1. Introduction
El Niño–Southern Oscillation (ENSO; see the appendix for full list of acronyms) has far-reaching systematic influence on the global atmospheric circulation (Oort and Yienger 1996; Alexander et al. 2002), energetics (Mayer and Haimberger 2012; Mayer et al. 2013; Trenberth and Fasullo 2013), and regional hydrological variability (Ropelewski and Halpert 1987; Dai and Wigley 2000; Trenberth et al. 2002; Gu et al. 2007; Miralles et al. 2014). The remote response to ocean temperatures related to ENSO has been well documented in observations, as well as in coupled and forced atmospheric models. In a multimodel sensitivity experiment, Schubert et al. (2016) summarize regional land precipitation responses around the Earth to global SST patterns, noting the linkages of many regions to ENSO. Pathways by which tropical oceans create remote responses are fundamentally dynamic in nature (Walker 1924; Hoskins and Karoly 1981; Alexander et al. 2002; Trenberth et al. 2002; Alexander et al. 2009). Relative to El Niño local SST warming and tropical convection, remote oceanic regions of descending motion show reductions in clouds and increasing surface radiative forcing, predominantly from the shortwave (Stephens et al. 2018).
In the case of El Niño, warming in the eastern and central equatorial Pacific precedes broader remote ocean warming (Klein et al. 1999; Yu and Rienecker 1999; Alexander et al. 2002; Mayer et al. 2013). Tropospheric temperatures over the global tropics rise in response to warm SST events within a period of a couple months; deep convective mixing of moist entropy throughout the troposphere provides a heat source that is dispersed longitudinally by inertia–gravity wave propagation (Charney 1963; Yulaeva and Wallace 1994; Sobel et al. 2002). Widely dispersed tropospheric warmth spreads over tropical land (Chiang and Lintner 2005). Reductions in moisture transport to land and associated reductions in precipitation and evapotranspiration predominantly result from these diabatically driven circulation anomalies (Gu et al. 2007; Miralles et al. 2014). With the complexity of the response and diversity of any given El Niño event, there are many outstanding questions regarding the El Niño remote response and the redistribution of heat. Here we are focused on one aspect of SST-driven climate effects—El Niño-related changes in the land surface energy balance within the tropical belt, which we take as within 30° latitude of the equator.
A notable aspect of these adjustments between the land and ocean domains is that land surface temperatures tend to be amplified over those of the global ocean (Chiang and Lintner 2005; Dommenget 2009; Tyrrell et al. 2015)—this amplification factor being typically about 1.2–1.5 over that of SSTs is close to upper tropospheric warmth dynamically spread over the tropics in response to El Niño events. Modeling studies by Tyrrell et al. (2015) suggest that amplified land warming associated with shorter-term interannual climate variability appears to behave in similar fashion to that occurring in model simulations of greenhouse gas–induced climate change (Manabe et al. 1991; Lambert and Chiang 2007; Sutton et al. 2007; Joshi et al. 2008; Fasullo 2010; Lambert et al. 2011). Sutton et al. (2007) employed the IPCC AR4 modeling suite to study relaxation to equilibrium after changes in doubled CO2 forcing experiments, emphasizing the important role that surface moisture availability plays in adjustments. They particularly noted the opposite change in Bowen ratio response over the land versus ocean. Increased sensible heat flux loss over land exceeded that of latent heat flux (0.86 and 0.83 W m−2, respectively). This result lends support to the analysis of Chiang and Lintner (2005), who concluded from mechanistic modeling studies of the 1997/98 El Niño that sensible heat flux regulates the warming over land. On the other hand, Compo and Sardeshmukh (2009) argue that observed multidecadal temperature over land can be replicated by AMIP experiments forced by observed SSTs and that enhanced downward longwave radiation from a warmer, moister troposphere is a primary surface warming agent.
In further modeling studies using specified SST and slab ocean models, Tyrrell et al. (2015) also place emphasis on the delayed tropospheric warming accompanying remote ocean SST changes that act to require adjustments in the land surface energy balance. Their results indicate a prominent role for downward longwave flux from a moistened lower troposphere, a result consistent with that of Compo and Sardeshmukh (2009). However, they also noted more cloudiness and a reduced shortwave flux into the land surface, a result at odds with the previously noted studies. Clearly the role of different mechanisms controlling the tropical land surface energy balance changes needs further examination.
