Contrasting Cloud Radiative Feedbacks during Warm Pool and Cold Tongue

The study presents the contrasting characteristics of cloudradiative feedbacks to the cold tongue (CT) and warm pool (WP) El Niño (EN). The maximum sea surface temperature anomalies (SSTA) of the CT-EN are located in the far-eastern Pacific. However, the maximum responses of the shortwaveand longwavecloud-radiative forcing (SWCRF and LWCRF) to the CT-EN warming are centered near the dateline, showing 70° westward shift relative to the maximum SSTA center of CT-WN. In contrast, the maximum responses of the SWCRF and LWCRF to the WP-EN warming show only slight westward shift relative to the maximum SSTA center. The contrasting cloud-radiative feedbacks to the two types of ENs can be traced back to the contrasting precipitation feedbacks, which is associated with the convection threshold. When the warm SSTA of CT-EN occurs in the relatively cold eastern Pacific, the total SST in-situ may not exceed the convection threshold. Therefore, the induced precipitation anomaly would occur towards the warm western Pacific, and the corresponding cloud cover and cloud-radiative feedbacks would exhibit an apparent westward shift. As the warm SSTA of WP-EN occurs in the relatively warm central Pacific, the corresponding responses of the anomalous fields to the WP-EN show only slight westward displacement. (Citation: Chen, L., L. Wang, T. Li, and D.-Z. Sun, 2018: Contrasting cloud radiative feedbacks during warm pool and cold tongue El Niños. SOLA, 14, 126−131, doi:10.2151/sola.2018-022.)


Introduction
Cloud is the prominent modulator in the climate system (Hartmann and Short 1980; Kiehl and Trenberth 1997;Houghton et al. 2001). Through the radiative effects, the cloud could regulate the global energy distribution and thereby alter the atmospheric circulation, which in turn modulates their effects (Stephens 2005). Thus it is critical to understand the cloud-radiative feedback in the climate system Ramanthan et al. 1989;Sun and Held 1996).
A number of previous studies have devoted great efforts into investigating the cloud-radiative feedbacks in climate system (Ramanthan et al. 1989;Cess et al. 1990;Sun and Held 1996;Soden 1997;Schneider et al. 1999;Soden and Held 2006;Sun et al. 2003Sun et al. , 2006Sun et al. , 2009Zhu et al. 2007;Zhang and Sun 2008;Wu et al. 2011;Zheng et al. 2014;Li et al. 2014Li et al. , 2015. As El Niño-Southern Oscillation (ENSO) is the greatest natural climate variability on interannual timescale and exerts profound impacts on the global climate and weather (Philander 1990; Chen 2016, 2017a, b;Yang et al. 2018), a frequently used method to quantify the cloud-radiative feedbacks is to calculate the response of cloud-radiative forcing to the sea surface temperature (SST) changes on the time scale of ENSO (Sun and Held 1996;Soden 1997;Sun et al. 2003Sun et al. , 2006Sun et al. , 2009Chen et al. 2013Chen et al. , 2015Chen et al. , 2017Hu et al. 2017). For example, through calculating the response of cloud-radiative forcing to El Niño (hereinafter EN) warming, some previous studies evaluated the cloud-radiative feedbacks in the ENSO cycle in coupled model simulations (e.g., Sun and Held 1996;Soden 1997;Sun et al. 2006;Sun et al. 2009;Chen et al. 2013Chen et al. , 2016a. More recently, plenty of studies pointed out that there is a new type of EN which tends to occur more frequently since the late 1990s (Chung and Li 2013;Xiang et al. 2013;Zhou et al. 2014). In contrast to the canonical EN featured by the prominent warming in the eastern equatorial Pacific and the coast of South America, the warming center of the new type EN occurs in the central Pacific (e.g., Ashok et al. 2007;Kao and Yu 2009;Kug et al. 2009). Such a new type of EN has various terminologies, such as dateline El Niño, El Niño Modoki, CP El Niño, and warm pool El Niño. In the following discussion, we use the warm pool El Niño (hereinafter WP-EN) to term the new type of EN and the cold tongue El Niño (hereinafter CT-EN) to term the canonical EN. The distinctive EN warming patterns may induce different teleconnections (e.g., Ashok et al. 2007;Weng et al. 2007;Feng et al. 2016aFeng et al. , 2016bFeng et al. , 2017Xu et al. 2017aXu et al. , 2017bXu et al. , 2018. For instance, Hu et al. (2016) pointed out contrasting contributions from the thermodynamic and dynamical processes to the SSTA associated with the two types of ENs, through performing a feedback attribution analysis. Hence, it is conceivable that the responses of cloud-radiative forcing to the two types of EN may differ from each other. However, it is not clear about the specific characteristics of the response of cloud-radiative effects to the new type of EN and whether it is different from that to conventional CT-EN.
