiated from the realistic initial condition at November 1980. The monthly gridded Climate Prediction Centre Merged Analysis of Precipitation (CMAP) (Xie and Arkin 1997) and the NCEP/National Center for Atmospheric Research (NCEP/NCAR) global atmospheric reanalysis (Kalnay et al. 1996) for a 30-year period (1981–2010) are used for comparison. The National Oceanic and Atmospheric Administration (NOAA) optimum interpolation (OI) SST analysis, version 2 (OISST v2), for the period of record 1982–2010 has been used in the present study (Reynolds and Smith 1995). Both the CMAP data and the NCEP/ NCAR reanalysis are gridded at a resolution of 2.5° × 2.5° latitude/longitude while the OISSTv2 has a 1° × 1° resolution. For quantitative model-data comparisons, the CFSv2 outputs and observational data are first bi-linearly interpolated onto a common grid of 1° × 1° resolution, which is closer to the model resolution. We make this choice to avoid aliasing errors of the model fields if they were interpolated to a courser resolution. Although this interpolation strategy inflates the resolution of the observational analyses nominally, it does not add smaller-scale structures in the data. Monthly anomalies are derived for each dataset with respect to its own monthly climatology. Season mean anomalies are then calculated from the monthly anomalies. Regional ISM patterns of precipitation are derived from an empirical orthogonal function (EOF) analysis performed on the seasonal summer (JJAS) rainfall anomalies within the domain (1.5°N–37.5°N, 59.5°E–99.5°E), covering the extended Indian monsoon region and some parts of adjacent oceanic region, to be referred as “major ISM region” as in Shukla and Huang (2015) hereafter. We have adopted

13

3848 Fig. 1  The spatial distributions of seasonal (JJAS) mean climatology of rainfall (colored shading) and SST (contours) in a observations, b CFSv2 and c the climatological biases relative to observations and intervals of contour lines for SST bias are −2.5, −2.0, −1.5, −1.0, −0.5, 0, 0.5, 1, 1.5 and 2.5 °C. The precipitation unit is “mm/day” and SST unit is “degree Celsius”

R. P. Shukla, B. Huang

(a)

(b)

(c)

the EOF analysis approach of Shukla and Huang (2015) in this paper. To connect the regional ISM patterns to the global summer monsoon regimes, NINO3.4 (an index of El Niño and the Southern Oscillation, ENSO) are also derived in observation and the CFSv2. Most of the results shown in the paper are based on the dominant modes for major ISM region. Correspondingly patterns for other variables are derived from lead-lag regressions and/or correlations with respect to the leading principal components of these EOF modes, as well as other predefined climate indices. The statistical significance of these patterns is measured pointwise using a 2-sided student t test, given the sampling size, a correlation value of ±0.346 is statistically significant at 95 % confidence level.

13

3 Mean state and biases in CFSv2 In this section, we discuss the characteristic features of the ISM climatology in the context of the global monsoon precipitation (Wang and Ding 2006), especially ASM and those in the Indo-Pacific domain. Figure 1 shows the JJAS mean precipitation (shading) and SST (contour) in observations (Fig. 1a) and CFSv2 simulation (Fig. 1b) in this broader area. The corresponding model biases (CFSv2 minus observations) are given in Fig. 1c. Qualitatively, CFSv2 (Fig. 1b) captures the main observed precipitation pattern (Fig. 1a) of the ISM and its surrounding area, including its action centers over the western India, the

Mean state and interannual variability of the Indian summer monsoon simulation by NCEP CFSv2 Fig. 2  The spatial distributions of seasonal (JJAS) mean climatology of H850 (colored shading) and 850 hPa-wind (vectors) in a observations and b CFSv2. c The climatological biases relative to observations for H850 (colored shading) and 850-hPa wind (vectors)

3849

(a)

(b)

(c)

northern Bay of Bengal, and the equatorial Indian Ocean off the Sumatra coast. In the broader Indo-Pacific domain, major precipitations are over the oceanic warm pool encompassed by the 28 °C isotherm of SST, some of which expand into the Asian Continent (Fig. 1a). These precipitation centers are also reproduced by the CFSv2 to a certain extent (Fig. 1b), such as those located over the South China Sea and the northwestern Pacific, as well as the ITCZ and the South Pacific convergence zone (SPCZ). Quantitatively, the model demonstrates a severe dry bias over the northern India and in the western part of the Bay of Bengal, with precipitation deficit up to 4 mm/day. In the mean time, it also shows an excessive rainfall in the western coast of Thailand, mainly in the Andaman Sea (Fig. 1c). Some of these coastal precipitation errors may be associated with the inadequate simulation of the regional

circulation at the windward side of narrow mountains (e.g., Kim et al. 2008). However, major CFSv2 precipitation bias in the ISM region seems to be associated with a bias of the basin-wide atmospheric circulation over the Indian Ocean. Although the model simulates reasonably realistic interhemispheric monsoon circulation in the lower atmosphere, composed of the easterlies over the South Indian Ocean, the Somali Jet and the westerly winds passing through the Indian subcontinent (Fig. 2a, b), its easterly branch tends to turn more equatorward along its route (Fig. 2b). This generates the northwesterly wind bias over the Indian Ocean in 15°S–10°N (Fig. 2c) and causes excessive atmospheric convergences appear in the central equatorial Indian Ocean and near the eastern coast extending from the surface (Fig. 3a, b) to the lower troposphere (Fig. 3c, d).