While surface observations confirm the relationship of ocean and land surface temperatures and precipitation, many of the studies relating tropical land warming and drying to the ocean temperatures are conducted with atmospheric or coupled numerical model ensemble simulations. Here, we will use the Modern-Era Retrospective Analysis for Research and Applications, version 2 (MERRA-2), reanalysis to provide additional information, surface fluxes in particular, that are not easily observed in order to evaluate the atmospheric circulation and land response over tropical continents to El Niño; we also use a 10-member AMIP ensemble with the MERRA-2 model to distinguish the model-simulated response from the analyzed atmospheric response in the reanalysis. Reanalysis quality broadly depends on the availability of observations assimilated into the system (e.g., Bosilovich et al. 2011, 2015), which may be challenging for tropical land areas where in situ observations are fewer than in the northern midlatitudes. This necessitates an evaluation of monthly mean SST, T2m, and precipitation anomaly behavior in comparison with available observations. We then analyze a composite of El Niño events, examining the evolution of radiative and turbulent energy fluxes and related quantities. This is done for tropical land mean values with additional diagnostics to highlight regional contributions to the global tropical signals. We then use pentad MERRA-2 data to examine more closely the lag relationships to SST warming. Last, we examine the 2015/16 El Niño, withheld from the composite analysis, to put this extreme event in context with composite results.
2. Data
a. MERRA-2
The MERRA-2 reanalysis (Gelaro et al. 2017) is a state-of-the-art global data product that includes numerous diagnostics (e.g., vertically integrated transport, cloud-free radiation, and surface fluxes) that can be used to investigate El Niño heating response, especially in the energy and water budgets. A preponderance of atmospheric satellite and in situ observations are assimilated in MERRA-2 (McCarty et al. 2016). Numerous model physics advancements have been included (Molod et al. 2015). The assimilation of atmospheric mass (including water vapor) is constrained to balance at the global scale (Takacs et al. 2016), which provides continuity of the mass through data assimilation. The climatology of the MERRA-2 data is discussed by Bosilovich et al. (2015) while Collow et al. (2017) provide some discussions on ENSO and SST forcing.
In the MERRA-2 reanalysis, observation-corrected precipitation is input to force the land surface parameterization, rather than the model simulated precipitation [Reichle et al. (2017) provide details on the observation-corrected precipitation]. The benefit here is that surface heat fluxes, temperature, and soil moisture are more closely constrained by the observed precipitation. However, the atmospheric model’s precipitation can be different and related to physical processes (e.g., convergence). This is an important consideration in evaluating the land coupling to the atmosphere. Substantial biases exist in the MERRA-2 model-generated tropical precipitation (Bosilovich et al. 2017). Such biases are common in reanalyses (Trenberth et al. 2011; Bosilovich et al. 2008). We will identify whether observation-corrected precipitation or model-generated precipitation is used in subsequent analysis of the data, as the choice of which depends on if the forcing on the land surface or the atmospheric condensation is more relevant to the discussion.
Lim et al. (2017) used MERRA-2 and ocean reanalyses to evaluate the tropical Pacific Ocean evolution of the strongest El Niño case studies in the ocean and atmosphere. The MERRA-2 representation of El Niño events was validated against cloud and precipitation observations. Each event is unique, although the evolutions of the SST, clouds, convection, and precipitation in the tropical Pacific are similarly connected. Even with a thorough dissection of each event in the atmosphere and the ocean, the authors note that there is still more to be done to complete the understanding of the events.
b. M2AMIP
To help to evaluate the model processes response to SST, as it differs from the assimilated state fields, a numerical model integration was carried out, following the MERRA-2 atmospheric model setup exactly (Molod et al. 2015), including its climate forcing data (Collow et al. 2017). Ten simulation members were made to allow for an ensemble of global simulations, using the observed SST, aerosol, and solar forcing that were used in MERRA-2. The data formats and output variables are identical to MERRA-2. Collow et al. (2017) evaluate the M2AMIP ENSO connection to the continental United States regional climate. Here we focus on tropical land El Niño response, incorporating the ensemble mean of the experiment into the comparison to consider the role that the prescribed SSTs has in governing the background model used in MERRA-2.
c. ERA5
The latest reanalysis from ECMWF, ERA5, includes physics and analysis updates (Hersbach et al. 2018) and will provide a point of comparison for the MERRA-2 evaluation. All ECMWF reanalysis products have characteristically produced superior near-surface air temperature owing to the analysis of surface meteorology stations (Simmons et al. 2017), which is continued in ERA5 at 1-hourly frequency (Hersbach et al. 2018). In addition to the significant advancement in data assimilation, ERA5 also includes model improvements in radiation, land, and convection processes.
d. Observations
One purpose of this study is to evaluate the MERRA-2 reanalysis and M2AMIP model capability to replicate El Niño and to use their unique data to better understand the subsequent warming and drying of the tropical land areas. Several key quantities in this process have very good quality observational climate data records to compare with the model-generated quantities. The MERRA-2 land surface air temperature is influenced by the observational data assimilation and the input observation-corrected precipitation forcing, although the model land parameterizations play a role.