To shed some light on the characteristics of cloud-radiative feedbacks to the new type of EN, we may use the aforementioned method that was previously employed to detect the cloud-radiative feedbacks in ENSO cycle. With the advent of satellite data, it is possible to quantify the cloud radiation forcing effects at the top of atmosphere (TOA) more accurately than the results derived from the surface heat flux reanalysis. Through classifying the ENSO into two specific types, the current study aims to identify the primary characteristics of the cloud-radiative feedbacks to the two types of EN.
The rest of this paper is organized as follows. The datasets and methods are described in Section 2. The contrasting characteristics of the cloud-radiative feedbacks to the two types of EN are presented in Section 3. The possible explanation of the contrasting characteristics of the feedbacks to the two types of EN are not sensitive to the selection method that is used to classify the two types of EN. The Niño3 index and EMI index have also been used to calculate the corresponding results of the feedbacks to the two types of EN (figure not shown), which resemble that shown as below.

Contrasting characteristics of cloud-radiative feedbacks to the two types of EN warming
The shortwave cloud radiative forcing (SWCRF) at TOA is defined as below: where SW and SW c represent the TOA net downward solar radiation for the average cloudy conditions and clear sky, respectively. Following previous studies (Chen et al. 2013;Li et al. 2014Li et al. , 2015, we obtained the SWCRF feedbacks to the two types of EN warming through regressing the anomalous SWCRF onto the CT-EN index and WP-EN index, respectively. Figure 2 presents the basin-wide view of the response of SWCRF (hereafter SWCRF feedback) to the two types of ENs. Pronounced negative SWCRF anomalies are noted over the central and eastern equatorial Pacific, in response to the CT-EN warming (Fig. 2a). The SWCRF feedback to the CT-EN is characterized by an anomalous negative center of SWCRF to the east of the dateline (around 170°W), while the negative SWCRF feedback in the far-eastern Pacific is minor. Notably, the maximum center of SWCRF response exhibits a 70° longitude shift compared to the corresponding maximum SSTA center (located near 100°W, as shown in Fig. 1a). As shown in Fig. 2b, significantly negative SWCRF anomalies are found in the central equatorial Pacific, in response to the WP-EN warming. Furthermore, the maximum center SWCRF response to the WP-EN is located to the west of the dateline (around 170°E), showing only 20° displacement compared to the corresponding maximum SSTA center of WP-EN. Additionally, the maximum center of SWCRF response to CT-EN and WP-EN are, respectively, around −20 W m −2 K −1 and −36 W m −2 K −1 . This indicates a weaker SWCRF feedback over the central equatorial Pacific to CT-EN than that to WP-EN.
The longwave cloud radiative forcing (LWCRF) is defined as below: cloud-radiative feedbacks is provided in Section 4. Finally, a summary is shown in Section 5.

Data and methods
The monthly SST used in this study is from the HadISST dataset (Rayner et al. 2003 Dee et al. 2011). In this study, the anomalies indicate the interannual anomaly fields which were obtained through removing their corresponding climatological seasonal cycle.