13

3850

R. P. Shukla, B. Huang

(a)

(b)

(c)

(d)

Fig. 3  The spatial distributions of seasonal (JJAS) mean climatology of 1000 hPa-divergence (colored shading) in a observations and b CFSv2 and divergence is in units of 10−6 s−1. c, d as in a, b but for 850 hPa-divergence

Although the converging flow causes more intensive precipitation off the coastal regions from Sumatra to Thailand and Myanmar, the basin-wide wind bias also prevents the rainfall center near the Myanmar coast from expanding northwestward into the ISM region (Fig. 1b) as the observations do (Fig. 1a). We speculate that the early turning of the easterly winds over the South Indian Ocean also reduces the moisture transported to the Somali Jet, to be further carried into the ISM region. In addition, the Indian Ocean cold SST bias throughout the route of the monsoon winds, especially in the Arabian Sea, can also weaken the ISM precipitation. For instance, Levine et al. (2013) have reported that CMIP5 models having cold SST bias in the Arabian Sea generally have reduced moisture transport into the Indian subcontinent and weaker monsoon (Levine et al. 2013). The excessive convergence over the equatorial Indian Ocean in CFSv2 may in turn be caused by the errors in the mean sea level pressure (SLP, not shown) and the 850 hPa height within the equatorial zone (H850), which are significantly lower than those in the observations (e.g., Fig. 2a, b). The SLP and H850 errors in the Indian Ocean, however, seem to be part of a larger scale bias covering the whole Indo-Pacific domain. In fact, the low H850 bias appears in the equatorial zone (15°S–15°N) globally. It is largest in the central and eastern Pacific (Fig. 2c) where it is

13

associated with the northward cross-equatorial wind bias that generates overly strong convergence to the north of the equator extending from sea surface to 850 hPa (Fig. 3b, d). We argue that the cold SST bias in the equatorial Pacific Ocean plays a major role in generating this global-scale error pattern. The colder equatorial SST intensifies the northward surface meridional pressure gradient force near the equator throughout the Pacific basin, which accelerates the southerly winds in the lower atmosphere (Lidnzen and Nigam 1987). This in turn generates stronger model convergence (divergence) near 10°N (equator) in the lower atmosphere, which enhance the ITCZ precipitation and displace it further north (Fig. 1b). A major consequence of the intensified model ITCZ is a further reduction of the SLP (not shown) and H850 throughout the tropical Pacific between 15°S and 15°N (Fig. 2a, b; colored shading). Therefore, the oceanic and atmospheric biases can feedback to each other. We speculate that the global-scale systematic bias originating from the equatorial Pacific basin is a major source of the global monsoon bias, including the ISM (Fig. 2c). Therefore, we are urging that some controlled experiments are needed to further examine it. Correspondingly, a tropical-wise cold bias occurs in the upper troposphere, which is about 2°–4 °C on average for the 500–200 hPa mean temperature (Fig. 4c). As a result, the geopotential surface at 200 hPa is generally 60–120 m

Mean state and interannual variability of the Indian summer monsoon simulation by NCEP CFSv2

3851

(a)

(b)

(c)

Fig. 4  The spatial distributions of seasonal (JJAS) mean climatology of temperature between 500 and 200 hPa in a observations and b CFSv2. c The climatological biases relative to observations. Temperature unit is “degree Celsius”

lower in the model (Fig. 5a–c; color shading) in the tropics. Moreover, the north–south gradient of the 500–200 hPa mean temperature between the Indian subcontinent and the central Indian Ocean (known as TT), which is a very important component of ISM in order to sustain the monsoon circulation (Wu and Zhang 1998; Goswami and Xavier 2005), is underestimated in the model. The mean JJAS TT is dominated by elevated heat source of Tibetan plateau and magnitude of meridional heating gradient between Tibetan plateau and the central Indian is 3°–4 °C as seen in NCEP/NCAR reanalysis (Fig. 4a). Although CFSv2 (Fig. 4b) is able to simulate the warm center over

Tibetan plateau qualitatively, its magnitude of TT throughout the Indian subcontinent region is approximately 2 °C. Similarly, although the CFSv2 climatology (Fig. 5b) qualitatively reproduces the TEJ over southern India, the Bay of Bengal and adjoining Indian Ocean, the model TEJ has a magnitude approximately 5–8 m/s lower than that of the observations over the Indian Ocean and Indian continent region (Fig. 5c; in vector), consistent with its underestimate of TT. It may be possible that weak monsoon circulation pattern at 200 hPa leads to less rainfall over the Indian subcontinent, thus the dry bias over this region.

13

3852

R. P. Shukla, B. Huang

(a)

(b)

(c)

Fig. 5  The spatial distributions of seasonal (JJAS) mean climatology of geo-potential height at 200 hPa (colored shading) and 200 hPawind (vectors) in a observations and b CFSv2 and geo-potential

units of meter. c The climatological biases relative to observations for H200 (colored shading) and 200-hPa wind (vectors)

The mean state bias in the CFSv2 affects significantly the characteristics of its interannual variability. Although the typical features of observed interannual variability of rainfall anomalies (Fig. 6a) are well captured by the CFSv2 (Fig. 6b), the amplitude of the model variability is significantly larger than the observations over a portion of the western Pacific Ocean to the east of Indonesia and north of the equator (Fig. 6c). Maximum fluctuations of the precipitation over the equatorial Pacific are near the eastern boundary of the warm pool where the SST and its zonal gradient are high in observation (Fig. 6a, d). They are also most easily influenced by the fluctuation of

the cold tongue SST. We believe the characteristics of the CFSv2 mean state on these properties can largely explain its excessively active interannual variability of the precipitation over equatorial Pacific and the westward shift of its center. The CFSv2 cold tongue is too strong and extends westward (Fig. 6f), which pushes the boundary of the warm pool further west. The CFSv2 cold tongue SST fluctuations are also more intensive than the observations, which naturally transfer to the more active precipitation variability. The annual cycle of the standard deviation of the monthly Niño 3.4 SST index is plotted in Fig. 7. The annual cycle of the observed monthly Niño3.4 index (black line) shows a

13

3853

Mean state and interannual variability of the Indian summer monsoon simulation by NCEP CFSv2

(a)

(d)

(b)

(e)

(c)

(f)

Fig. 6  Interannual standard deviation of JJAS rainfall anomalies in a CMAP and b CFSv2 simulation. c Interannual standard deviation of JJAS rainfall anomalies difference between CFSv2 and CMAP. d, e as in a, b but for JJAS SST anomalies. f as in c but for JJAS SST

Fig. 7  Annual cycle of the variance of the monthly anomalies of the Nino3.4 index for observation (black line) and CFSv2 (red line)

maximum value in December and January and a minimum in April, demonstrating a strong ENSO phase lock to the seasonal cycle. CFSv2 (red line) is able to reproduce the “U” shape of monthly interannual standard deviation but it shows high value of variance during February to July in comparison to observations.