For surface air temperature, CRU time series version 4.01 (Harris and Jones 2017; Harris et al. 2014) provides land station dataset gridded to ½° spatial resolution with monthly means up to near current times, using quality control algorithms that have developed over decades. In precipitation, GPCP covers the globe and encompasses the modern satellite era by merging satellite and gauge observations to provide a continuous and consistent record of precipitation. While the resolution is somewhat coarser than current reanalysis systems, the high quality and global coverage are useful characteristics in evaluating large-scale modeling processes. The most recent version (2.3) includes several quality advancements over previous processing of the observations (Adler et al. 2017).
Because we are interested in the warming of the land surface during the life cycle of El Niño, the clouds over the land surface would likewise be closely related to the land surface response. Cloud observations and their comparison to model quantities are often uncertain owing to fundamental differences in the measured and modeled structural formulations. Using ISCCP HX satellite radiance, cloud, and surface information (Rossow and Schiffer 1999; Young et al. 2018), SRB provides the surface radiation and clear-sky estimates to determine the cloud radiative effect at a resolution of 1° (Stackhouse et al. 2011; Zhang et al. 2015, 2019). CERES Surface-EBAF (Edition 4; Kato et al. 2018) also provides estimates of the radiative effect using MODIS and CERES observations. In this way, the surface radiative effect can serve as a method to determine the influence of clouds on the land surface. Here, SRB (Release 4.0, Integrated Product) provides surface shortwave radiative flux estimates for 1983–2015, and CERES EBAF (Edition 4.0) is available from 2000 to the present (CERES Science Team 2018).
3. Time series
While steps were taken in the development of the MERRA-2 model and data assimilation to mitigate excessive unrealistic impact of water vapor assimilation and variations in the water vapor observing system, there were notable instances of changes apparent in the time-varying global water vapor (Bosilovich et al. 2017; Robertson et al. 2016). For example, test experiments without AIRS radiance assimilation demonstrated that MERRA-2’s already high precipitation over global land surfaces increases even more as AIRS is included. Tropical continental (30°S–30°N, land only) precipitation produced by MERRA-2 (Fig. 1a) does not increase sharply after AIRS introduction (November 2002) but does suggest larger interannual variability. ERA5 tends to have higher correlation to GPCP than MERRA-2, and a smaller bias. After 1998, MERRA-2 and ERA5 track closely to the GPCP anomalies. Interestingly, the M2AMIP mean precipitation high bias is similar to that of MERRA-2, but the interannual variability amplitude more closely follows GPCP, and is as well correlated as ERA5.
Comparison of anomalies of tropical (a) precipitation, (b) surface shortwave cloud radiative effect, and (c) near-surface air temperature over land (30°S–30°N) regions for MERRA-2, M2AMIP, and ERA5 data with observations from CERES/SRB, GPCP, and CRU, respectively. The anomalies remove the common mean annual cycle period (2000–14) and have a 12-month running mean applied. The time averaged mean of the data is reported in the legend. Niño-3.4 SST anomalies (K) are gray shaded, with the scale on the right axis (anomalies are 12-month running means). Mean values and correlation coefficients to the reference observations are shown in the legend.
Citation: Journal of Climate 33, 3; 10.1175/JCLI-D-19-0231.1
Shortwave cloud radiative effect at the ground surface (SWgCRE) is computed as the all-sky net shortwave radiation minus the clear-sky net shortwave radiation, so that positive anomalies indicate reduced cloudiness and more warming of the land surface. The reanalyses and model means SWgCRE anomalies in Fig. 1b are biased strongly relative to SRB values toward more clouds with ERA5 values showing more clouds than both observation datasets for tropical land. While the MERRA-2 mean SWgCRE is between the means of the observations (included in the legend of Fig. 1), the interannual variations after 2002 may be related to variations in observations (e.g., AIRS and SSM/I; see McCarty et al. 2016). A general decreasing trend (after 2000) is somewhat captured by MERRA-2 and ERA5, but not M2AMIP (or in the CERES EBAF record). The reduction of surface shortwave radiation attenuated by cloud effects during El Niño events is apparent, but also varies between the observations and numerical systems.