To estimate the responses of cloud-radiative forcing to the two types of EN regimes, we will apply the method used in previous studies (e.g., Sun et al. 2003Sun et al. , 2006Sun et al. , 2009Chen et al. 2013): quantifying the feedbacks through regressing the corresponding anomaly fields to the SSTA. In these previous studies, the SSTA averaged over a wide region (5°S−5°N, 150°E−110°W) was used to do the regression analysis. To identify the contrasting characteristics of the cloud-radiative feedbacks to the two types of EN, in this study the corresponding anomaly fields are, respectively, regressed to the CT-EN index and WP-EN index. Based on the definition proposed by previous study (see the equation 1 in Ren and Jin 2011), we obtained the CT-EN index and WP-EN index. Figures 1a and 1b show the spatial warming patterns of CT-EN and WP-EN, which are obtained through regressing the SSTA onto the CT-EN index and WP-EN index. Clearly, the CT-EN and WP-EN shows different spatial pattern in terms of locations of SSTA warming: the maximum center of SSTA warming of the former is located near 100°W, and the latter near 170°W. Such CT-and WP-warming patterns resemble that revealed by some other studies (e.g., Ashok et al. 2007;Kao and Yu 2009;Kug et al. 2009). The specific time series of CT-EN index and WP-EN index are shown in Figs. 1c and 1d. The correlation between the Niño3 index and CT-EN index is 0.98, and the correlation between the WP-EN index and the El Niño Modoki index (EMI, which is proposed by Ashok et al. 2007) is 0.75. Note that the primary where F and F c indicate the outgoing longwave radiation for the average cloudy condition and clear sky, respectively. Figure 3 provides the horizontal pattern of the response of LWCRF (hereafter LWCRF feedback) to the CT-EN and WP-EN. It is noted that positive LWCRF anomalies appear over the central and eastern equatorial Pacific (Fig. 3a), in response to the CT-EN warming. This LWCRF feedback to the CT-EN is characterized by an anomalous positive center of LWCRF to the east of the dateline (around 170°W), showing a 70° longitude shift compared to the corresponding maximum SSTA center (100°W) of CT-EN. In contrast, the maximum center of the response of LWCRF to WP-EN is located near 170°E (Fig. 3b), which shows slightly westward shift (around 20° longitude) relative to the maximum SSTA center of WP-EN. Additionally, the maximum center of the response of the LWCRF to the CT-EN and WP-EN are, respectively, 16 W m −2 K −1 and 30 W m −2 K −1 , showing a weaker LWCRF feedback over the central equatorial Pacific to CT-EN than that to WP-EN. The different displacement configuration and the contrasting magnitude of the cloud-radiative feedbacks between the two types of ENs, may have an impact on the EN development and the ultimate EN intensity (Xiang et al. 2012;Su et al. 2014;Chen et al. 2016b, c;Chen and Li 2017). Previous studies suggested that the SWCRF feedback, a dominant thermodynamic damping feedback during ENSO cycle, plays an essential role in suppressing the ENSO development (e.g., Li et al. 2014Li et al. , 2015. As demonstrated above, the SWCRF feedback to the two types of ENs show different extent of westward displacement relative to the corresponding maximum SSTA center. For the WP-EN, as the majority of the anomalous negative SWCRF is located in the central equatorial Pacific, it acts to directly damp the development of WP-EN SSTA signals in the central equatorial Pacific. But for the CT-EN, as the anomalous negative SWCRF occurs in the central equatorial Pacific and the majority of SSTA signals occur in the eastern Pacific, the anomalous negative SWCRF would not damp the development of CT-EN SSTA as effectively as that for the WP-EN. Moreover, the negative SWCRF feedback itself is weaker for the CT-EN than that for the WP-EN. This may play a partial role in leading to the difference in the EN intensity between the CT-EN and WP-EN (in observations, the CT-EN is usually stronger than the WP-EN, as analyzed by Zheng et al. 2014;Fang et al. 2015).

Possible explanation of the contrasting cloudradiative feedbacks
To understand the contrasting characteristics of the cloudradiative feedbacks to two types of ENs, we first examined the total cloud cover (TCC) feedback. As shown in Fig. 4, the spatial patterns of the response of TCC anomaly to the two types of EN warming mirror those of both the SWCRF and the LWCRF feedback patterns. Specifically, the maximum center of TCC feedback to CT-EN is located to the east of the dateline (near 170°W), showing about 70° westward shift relative to the maximum SSTA center of CT-EN. In contrast, the maximum center of TCC feedback to WP-EN is located near 170°E, showing a slightly westward shift (around 20° longitude) relative to the maximum SSTA center of WP-EN. Also, the maximum center of the TCC feedback to the CT-EN is weaker than that to the WP-EN. As the cloud could increase the planet albedo compared to cloudless sky, the contrasting TCC feedbacks to the two types of EN generally determine the different distribution and strength of the SWCRF feedbacks to the two types of EN. As the cloud could decrease the outgoing longwave radiation (i.e., heat preserving effect), the contrasting TCC feedbacks also cause the contrasting pattern and strength of the LWCRF feedbacks.