4 Leading mode of ISM in observations and CFSv2 and its connection with ENSO To investigate the regional dominate mode of variability on interannual time scales of ISMR, an empirical orthogonal

13

3854

R. P. Shukla, B. Huang

(a)

(b)

(c)

(d)

Fig. 8  Spatial distribution of leading empirical orthogonal functions (EOF1) mode of averaged summer rainfall anomalies covering the extended Indian monsoon region and some parts of adjacent oceanic

region for June–September (JJAS) in a CMAP and b CFSv2. Yearto-year variation of standardized PC1 and JJAS Niño 3.4 index in c observations and d CFSv2

function (EOF) analysis is performed on the JJAS seasonal anomalies of CMAP and CFSv2 over the domain (1.5°N–37.5°N, 59.5°E–99.5°E) covering the extended

Indian monsoon region and some parts of adjacent oceanic region. Figure 8 depicts the leading mode (EOF1) of JJAS seasonal anomalies of rainfall for CMAP (Fig. 8a)

13

Mean state and interannual variability of the Indian summer monsoon simulation by NCEP CFSv2

and CFSv2 (Fig. 8b). The dominant modes of JJAS seasonal anomalies of rainfall explain 27.05 and 18.07 % variance in CMAP and CFSv2 respectively. The leading EOF1 of CMAP (Fig. 8a) has two maxima over the Western Ghats of the Indian and Bay of Bengal and opposite sign over the northeast Indian, Myanmar and near the foothills of the Himalayas. The first EOF shows strong variability over monsoon trough region in CMAP. The amplitudes of the rainfall perturbations at the major centers of action of EOF1 range up to 1.5 mm/day per standard deviation of its standardized principal component (PC1). The spatial pattern of rainfall variability in the CFSv2 (Fig. 8b) is in good agreement with the observations over the Western Ghats of India and the Bay of Bengal as well as center of opposite sign over Myanmar. However, the CFSv2 fails to reproduce the northward expansion of rainfall anomalies from Myanmar, leading to positive anomalies over northeast India and Himalayas region and the model also depicts opposite sign of anomalies over the northern part of the central Indian Ocean. Hereafter, the PC1s of both CMAP and CFSv2 modes (black curves; Fig. 8c, d) will be used to represent the interannual variability of the Indian monsoon rainfall over the “major ISM region” defined before, will be referred to as the PC1. Shukla and Huang (2015) have discussed in details about regional dominant mode of summer rainfall anomalies in the observation on interannual time scale. The simultaneous correlation coefficient (CCs) between the normalized time series of PC1 (black curve, Fig.  8c, d) and JJAS Niño3.4 SST index (green curve; Fig.  8c, d) in CFSv2 is larger (−0.60) than the observation (−0.42). It can be seen (Fig. 8c) that there are many years (1984, 1985, 1992, 1994, 1997, 1999, 2000 and 2006) when normalized PC1 and JJAS Niño3.4 SST index are showing out-of-phase relationship in observation. We have also found that amplitude and structure of correlation patterns between JJAS rainfall anomalies over global tropics and standardized PC1 and JJAS Niño3.4 index in the CFSv2 (Fig. 9b, d) respectively are almost similar. Based on spatial and temporal correlation analysis, we may conclude that PC1 in CFSv2 simulation may be highly depend on ENSO in comparison to observation. On the other hand, it is observed that amplitude and structure of correlation patterns of standardized PC1 and JJAS Niño3.4 index in observation (Fig. 9a, c) are not agreement over the northern Indian Ocean, Arabian Sea, Western Ghats of India and the central western Pacific Ocean. It may conclude that regional mode in observation (Fig. 8a) not only depends on ENSO but also depends on regional processes associated with it (Shukla and Huang 2015). We have identified some major differences in correlation patterns between JJAS rainfall anomalies over the global tropics and normalized PC1 of observation and CFSv2 (Fig. 9a, b) respectively. We have found a sharp