The time series of tropical continental near-surface air temperature anomalies (Fig. 1c) demonstrates two dominant signals: an apparent global warming trend and ENSO. In subsequent analysis, we will focus on the ENSO signal, as the global warming signal should be addressed with longer time series of observations and reanalyses (e.g., Compo and Sardeshmukh 2009). Both MERRA-2 and M2AMIP represent a reasonable surface temperature response to ENSO over the tropical lands, despite some interannual variations being stronger than observed. ERA5 variations track the CRU observations well, owing to the analysis of surface air temperature and water vapor observations (Hersbach et al. 2018).
From here forward, we will emphasize the processes in MERRA-2 and M2AMIP with a focus on the continental global monsoon regions in the tropical (30°S–30°N) latitude band. The global monsoon regions are defined as in Wang et al. (2012) (see their Fig. 1; also see Fig. 2 herein). The definition uses the amplitude of the seasonal precipitation variations for monsoon development, and Wang et al. (2012) note that the monsoon precipitation is also affected by ENSO. This partitioning of the land area provides focus on the hydrological response and drought of the land areas during El Niño. Figures 3a–c compare the SST averaged over the Niño-3.4 region (170°–120°W, 5°S–5°N) with the area-averaged tropical land near-surface air temperature in the global monsoon areas. In this figure (and in the remainder of the paper, with any exceptions noted), the time series have been detrended to emphasize ENSO variability and the relationship between the SST and continental near-surface air temperature and other properties.
Global monsoon areas (blue-shaded regions) as defined by MERRA-2 seasonal amplitude of precipitation using the method developed by Wang et al. (2012). The thick continental outlines depict regions that will be evaluated more closely.
Citation: Journal of Climate 33, 3; 10.1175/JCLI-D-19-0231.1
Time series of Niño-3.4 SST and near-surface tropical continental air temperature (T2m) in global monsoon areas for (a) HadISST and CRU observations, (b) MERRA-2, and (c) M2AMIP sea surface temperature (Ts). MERRA-2 and M2AMIP SSTs are from the same source, which is documented by Bosilovich et al. (2015). The time series data have been deseasonalized, detrended, and then filtered with a 12-month running mean. (d) The lead/lag correlations of these time series, computed from the monthly mean time series. Units are kelvins, except for correlations, which are dimensionless.
Citation: Journal of Climate 33, 3; 10.1175/JCLI-D-19-0231.1
Comparing HadISST (Rayner et al. 2003) with CRU tropical continental air temperature observations, the land response to El Niño warming varies in magnitude from event to event. This is to be expected, as essentially each El Niño is different, despite attempts to classify them by similar characteristics (e.g., Trenberth and Smith 2006; Pillai et al. 2017). However, major El Niño events generally lead to tropical continental warming within a few months. Computing lead and lag correlations from monthly means (Fig. 3d), the CRU air temperatures reach a peak warming 3 months after the peak SST warming, although the lag correlation value at 4 months after is very close to that peak. MERRA-2 and M2AMIP (Figs. 3b,c) tend to follow the same relationship, however; both systems have generally higher continental air temperature correlations to the Niño-3.4 temperature anomalies (Fig. 3d). The shape of the lead/lag curve (Fig. 3d) is much the same in each system, and the significant positive correlations demonstrate the systematic El Niño SST controls on land surface temperatures. While acknowledging that so-called central Pacific or Modoki El Niño events may originate from somewhat different processes than their eastern Pacific counterparts (e.g., Kao and Yu 2009; Hu et al. 2016), a compositing approach provides a compact way of examining first-order characteristics and evolution of these events.
4. Composite El Niño
To composite El Niño events, we first determine a central time, when the event is at its peak (will also be referred to as M = 0). Using the monthly mean MERRA-2 SSTs, area averaged for the Niño-3.4 region (170°–120°W, 5°S–5°N), we remove the mean annual cycle (base period 1981–2010) and then average for 3-month seasonal anomalies. The peak seasonal anomaly for each event (if > 1 K) is used in the composite El Niño. Peak dates and anomalies are listed in Table 1. The composite mean is developed for 25 months beginning at the month M = −12 relative to the M = 0 month, ending at M = +12 months. The DJF 2015/16 event is excluded in order to compare the composite to that event (section 5b). Note that by compositing only warm events, averaging the 25-month anomalies does not yield zero-mean in the composite figures. This compositing of 2-yr periods also helps to mitigate the influence of discontinuities in the data. Statistical significance and composite standard deviation are presented in the composite time series. These statistics help to focus the interpretation of the results.