It is well known that the TCC feedback is closely linked to the precipitation feedback during ENSO cycle (e.g., Sun et al. 2009;Li et al. 2014). As shown in Fig. 5, the horizontal pattern of the response of precipitation (hereafter precipitation feedback) to CT-EN resembles the corresponding TCC feedback to CT-EN. So does the precipitation feedback to WP-EN. This indicates that the contrasting characteristic of the cloud-radiative and TCC feedbacks can be traced back to the different precipitation feedbacks. We further examined the response of the atmospheric circulation to the two types of EN. Figure 6 shows the spatial pattern of  Fig. 1a. (b) same as (a), but for the response of anomalous SWCRF to the warm pool El Niño warming that was obtained through regressing the SWCRF anomaly onto the WP-EN index. The stippling indicates the regression coefficient exceeding a confidence level of 99% using Student's t test. the response of vertical motion at 500 hPa to the two types of EN warming. As shown in Fig. 6, the vertical motion anomalies show spatial patterns that are dynamically coherent with the precipitation feedbacks. In response to the CT-EN, anomalous ascending motion occurs in the central-eastern equatorial Pacific with a maximum center near 170°W (Fig. 6a), which exhibits an apparent westward shift relative to the maximum SSTA center of CT-EN. For the response to the WP-EN, the maximum center of the anomalous ascending motion is located near 170°E (Fig. 6b), which exhibits a slightly westward shift relative to its maximum SST center. Thus, the contrasting responses of atmospheric circulation to the two types of ENs match well with the contrasting precipitation feedbacks.
Regarding the contrasting characteristics of the precipitation feedbacks to the two types of EN, this may be closely associated with the tropical convection threshold. As presented in Figs. 5a and 5b, the contours indicate the total SST during the CT-EN warming and WP-EN warming. Clearly, the response of precipitation anomaly to either CT-EN or WP-EN generally occurs within the area where the total SST exceeds 28°C (Fig. 5), rather than the area where the SSTA dominates as shown in Figs. 1a and 1b. Additionally, both the maximum convection centers in response to the two types of ENs roughly coincide with the area exceeding 30°C. The well matching between the 28 isotherm and the anomalous precipitation response pattern indicates that the contrasting characteristics of the precipitation feedbacks, and the resultant TCC feedbacks and the cloud-radiative feedbacks may root in the deep convection threshold. As we know, the deep convection generally occurs when the total SST exceeds 28°C. Since the warming SSTA of the CT-EN occurs in the eastern equatorial Pacific, overlaying relatively cold climatological SST, the total SST (i.e., SSTA plus mean SST) in the eastern Pacific may not exceed the convection threshold. Thus, the occurrence of the convection would ultimately shift toward the western equatorial Pacific region, where the mean SST is warmer than that in far-eastern Pacific. Consistent with the response of precipitation to the CT-EN, the corresponding responses of the atmospheric circulation, cloud cover and cloud-radiative effects also exhibit an evident westward shift relative to the maximum center of SSTA. In contrast, the SSTA warming of WP-EN occurs in the central equatorial Pacific, overlaying relatively warm climatological SST, hence, the corresponding responses of the anomalous fields exhibit only a slightly westward shift relative to the WP-EN warming.
The difference in the magnitudes of the cloud-radiative feedback is associated with the contrasting displacement configuration between the two types of ENs. For the WP-EN, the precipitation   Fig. 2, but for the response of precipitation (unit: mm day −1 K −1 ). The contours indicate the corresponding total SST (the values are 28, 29, and 30°C) during CT-EN (Fig. 5a) and WP-EN (Fig. 5b). Here the total SST in Fig. 5a was estimated based on the following formula: SST_total CT = SST + SSTA CT * Scale_ coeff CT , where SST_total CT denotes the estimated total SST during CT-EN, SST denotes the climatological mean SST, SSTA CT denotes the spatial pattern of SSTA as shown in Fig. 1a, and the Scale_coeff CT denotes the scale up coefficient, i.e., the average of the specific CT-EN index that exceeds 1.5 standard deviation of CT-EN index as shown in Fig. 1c. Similarly, the estimated SST_total WP was obtained. response to WP-EN warming exhibits only slight displacement relative to the corresponding maximum SSTA. In contrast, for the CT-EN, the maximum precipitation response occurs near the dateline (around 170°E−150°W on the equator), but the warm SSTA signals are relatively weak near the dateline, as the region is outside the core region of CT-EN warming. Thus, the magnitudes of the precipitation feedback and the corresponding cloud-radiative feedbacks to CT-EN warming are weaker than the counterparts to WP-EN warming.