3855

contrast between positive correction over the maritime continent, Arabian Sea, Bay of Bengal and negative correction over the central Pacific Ocean, South China Sea, equatorial Pacific Ocean in observations (Fig. 9a). The amplitude and structure of correction pattern over the northerncentral western Pacific Ocean (4°N–20°N, 125°E–180°E) and north of the equator in the central Pacific Ocean simulated by CFSv2 (Fig. 9b) is not in agreement with CMAP (Fig. 9a). On close inspection, we have found that CFSv2 produces wrong center of action in the central Indian Ocean, mainly (5°S–6°N, 55°E–90°E) (Fig. 9b). Following reviewer suggestion, an EOF analysis is performed on the JJAS seasonal anomalies of the Global Precipitation Climatology Project (GPCP v2.2; Adler et al. 2003) over the domain (1.5°N–37.5°N, 59.5°E–99.5°E) and compared with the one with those derived from the CMAP data. The dominant modes of JJAS seasonal anomalies of rainfall explain comparable percentages of the total variances in this basin, 27.05 and 25.72 % variance in CMAP and GPCP respectively. The structure and amplitude of rainfall anomalies of GPCP is also in a good agreement with CMAP except that the pattern of EOF-1 (not shown) has weaker amplitudes over Western Ghats of India and the Bay of Bengal, and slightly stronger negative anomalies over some part of the eastern India. The simultaneous correlation coefficients between the normalized time series of PC1 and JJAS Niño3.4 SST index are −0.42 for the CMAP and −0.43 for the GPCP. In our further analysis, we performed EOF analysis of JJAS precipitation anomalies to identify the primary spatial structure and temporal evolution of interannual variability over the region encompassing the ISM and ENSO (30°S–30°N, 30°E–289°E) in GPCP. Again, structure and amplitude of rainfall anomalies of GPCP (not shown) associated with global leading mode of summer rainfall anomalies is in a good agreement with CMAP. The simultaneous correlation coefficients between normalized time series of global leading mode and JJAS Niño3.4 SST index are −0.92 for the CMAP and −0.93 for the GPCP. Based on this EOF analysis, we conclude that overall CMAP and GPCP reproduced similar structure associated with regional and global leading mode of JJAS rainfall anomalies in our case. We will not discuss further about GPCP in this paper because here we employed CMAP as the principal data. Now, we have examined the structure of the patterns in the SST, SLP, H850, H200 and wind at 850 and 200 hPa associated with the PC1 in both observation and CFSv2. Figure  10 depicts the spatial distribution of correlation coefficient between March–May (MAM), JJAS and September–November (SON) SST anomalies and PC1 in observation and CFSv2. The simultaneous correlation map with JJAS SST in observation (Fig. 10c) is characterized by prominent SST cooling over the equatorial Pacific Ocean

13

3856

R. P. Shukla, B. Huang

(a)

(b)

(c)

(d)

Fig. 9  Spatial distributions of the correlation coefficient between JJAS rainfall anomalies in the global tropics and standardized PC1 in a observation and b CFSv2. c, d as in a, b but for standardized JJAS Niño 3.4 SST index

13

Mean state and interannual variability of the Indian summer monsoon simulation by NCEP CFSv2

(a)

(b)

(c)

(d)

(e)

(f)

3857

Fig. 10  Spatial distributions of the correlation coefficient between MAM SST anomalies and standardized PC1 in a observation and b CFSv2. c, d as in a, b but for JJAS SST anomalies. e, f as in a, b but for SON SST anomalies

whereas SST warming over the western Pacific Ocean, South China Sea and the Arabian Sea. The correlation map of SON SST (Fig. 10e) associated with PC1 indicates that the central Pacific cooling signifies a developing La Niña. On the other hand, the CFSv2 reproduces correctly sign of SST anomalies in the eastern Pacific Ocean during JJAS and SON season (Fig. 10d, f) but during these periods MAM to SON, model also produces too strong negative SST anomalies (Fig. 10b, d, f) in the eastern equatorial Pacific in comparison to observation (Fig. 10a, c, e). We have obtained similar CFSv2-observation discrepancies between SST anomalies and normalized Niño 3.4 SST index in these seasons (not shown). Wang et al. (2004, 2005) pointed out that generally a positive correlation between seasonal mean SST and rainfall anomalies may indicate that the atmosphere responses is forced by SST anomalies; conversely a negative correlation may indicate that SST anomalies is forced by the atmosphere. Shukla and Huang (2015) identified that SST condition over the Arabian Sea may be an important contributor to ISM predictability from monthly to seasonal scales during summer monsoon. In observation (Fig. 10c), the sign of the correlation is positive over the Arabian Sea,

while CFSv2 (Fig. 10d) simulation erroneously produces a negative correlation over the Arabian Sea during JJAS season. This may also be a factor that effect the ISM in the CFSv2 simulation. Figure  11 depicts the simultaneous correlation patterns for the JJAS 850-mb wind anomalies field, which is superimposed upon the correlation pattern of JJAS H850 anomalies associated with the PC1 in both observation and CFSv2. Statistically significant negative correlation of H850 over the Arabian Sea and the central Indian Ocean, and positive correlation over the western and eastern Pacific Ocean region is a signature of typical ENSO response in the observations (Fig. 11a) (e.g. Wallace et al. 1998). Consistent with this result, correlations pattern of JJAS anomalies of wind-850 hPa (Fig. 11a; vector plot) shows a band of easterly wind anomalies from the central western Pacific Ocean to the Bay of Bengal and the Indian Ocean with peak amplitudes on the order of 1.5 m/s (not shown). The strongest easterlies are located between bands of 5°S to 15°N degree of latitude. It transports the moisture from the western Pacific Ocean to the Indian Ocean, and enhances convection over the Indian Ocean and the Arabian Sea. On closer inspection, we have observed anticyclone over the

13

3858

R. P. Shukla, B. Huang

(a)

(b)

Fig. 11  Spatial distributions of the correlation coefficient between JJAS seasonal anomalies of H850 (colored shading) and 850 hPa-wind (vectors) over the Indo-Pacific basin and standardized PC1 in a observation and b CFSv2

central western Pacific region, mainly around 15°N, 145°E (Fig.  11a). The anticyclone structure (Fig. 11a, in vector) over the western Pacific Ocean associated with less rainfall (Fig. 9a) over the central western Pacific Ocean, South China Sea and the northern Bay of Bengal and also favor the above normal H850 anomalies (Fig. 11a, in shaded) over the central western Pacific Ocean and South China Sea. CFSv2 (Fig. 11b) simulation correctly produces a negative correlation of H850 over the Arabian Sea and the central Indian Ocean with slightly stronger magnitude in comparison to observation (Fig. 11a) but it is unable to reproduces positive correlation over the northern western Pacific Ocean, South China Sea and Philippines Sea mainly (10°N–23°N, 125°E–175°E) in comparison to observation (Fig. 11a). The CFSv2 (vector; Fig. 11b) is able to reproduce a band of strong easterly anomalies extending from the western Pacific into the central Indian Ocean between 5°S and 15°N. On close inspection, we have found a