Coordinates and magnitude of the Niño-3.4 temperature anomalies that are the center times included in the composite El Niño.
a. Tropical composite
Niño-3.4 SST and tropical land global monsoon area 2-m air and midtropospheric temperature time series composites are given in Fig. 4. Shading indicates the envelope of one standard deviation of the composite average for Niño-3.4 SST and T2m. Note that MERRA-2 and M2AMIP use the same SST data. The peak land temperature response, as shown above with correlations, occurs in the 3–4-month time frame, after the peak of Niño-3.4 anomalies. Initial land warming slows through the SST peak month before again warming rapidly. MERRA-2 warming of tropical continental surface air temperature exceeds that in CRU observations and is perhaps a little earlier. On the other hand, MERRA-2 midtropospheric temperatures are cooler than ERA5, but the timing of the warming in both composites is similar (Figs. 4a,b). Midtropospheric temperature maxima generally follow close to those of T2m, lagging T2m by one month for MERRA-2. The M2AMIP has a smoother time evolution owing to the ensemble simulation. Its land response is driven by the atmospheric modeling, and maintains a similar evolution of the land temperatures as CRU. For all three cases, the amplitude of midtropospheric warming in the six months preceding the T2m maximum exceeds that of the near surface. This is consistent with behavior reported by Chiang and Lintner (2005), Fasullo (2010), and Tyrrell et al. (2015).
El Niño composite anomalies for Niño-3.4 SST (black) and 2-m air temperature for all tropical lands identified in global monsoon regions (T2m; red). The solid line indicates the composite average, and shading indicates ±1 standard deviation. Units are kelvins, and the scales are color coded to each corresponding line. The dashed red lines indicate 500-hPa temperature anomalies (where ERA5 reanalysis is shown for comparison purposes in the observations panel). M2AMIP uses the same prescribed SST as MERRA-2, and the observations are from HadISST. Dots indicate anomalies significantly different from zero (using a t test at 90% confidence).
Citation: Journal of Climate 33, 3; 10.1175/JCLI-D-19-0231.1
Figure 5 shows the SWgCRE and precipitation composite evolution for the tropical land areas. A double maximum (minimum) is apparent in the SWgCRE (precipitation) time series. The initial reduction in precipitation and associated cloudiness occurs from M = −4 to −6 months. Following this SWgCRE weakens near M = 0 as precipitation deficits also weaken somewhat. A second maximum in SWgCRE and precipitation reduction occurs at M = 3–4 months in the observations and in MERRA-2. For each of these two phases SWgCRE changes coincide closely with the precipitation changes, with a suggestion that SWgCRE leads the precipitation. The entire sequence is somewhat weaker in M2AMIP and the second maximum occurs much earlier (M = 0).
As in Fig. 4, but for absorbed shortwave radiation cloud radiative effect (SWgCRE, made negative so that the lower values correspond to less cloud; black) and precipitation (Prec; red), where units are watts per meter squared for SWgCRE and millimeters per day for precipitation. Note that the MERRA-2 precipitation is derived from the atmospheric model prediction.
Citation: Journal of Climate 33, 3; 10.1175/JCLI-D-19-0231.1
Figures 6 and 7 provide regional structure of temperature and precipitation anomalies at three times in the composite complementing the time series in Fig. 4. At M = −4 as near-equatorial SSTs have already cooled west of the date line and are rapidly warming over the eastern tropical Pacific, near-surface warming is widespread over the Indo-Pacific rim countries and most of Australia. Only weak midlevel warming is present in the southeastern subtropical Pacific, yet other regions of the subtropics at 500 hPa have already started to cool, especially in the Southern Hemisphere where a band of midlevel cold anomalies stretches from subtropical South America southeastward around the globe. MERRA-2 and ERA5 agree on this spatial structure, but it is largely absent in the M2AMIP simulations. Oceanic precipitation has migrated eastward over the tropical Pacific (Fig. 7) and, consistent with SST declines just south of the Maritime Continent, predominantly negative precipitation anomalies have developed over that region and parts of Africa, India, and Australia. These decreases in precipitation over the Indo-Pacific rim land areas before El Niño have been well documented (e.g., Ropelewski and Halpert 1987; Curtis and Adler 2003). While MERRA-2 is representing this feature compared to GPCP, the M2AMIP simulation also develops a strong decrease on precipitation toward the north, into the Maritime Continent. This precipitation decrease in the Maritime Continent is evident in MERRA-2 and GPCP at M = −3 (not shown), indicating that the simulated response in M2AMIP is occurring earlier than observed.