Summary and discussion
The study presents the contrasting characteristics of cloudradiative feedbacks to the cold tongue and warm pool ENs. The warm SSTA of the conventional CT-EN is confined to the fareastern Pacific, but the maximum center of the cloud-radiative feedbacks (both SWCRF and LWCRF feedbacks) to the CT-EN is located near the dateline, showing about 70° westward shift relative to its maximum SSTA center. In contrast, the maximum center of the cloud-radiative feedbacks to the WP-EN warming shows only a slightly westward displacement. Additionally, the magnitude of the SWCRF and LWCRF feedbacks to the CT-EN are weaker than that to WP-EN.
The contrasting characteristics (e.g., location and strength) of the cloud-radiative feedbacks to the two types of EN are linked to the contrasting TCC and precipitation feedbacks. Matching well with the contrasting cloud-radiative feedbacks, the TCC feedbacks also exhibit the contrasting characteristics between the two types of ENs. The different extent of the westward displacement of the TCC feedbacks relative to the corresponding maximum SSTA center are responsible for the contrasting cloud-radiative feedbacks, as the cloud is the primary contributor to the planetary albedo and has the heat preserving effect through reducing the outgoing longwave radiation. Furthermore, the contrasting TCC feedbacks can be traced back to the contrasting precipitation feedbacks. The maximum response centers of the precipitation to the two types of ENs exhibit different extent of westward displacement relative to the corresponding maximum SSTA centers. Such contrasting precipitation feedbacks are responsible for the contrasting characteristics of the cloud cover and the cloud-radiative feedbacks between the two types of ENs.
The different extent of the westward displacement of the precipitation feedbacks relative to the corresponding maximum SSTA center are associated with the tropical deep convection. The spatial patterns of the anomalous precipitation response to the two types of ENs generally match well with the 28 isotherm. When the warming SSTA of the CT-EN occurs in the eastern equatorial Pacific where the climatological SST is cold, the total SST in eastern Pacific may not exceed the convection threshold. The occurrence of the convection would ultimately shift toward the western equatorial Pacific where the climatological SST is warmer than eastern Pacific. Therefore, the response of precipitation and the corresponding responses of the large-scale atmosphere circulation, cloud cover and cloud-radiative effects show an evident westward shift relative to the maximum center of SSTA, and exhibit relatively weak intensity. In contrast, when the SSTA warming of WP-EN occurs in the central equatorial Pacific where climatological SST is relatively warm, the corresponding responses of the anomaly fields show only slightly westward shift relative to the WP-EN warming, and exhibit a relatively strong intensity.
Understanding the different ENSO responses to the two types of ENs are of importance, as the contrasting characteristics of the cloud-radiative feedbacks could make a contribution to the contrasting properties between CT-EN and WP-EN. For example, the different displacement configuration and the contrasting magnitude of the negative SWCRF feedback may play a partial role in causing the relatively strong CT-EN but relatively weak WP-EN, as also discussed by Zheng et al. (2014) and Fang et al. (2015). Hu et al. (2016) suggested that the strong water-vapor feedback over the eastern Pacific and the strong negative feedbacks over the central equatorial Pacific can help relocate the maximum warming center from the central Pacific to the eastern basin for the CT-EN. The present finding of the spatial shift of the negative SWCRF feedback in response to the CT-EN suggests that the understanding about the role of SWCRF feedback in the life cycle of El Niño only applies well to the WP-EN. It appears that the SWCRF feedback amplifies the canonical CT-EN in a relative sense because it damps the west side of CT-EN warming while it has relatively little effect on the far east side of the warming.