13

cyclonic structure over the central western Pacific Ocean mainly around 15°N, 145°E in CFSv2 (Fig. 11b; in vector). Based on these results, we may conclude that CFSv2 produces weaker ISM phase relationship with respect to the H850 anomalies over the northwestern Pacific Ocean in comparison to the observations during summer season. This is possibly because ENSO is overly active in CFSv2 (Fig. 10b, d, f) simulation during and before the monsoon season and persist for a longer time period in comparison to observation. Corresponding correlation patterns of H850 (Fig. 11a) and H200 (Fig. 12a) associated with PC1 in observation show that the equatorial atmospheric anomalies generally have opposite signs between lower and upper troposphere, especially over the marine continent and the southern Indian Ocean, with a typical baroclinic structure. The pattern of correlation between PC1 and 200 hPawind (Fig. 12a; in vector) shows a band of westerly wind

Mean state and interannual variability of the Indian summer monsoon simulation by NCEP CFSv2

3859

(a)

(b)

Fig. 12  Spatial distributions of the correlation coefficient between JJAS seasonal anomalies of H200 (colored shading) and 200 hPa-wind (vectors) over the Indo-Pacific basin and standardized PC1 in a observation and b CFSv2

anomalies from the maritime continent to central western Pacific Ocean in observations. The strongest westerlies are located between bands of 12°S to 15°N degree of latitude. The easterly wind is converging towards the maritime continent at 850-hPa (Fig. 11a) and diverging away from the maritime continent to the central western Pacific Ocean at 200-hPa (Fig. 12a) in the observation. These wind anomalies are consistent with positive correlation in rainfall over maritime continent, and negative correlation in rainfall over the central western Pacific Ocean (Fig. 9a). In CFSv2, the pattern of correlation between H200 and PC1 (Fig. 12b) shows negative correlation over extending from the Indian Ocean to the eastern equatorial Pacific, typical of the ENSO response. CFSv2 is unable to capture a band of westerly wind anomalies from the maritime continent to central western Pacific Ocean. It is necessary to mentioned that sign of correlation coefficient in geo-potential height anomalies in lower (850 hPa) to upper (200 hPa)

troposphere associated to PC1 in the CFSv2 (Figs. 11b, 12b) over the Indian Ocean and marine continent is same with a typical “barotropic structure”. Why does the CFSv2 reproduce large discrepancy at 200 hPa level in correlation pattern of H200 (200-wind) anomalies associated PC1 in comparison to observation? To answer this question, we have calculated correction coefficient (Fig. 13) between JJAS H200 (200 mb-wind) anomalies and JJAS Niño3.4 index (the sign of the Niño3.4 index is reversed to be consistent with the PC1 pattern) for both CFSv2 and the observations. We can clearly see that the correlation patterns in H200 anomalies associated with PC1 (Fig. 12b) and JJAS Niño3.4 index (Fig. 13b) are very similar in the CFSv2 over the entire tropics. More interestingly, the NINO3.4 correlation patterns from the observations (Fig. 13a) show certain resemblance to those with respect to the ISM PC1 (Fig. 12b) and NINO3.4 (Fig. 13b). They demonstrate that the ENSO signals dominate the

13

3860

R. P. Shukla, B. Huang

(a)

(b)

Fig. 13  Spatial distributions of the correlation coefficient between JJAS seasonal anomalies of H200 (colored shading) and 200 hPa-wind (vectors) over the Indo-Pacific basin and standardized JJAS Niño 3.4 SST index in a observation and b CFSv2

200 hPa circulation patterns for the ISM precipitation in CFSv2 while the PC1 of the observed ISMR are influenced by other processes (Shukla and Huang 2015). The ENSO–monsoon relation in the CFSv2 is further evaluated by computing lead-lag correlations between 3-month averaged monthly NINO3.4 index (SSTA average in 170°–120°W, 5°S–5°N) and the June–September (JJAS) averaged extended Indian monsoon rainfall index (EIMR; Wu and Kirtman, 2003). The EIMR is represented by an average of summer rainfall anomalies over the region (5°–25°N, 60°–100°E), following Wu and Kirtman (2003) and Shukla and Kinter (2014). Observationally, the EIMR is positively correlated to the NINO3.4 of the previous fall and winter but has the largest negatively correlations with the NINO3.4 during September–October following the monsoon season in observations (black line in Fig. 14, see also Shukla and Kinter 2014). The evolution of the correlation in the CFSv2 long run (red line in Fig. 14) resembles

13

the observation (black line in Fig. 14) in that the largest negative correlation occurs to the NINO3.4 index following the monsoon season in late fall and early winter, which is higher than the observed but slightly delayed. The largest discrepancy between the observations and CFSv2 occurs in year(−1), when the observed correlation is significantly positive, but the correlation between the EIMR and NINO3.4 in CFSv2 is insignificantly negative at all lags before the monsoon season. This is mainly because the model ENSO is more persistent and does not show a quick transition in winter–spring seasons. In our previous article, Shukla and Kinter (2014) mentioned that the evolution of the correlation in the CFSv1 long run does not mimic the observed relationship, having insignificant correlations at all lags between EIMR and NINO3.4 index. In this work, we have found that the CFSv2 is able to reproduced EIMR and NINO3.4 index relationship with some extent.