Monthly mean 2-m temperature (K) El Niño composite anomalies for (top) CRU, (middle) MERRA-2, and (bottom) M2AMIP in color shading. Corresponding 500-hPa temperature anomalies are contoured (at 0.5-K intervals), where ERA5 reanalysis 500-hPa height anomalies are included with CRU observations. The columns represent the composite El Niño at (left) 4 months before the peak, (center) the month of the peak, and (right) 4 months after the peak. As in the time series composite, these fields have been deseasonalized and detrended.
Citation: Journal of Climate 33, 3; 10.1175/JCLI-D-19-0231.1
Monthly mean precipitation (mm day−1) El Niño composite anomalies for (top) GPCP, (middle) MERRA-2, and (bottom) M2AMIP. The columns represent the composite El Niño at (left) 4 months before the peak, (center) the month of the peak, and (right) 4 months after the peak. As in the time series composite, these fields have been deseasonalized and detrended. MERRA-2 is its modeled precipitation.
Citation: Journal of Climate 33, 3; 10.1175/JCLI-D-19-0231.1
By M = 0, when Niño-3.4 SST anomalies peak, MERRA-2 continues to agree with CRU showing surface warming in Australia and large parts of Africa. The surface temperature anomaly pattern across Southeast Asia is well reproduced in MERRA-2 but across the interior of South America the warming is too large compared to CRU. In the midtroposphere, a quadrupole temperature pattern is close to its mature state; warm T500 anomalies sit astride the T2m anomalies in the eastern tropical Pacific Ocean while negative anomalies are present to the west over western Australia and southern Asia. This pattern reflects the movement of precipitation and vertical mixing of moist enthalpy and dynamical response to deep convective heating from the Indo-Pacific region eastward toward the date line. Prominent negative T500 anomalies are now at their peak over the Maritime Continent, Australia, and southern Africa (Fig. 6), consistent with reduced precipitation over this region. There is now a resurgence of precipitation over equatorial western Indian Ocean and East Africa with continued suppressed precipitation over most of Australia. This anomaly pattern is likely due not only to delayed Indian Ocean SST forcing by El Niño but also by the remnants of a positive Indian Ocean zonal mode (IOZM) (Saji and Yamagata 2003) SST structure that occurs preferentially during El Niño events (Cai et al. 2011). The 1997 El Niño event was accompanied by a very strong positive IOZM event (Webster et al. 1999) which may heavily weight the composite, but 1987, 1994, and 2006 also experienced significant positive IOZM events. The M2AMIP ensemble replicates MERRA-2 temperature and precipitation anomalies over the land surface except that the magnitudes are generally slightly weaker. In particular, M2AMIP produces too much warming in South America while ENSO is building. African temperature anomalies in M2AMIP are also not as strong as in MERRA-2 and CRU as El Niño builds. After the peak, warm temperatures occur everywhere, except the Sahara, which is observed to be cooling.
At M = 4 months, the mean tropical land has reached maximum warm anomalies (Fig. 4), but at this point not all regions are in phase with the mean. Southeast Asia, Africa, and central South America have all increased surface temperature anomalies. In the midtroposphere, anomalous warmth has been dispersed over the entire tropical band but a dipole maximum of T500 remains in the eastern Pacific with its easternmost extent now over South America. However, Australia land temperatures have decreased substantially (Fig. 6) in the presence of a return to positive precipitation anomalies (the remaining warm anomalies in the extreme north are included in the global monsoon area). The wet regime in areas surrounding the Caribbean has diminished in intensity. MERRA-2 follows the progression of the observations well in most of the continents, including the cooling in Australia. Africa, however, warms more than observations. The M2AMIP handles the warming regions at M = 4 well, but also maintains a warm anomaly through all of Australia, instead of cooling as observed.
These results show that the global mean temperature evolution (Fig. 4) results from combined regional responses that can differ in evolution according to whether they are primarily under the influence of adjacent El Niño SST forcing (South America) or to SST and circulation anomalies that develop in concert with far-field adjustments during El Niño (e.g., Australia, East Africa, and other Indo-Pacific rim regions). MERRA-2 compares reasonably with the observed precipitation anomalies, for land and ocean, with some regional higher intensity owing to its higher resolution (as compared to the GPCP data). The M2AMIP anomalies are smoother than observed (related to the ensemble averaging) and generally comparable, although some regions are not well represented (e.g., Australia after the peak in Niño-3.4).
The spatial patterns in MERRA-2 and M2AMIP are positively correlated to the GPCP observed anomalies (Fig. 8). The observational analysis in MERRA-2 provides substantial improvements in the representation of the precipitation across the composite time periods (compared to the AMIP simulations), including the times when the SST forcing is weak, and also over continental regions. The influence of El Niño SSTs leads to peak positive correlation in M2AMIP larger than most other months, but this benefit is limited to the few months nearest the peak SST anomalies and less prominent over tropical land areas.
Spatial anomaly correlation in the tropics (30°S–30°N) of GPCP with MERRA-2 and M2AMIP precipitation for El Niño composite averages. MERRA-2 and M2AMIP are interpolated to GPCP’s grid to compute the correlation. Solid lines represent all tropics, and dashed lines are land-only correlations. The correlations are dimensionless.
Citation: Journal of Climate 33, 3; 10.1175/JCLI-D-19-0231.1
These initial evaluations show that MERRA-2 global and even most regional tropical land and responses correspond well to observations; M2AMIP comparisons too are reasonable though somewhat less successful in the details of precipitation pattern change.
b. Energy fluxes
The foregoing consistency check between observation, reanalysis, and AMIP temperature and precipitation relationships provides confidence that we can now use these tools to pursue interpreting the role of various fluxes in determining moisture and temperature changes. Figures 9a–d show anomalous downward motion and increased shortwave radiation at the surface, as the precipitation is reduced and temperatures rise, across tropical land areas. Reduced total cloudiness occurs over tropical land extending from M = −8 to M = 8 months, although a substantial recovery toward climatological values is present from M = −3 to M = 0. The surface warming, which is an integrated response to all fluxes, begins in earnest after the first increase in SWgCRE, strongly suggesting the importance of cloudiness reductions in the land surface energy budget during El Niño events. The SWgCRE evolution is also tightly correlated to that of midtropospheric vertical motion (Figs. 9c,d), with downward vertical motion anomalies leading reductions in cloudiness. These vertical motion changes represent dynamical linkages to the SST-induced convection over the tropical Pacific, the Niño-3.4 region, and other regions where the Walker and Hadley circulations are perturbed by anomalous convective heating.
Tropical global monsoon land areas (30°S–30°N) composite El Niño anomaly time series for (left) MERRA-2 and (right) M2AMIP for several key quantities listed on the figure. The solid line indicates the composite average, and shading indicates ±1 standard deviation of the composite mean. Scales are color coded for each line. Dots indicate anomalies that are significantly different from zero (at 90% confidence). The variables are near-surface air temperature (T2m), precipitation, SWgCRE, vertical velocity (−ω500, directed positive for upward motion, at the 500-hPa level), surface temperature (Ts), sensible heat flux (Hs), downward longwave radiation (LW↓), and net surface longwave radiation (LWnet; directed downward is positive).
Citation: Journal of Climate 33, 3; 10.1175/JCLI-D-19-0231.1
Climatologically in MERRA-2 (not shown), there is net energy export from tropical land regions associated with mean moisture convergence but a dominant export of dry static energy (DSE) and kinetic energy (KE). El Niño–related dynamical changes weaken these energy flows yielding reduced flux convergence of moisture, correspondingly reduced divergence of (DSE + KE); little or no net energy export anomaly through the peak of El Niño results (Fig. 10a). These circulation and energy transport anomalies are consistent with tropical land precipitation reductions (Fig. 5), which correlate strongly with midtropospheric vertical velocity ω500 changes. Note that the spread in MERRA-2 transports among composite events reduces the statistical significance. The M2AMIP ensemble shows similar relationships except that the slight recovery of the transports toward climatological values occurs about three months earlier.
As in Fig. 9, but for additional quantities: DSE divergence (∇ ⋅ CpT + ∇ ⋅ gZ + ∇ ⋅ KE), heating due to water vapor divergence (∇⋅ Lqυ), surface evaporation (Evap), and surface soil wetness. DSE and water vapor convergences are computed from the model output fluxes of qυ, CpT, and gZ and do not include mass corrections to the wind.