Mean state and interannual variability of the Indian summer monsoon simulation by NCEP CFSv2

3861

Fig. 14  The evolution of lead-lag correlation coefficients between 3-month averaged monthly Niño 3.4 (170°W–120°W, 5°S–5°N) SST index and JJAS extended Indian monsoon rainfall index (EIMRI; Wu and Kirtman 2003). The black lines denote the observation and the red line denotes CFSv2 simulation. The green lines denote the 95 % significance level

The second modes (EOF2) (not shown) of JJAS seasonal anomalies explain 9.81 and 12.95 % domain-accumulated variances in CMAP and CFSv2 respectively. Therefore the EOF2s are distinctive of their respective EOF1s, according to the rule of thumb for uniqueness by North et al. (1982). The patterns of the EOF2s are also physically meaningful. The spatial pattern of precipitation in CMAP places the heaviest loading over the central India. This structure is closer to a major phase of monsoon intraseasonal oscillation (MISO) (Shukla 2014; Shukla and Zhu 2014). CFSv2 reproduces sign of loading over central India, southern India and the eastern Arabian Sea. We have identified an important feature associated with PC2 of observation and CFSv2 when we project JJAS rainfall anomalies in global tropics upon the standardized PC2 of CMAP and CFSv2 (not shown). We have found positive loading over the central-eastern Indian Ocean and negative loading over the central western Pacific region in CMAP. The CFSv2 successfully reproduced positive and negative loading of rainfall over the central-eastern Indian Ocean and central western Pacific region respectively. Since MISO is beyond the scope of this paper, we will further analyze the EOF2 of JJAS rainfall anomalies in a further study.

5 Summary and discussion In this study, we examine the capability of the NCEP climate forecast system version 2 (CFSv2) in simulating the ISM, with an emphasis on its relationship to the global monsoon in the Indo-Pacific domain and the remote forcing

from the ENSO variability. We have employed 30-year data for EOF analysis in both observations and CFSv2. The record length may be too short to get robust spatial pattern but it is long enough to begin to consider them for EOF analysis. Correlation coefficients at most of the dominant centers of action in this paper are above the reference value (0.346) for 95 % confidence level based on student 2-sided test. It is demonstrated that the CFSv2 captures the spatial structure of the mean ISM precipitation centered over the western India and the northern Bay of Bengal, as well as the equatorial Indian Ocean off the Sumatra coast. Quantitatively, however, the model demonstrates a severe dry bias over the Indian subcontinent, which may be connected to its deficiencies in simulating the inter-hemispheric monsoon circulation over the Indian Ocean. For instance, the model easterly winds over the central Indian Ocean to the south of the equator turn too early equatorward along its route to the Somali Jet, which generates excessive convergence over the central and eastern equatorial Indian Ocean. This northwesterly wind bias over the central equatorial Indian Ocean intensifies the precipitation near the coasts from Sumatra to Thailand and prevents the northwestward spread of the rainfall from Myanmar. It also reduces the moisture transport through the Somali Jet into the Indian subcontinent. A secondary source of the ISM dry bias may be the cold SST bias throughout the Indian Ocean along the route of the anticlockwise monsoon circulation, especially those in the Arabian Sea. The lower SST also reduces the surface evaporation and the moisture supply of the monsoon flow.

13

3862

There exists an excessively low atmospheric pressure around the equatorial Indian Ocean that drives equatorial convergence and the wind bias there. From a global perspective, this excessively low pressure is a part of a prevailing pattern throughout the tropical Indo-Pacific domain with the largest errors in the central and eastern Pacific. It is suggested that a potential cause behind this basin-wide pressure bias is the cold SST bias in the equatorial Pacific. The stronger meridional SST gradient caused by cold equatorial bias forces excessive southerly winds across the equator in the Pacific Ocean (e.g., Lidnzen and Nigam 1987). The resulting convergence generates an overly strong precipitation in the inter-tropical convergence zone (ITCZ) while displacing it northward. It is noticeable that a positive air–sea feedback is induced through this process. The heating sources associated with the displaced deep convection may generate remote forcing in the Indian Ocean basin. The intensified cold tongue in the equatorial Pacific also causes tropical-wise cooling throughout the tropospheric atmosphere (e.g., Yulaeva and Wallace 1994), which may further suppress the ISM precipitation. In this sense, the model bias in the tropical Pacific influences those in the Indian Ocean-ISM region substantially. Although the monsoon trough over the ISM are relatively well simulated, its connection with the tropical Pacific circulation may not be as realistic. The leading mode of the June–September averaged CFSv2 rainfall anomalies covering the ISM and its adjacent oceanic regions is qualitatively similar to that of the observations, characterized by a spatial pattern of strong anomalies over either side of the Indian peninsula as well as center of opposite sign over Myanmar. However, the model fails to reproduce correct sign of rainfall anomalies over northeast India and Himalayas region. A substantial amount of the anomalous fluctuation is attributed to the ENSO, although the model variability depends almost exclusively on ENSO while the observed one is also affected by other more regional sources. The active regional influences in the observations may account for its baroclinic vertical structure of the geopotential height anomalies in the ISM region, compared with the predominantly barotropic one in CFSv2. Model ENSO deficiencies strongly influence its ISM simulation. It is demonstrated that the model ENSO is not as strongly phase-locked to the annual cycle as the observations do. Some of the ENSO events seem to initiate earlier while others may persist longer. As a result, the SST anomalies in the NINO3.4 region during summer season are more active than the observations. These model discrepancies may affect its ENSO–monsoon relationship, as well as its connections to the summer monsoons in other regions. For example, CFSv2 produces a weaker ISM phase relationship with respect to the H850 anomalies over the northwestern Pacific Ocean in comparison to the observations

13

R. P. Shukla, B. Huang

during summer. Further, it is shown that during and before the monsoon season (JJAS), ENSO is overly acting in CFSv2 simulation and persist for a longer time period in comparison to observation. Acknowledgments  Funding for this study was provided by Grants from the National Science Foundation (ATM-0830068, 0947837 and 1338427), the National Oceanic and Atmospheric Administration (NA09OAR4310058 and NA14OAR4310160), and the National Aeronautics and Space Administration (NNX09AN50G, NNX09AI84G and NNX14AM19G). The authors thank Prof. Jagadish Shukla, Prof. James L. Kinter III and national monsoon mission (NMM) group at COLA/GMU for their helpful comments. We also thank Dr. M. A. Balmaseda from ECMWF for providing their ocean initial conditions and L. Marx’s help in the experiment setup. Computing resources provided by the Extreme Science and Engineering Discovery Environment (XSEDE) division are also gratefully acknowledged. Compliance with ethical standards  Conflict of interest  The authors declare that they have no conflict of interest.

References Achuthavarier D, Krishnamurthy V, Kirtman BP, Huang B (2012) Role of Indian Ocean in the ENSO–Indian summer monsoon teleconnection in the NCEP climate forecast system. J Clim 25:2490–2508 Adler RF, Huffman GJ, Chang A, Ferraro R, Xie PP, Janowiak J, Rudolf B, Schneider U, Curtis S, Bolvin D, Gruber A, Susskind J, Arkin P, Nelkin E (2003) The version-2 Global Precipitation Climatology Project (GPCP) monthly precipitation analysis (1979–present). J Hydrometeorol 4:1147–1167 Ashok K, Guan Z, Yamagata T (2001) Impact of the Indian Ocean dipole on the relationship between the Indian monsoon rainfall and ENSO. Geophys Res Lett 28:4499–4502 Charney JG, Shukla J (1981) Predictability of monsoons. In: Lighthill J, Pearce RP (eds) Monsoon dynamics. Cambridge University Press, Cambridge, pp 99–109 Chaudhari HS, Pokhrel S, Saha SK, Dhakate A, Yadav RK, Salunke K, Mahapatra S, Sabeerali CT, Rao SA (2013) Model biases in long coupled runs of NCEP CFS in the context of Indian summer monsoon. Int J Climatol 33:1057–1069 Findlater J (1969) A major low-level air current near the Indian Ocean during the northern summer. Q J R Meteorol Soc 95:362–380 Goswami BN, Xavier PK (2005) Dynamics of internal interannual variability of the Indian summer monsoon in a GCM. Geophys Res Lett 110:D24104. doi:10.1029/2005JD006042 Guilyardi E, Gualdi S, Slingo J, Navarra A, Delecluse P, Cole J, Madec G, Roberts M, Latif M, Terray L (2004) Representing El Niño in coupled ocean–atmosphere GCMs: the dominant role of the atmospheric component. J Clim 17:4623–4629 Huang B, Kinter JL III (2002) Interannual variability in the tropical Indian Ocean. J Geophys Res 107(C11):3199. doi:10.1029/200 1JC001278 Huang B, Shukla J (2007a) Mechanisms for the interannual variability in the tropical Indian Ocean—part I: the role of remote forcing from the tropical Pacific. J Clim 20:2917–2936 Huang B, Shukla J (2007b) Mechanisms for the interannual variability in the tropical Indian Ocean—part II: regional processes. J Clim 20:2937–2960

Mean state and interannual variability of the Indian summer monsoon simulation by NCEP CFSv2 Huang B, Zhu J, Marx L, Wu X, Kumar A, Hu ZZ, Balmaseda MA, Zhang S, Lu J, Schneider EK, Kinter JL (2015) Climate drift of AMOC, North Atlantic salinity and arctic sea ice in CFSv2 decadal predictions. Clim Dyn 44:559–583. doi:10.1007/ s00382-014-2395-y Jiang X, Yang S, Li Y, Kumar A, Liu X, Zuo Z, Jha B (2013) Seasonalto-interannual prediction of the Asian summer monsoon in the NCEP climate forecast system version 2. J Clim 26:3708–3727 Kalnay E et al (1996) The NCEP/NCAR 40-year reanalysis project. Bull Am Meteorol Soc 77:437–471 Kim HJ, Wang B, Ding Q (2008) The global monsoon variability simulated by CMIP3 coupled climate models. J Clim 21:5271–5294. doi:10.1175/2008JCLI2041.1 Kinter JL III, Miyakoda K, Yang S (2002) Recent change in the connection from the Asian Monsoon to ENSO. J Clim 15:1203–1215 Kirtman BP, Shukla J (2000) Influence of the Indian summer Monsoon on ENSO. Q J R Meteorol Soc 126:213–239 Krishna KK, Rajagopalan B, Cane M (1999) On the weakening relationship between the Indian monsoon and ENSO. Science 284:2156–2159 Krishnamurti TN, Bhalme HN (1976) Oscillations of a monsoon system—part I: observational aspects. J Atmos Sci 33:1937–1954 Lau NC, Nath MJ (2000) Impact of ENSO on the variability of the Asian-Australian monsoons as simulated in GCM experiments. J Clim 13:4287–4307 Levine RC, Turner AG (2012) Dependence of Indian monsoon rainfall on moisture fluxes across the Arabian Sea and the impact of coupled model sea surface temperature biases. Clim Dyn 38:2167–2190. doi:10.1007/s00382-011-1096-z Levine RC, Turner AG, Marathayil D, Martin GM (2013) The role of northern Arabian Sea surface temperature biases in CMIP5 model simulations and future projections of Indian summer monsoon rainfall. Clim Dyn 41:155–172. doi:10.1007/ s00382-012-1656-x Lidnzen RS, Nigam S (1987) On the role of sea surface temperature gradients in forcing low-level winds and convergence in the tropics. J Atmos Sci 44:2418–2436 LinHo, Wang B (2002) The time–space structure of the Asian-Pacific monsoon: a fast annual cycle view. J Clim 15:2001–2019 Mishra V, Li H (2014) The seasonal predictability of the Asian summer monsoon in a two-tiered forecast system. Clim Dyn. doi:10.1007/s00382-013-1838-1 North R, Bell T, Cahalan R, Moeng F (1982) Sampling errors in the estimation of empirical orthogonal functions. Mon Weather Rev 110:699–706 Pokhrel S, Chaudhari HS, Saha SK et al (2012) ENSO, IOD and Indian summer monsoon in NCEP climate forecast system. Clim Dyn 39:2143–2165 Rasmusson EM, Carpenter TH (1983) The relationship between eastern equatorial Pacific sea surface temperature and rainfall over India and Sri Lanka. Mon Weather Rev 111:517–528 Reynolds RW, Smith TM (1995) A high-resolution global sea surface temperature climatology. J Clim 7:1571–1583 Saha K (1970) Zonal anomaly of sea surface temperature in equatorial Indian Ocean and its possible effect upon monsoon circulation. Tellus 22:403–409 Saha S et al (2010) The NCEP climate forecast system reanalysis. Bull Am Meteorol Soc 91:1015–1057 Saha SK, Pokhrel S, Chaudhari HS, Dhakate A, Shewale S, Sabeerali CT, Salunke K, Hazra A, Mahapatra S, Rao AS (2014a) Improved simulation of Indian summer monsoon in latest NCEP climate forecast system free run. Int J Climatol 34:1628–1641 Saha S, Moorthi S, Wu X, Wang J, Nadiga S, Tripp P, Behringer D, Hou Y-T, Chuang H, Iredell M, Ek M, Meng J, Yang R, Malaquías Mendez P, van den Dool H, Zhang Q, Wang W, Chen

3863

M, Emily B (2014b) The NCEP climate forecast system version 2. J Clim 27:2185–2208 Saji NH, Goswami BN, Vinayachandran PN, Yamagata T (1999) A dipole mode in the tropical Indian Ocean. Nature 401:360–363 Shin C-S, Huang B (2015) Slow and fast annual cycles of the Asian summer monsoon in the NCEP CFSv2. Clim Dyn (under review) Shukla J (1975) Effect of Arabian sea-surface temperature anomaly on Indian summer monsoon: a numerical experiment with the GFDL mode. J Atmos Sci 32:503–511 Shukla J (1987) Interannual variability of monsoons. In: Fein JS, Stephens PI (eds) Monsoons. Wiley, New York Shukla RP (2014) The dominant intraseasonal mode of intraseasonal South Asian summer monsoon. J Geophys Res Atmos. doi:10.10 02/2013JD020335 Shukla RP, Huang B (2015) Interannual variability of the Indian summer monsoon associated with the air–sea feedback in the northern Indian Ocean. Clim Dyn. doi:10.1007/s00382-015-2687-x Shukla RP, Kinter JL III (2014) Simulation of the Asian Monsoon using a regionally-coupled global climate mode. Clim Dyn. doi:10.1007/s00382-014-2188-3 Shukla J, Paolin DA (1983) The Southern oscillation and long range forecasting of the summer monsoon rainfall over India. Mon Weather Rev 11:1830–1837 Shukla RP, Zhu J (2014) Simulations of boreal summer intraseasonal oscillations with CFSv2 over India and western Pacific: role of air–sea coupling. Atmos Ocean. doi:10.1080/07055900.2014.93 9575 Shukla RP, Tripathi KC, Pandey AC, Das IML (2011) Prediction of Indian Summer monsoon rainfall using Niño indices: a neural network approach. Atmos Res 102(1):99–109. doi:10.1016/j. atmosres.2011.06.013 Wallace JM, Rasmusson EM, Mitchell TP, Kousky VE, Sarachik ES, vonStorch H (1998) On the structure and evolution of ENSO-related climate variability in the tropical Pacific: lessons from TOGA. J Geophys Res 103(C7):14241–14259. doi:10.1029/97JC02905 Wang B, Ding Q (2006) Changes in global monsoon precipitation over the past 56 years. Geophys Res Lett 33:L06711. doi:10.102 9/2005GL025347 Wang B, Kang IS, Lee J-Y (2004) Ensemble simulations of Asian-Australian monsoon variability by 11 AGCMs. J Clim 17:803–818 Wang B, Ding Q, Fu X, Kang I-S, Shukla J, Doblas-Reyes F (2005) Fundamental challenge in simulation and prediction of summer monsoon rainfall. Geophys Res Lett 32:L15711. doi:10.1029/2 005GL022734 Webster PJ, Yang S (1992) Monsoon and ENSO: selectively interactive systems. Q J R Meteorol Soc 118:877–926 Webster PJ, Magana VO, Palmer TN, Shukla J, Tomas RA, Yanai M, Yasunari T (1998) Monsoons: processes, predictability and the prospectsfor prediction. J Geophys Res 103:451–510 Wu R, Kirtman BP (2003) On the impacts of the Indian summer monsoon on ENSO in a coupled GCM. Q J R Meteorol Soc 129B:3439–3468 Wu R, Kirtman BP (2005) Roles of Indian and Pacific Ocean air– sea coupling in tropical atmospheric variability. Clim Dyn 25:155–170 Wu R, Kirtman BP (2007) Regimes of seasonal air–sea interaction and implications for performance of forced simulations. Clim Dyn 29:393–410 Wu G, Zhang Y (1998) Tibetan plateau forcing and the timing of the monsoon onset over South Asia and the South China Sea. Mon Weather Rev 126:913–927 Xie P, Arkin PA (1997) Analysis of global monthly precipitation using gauge observations, satellite estimates and numerical predictions. J Clim 9:840–858

13

3864 Yang S, Zhang Z, Kousky VE et al (2008) Simulations and seasonal prediction of the Asian summer monsoon in the NCEP climate forecast system. J Clim 21:3755–3775 Yulaeva E, Wallace JM (1994) The signature of ENSO in global temperature and precipitation fields derived from the microwave sounding unit. J Clim 7:1719–1736 Zhu J, Shukla J (2013) The role of air–sea coupling in seasonal prediction of Asia-Pacific summer monsoon rainfall. J Clim 26:5689–5697

13

R. P. Shukla, B. Huang Zuo Z, Yang S, Hu Z-Z, Zhang R, Wang W, Huang B, Wang F (2013) Predictable patterns and predictive skills of monsoon precipitation in Northern Hemisphere summer in NCEP CFSv2 reforecasts. Clim Dyn 40:3071–3088. doi:10.1007/s00382-013-1772-2