Citation: Journal of Climate 33, 3; 10.1175/JCLI-D-19-0231.1
Although SWgCRE relates closely to surface temperature changes, other surface energy budget components provide significant contributions. In response to reduced precipitation and altered shortwave forcing, surface soil moisture and evaporation begin a steady decrease after M = −9 and sensible heat flux increases, communicating this thermal energy surplus to the overlying boundary layer (Figs. 9e,f). Sensible heat flux variations are strongly correlated with both surface and near-surface air temperature, but lead these temperatures as well as downward longwave flux increases by about 1–2 months (Figs. 9e,g). The evaporation response lags the precipitation as soil moisture storage is depleted (Figs. 9a and 10c). Individually, the surface longwave flux components are contemporaneously correlated to surface temperature. Unlike SWgCRE, which exhibits two maxima dynamically coupled to vertical velocity, the downward longwave radiation has a single maximum that tracks surface temperature (and the upward longwave radiation). These relationships are replicated in the M2AMIP results, albeit with smaller magnitudes.
This analysis suggests a pathway for SST/convectively induced free tropospheric temperature anomalies directly controlling near-surface land temperatures: The dynamically induced precipitation reductions, allied cloud reduction, and increased shortwave radiation drive subsequent surface temperature and sensible heat flux increases. Surface upward and downward LW fluxes act with the sensible heat flux to keep Ts and T2m strongly coupled. Since turbulent processes dominating land planetary boundary layer (PBL) structure operate on subdiurnal time scales (not shown) the 850-hPa temperature also correlates closely with Ts and T2m.
The midtropospheric temperature increases act to offset the destabilizing effects of boundary layer temperature rise and, in conjunction with reduced moisture flux convergence, to suppress precipitation. However, these temperature increases are dynamically tied to reduced upward motion and cloudiness and increased SW forcing of surface temperature. M2AMIP results, where the SST forcing is the most significant observational input data, show similar though more muted connections to those of MERRA-2.
One expects these global tropical flux relationships to have substantial regional/temporal granularity as evidenced by Figs. 6 and 7. In the remainder of this section we explore the contributions of processes in two different regions to the tropical land signal. Then in section 5, we will use MERRA-2 data at pentad frequencies to better test the lag relationships.
c. South America
El Niño impacts on South American precipitation anomalies (global monsoon region, as in Fig. 2) via teleconnections associated with Walker circulation variations have been widely documented (Kousky et al. 1984; Ropelewksi and Halpert 1987; Grimm 2003). In MERRA-2, land temperature closely follows the timing of precipitation (observation corrected) decreases (Fig. 11a). Ts, T2m, and downward longwave flux are all strongly correlated and lead T500, −omega, and SWgCRE by 1–3 months. More shortwave radiation is reaching the surface (Fig. 11c) owing to reduced cloud effects which, along with downward longwave, sustains the warming against sensible heat losses (Fig. 11e) that increase nearly 5 W m−2 in the six months before M = 0. Downward longwave radiation at the surface correlates highly with surface outgoing longwave radiation yielding a net cooling effect for the surface, and that cooling effect continues during El Niño.
South America global monsoon region composite El Niño anomaly time series for (left) MERRA-2 and (right) M2AMIP for several key quantities. Vertical velocity (−ω500) is directed positive for upward motion, determined at the 500-hPa level. The solid line indicates the composite average, and shading indicates ±1 standard deviation of the composite mean. Scales are color coded for each line. Red dashed lines in the top row indicate the 500-hPa temperature (T500). Dots indicate anomalies that are significantly different from zero (at 90% confidence).
Citation: Journal of Climate 33, 3; 10.1175/JCLI-D-19-0231.1
The reduced cloud effect in M2AMIP is also closely related to reduced upward vertical motions and precipitation (Figs. 11b,d) and T2m also leads T500. In MERRA-2, the increase in SWgCRE and decrease of upward motion at M = 0 is apparent, but also shows more composite variance, leading to limited statistical significance.
While the M2AMIP experiment exhibits several features similar to MERRA-2, there are some key differences worth noting. In South America for example, the anomalies tend to be more pronounced closer to the time of the peak Niño-3.4 SST anomaly. This is related to the interensemble variability when the forcing is still developing. In the model, the lowest precipitation deficit leads the warm temperatures by 2 months (Fig. 11b). In general, the M2AMIP El Niño signal starts later than the reanalysis and ends earlier, emphasizing the model’s response to the strongest SST forcing, but diminished in the buildup and decay phase when SST forcing is weaker.
d. Australia
While we have primarily focused on global monsoon areas, Australia only has a small area that by definition registers as global monsoon, yet the whole continent has strong connection to El Niño in this composite (Figs. 6 and 7). Here, we expand the composite El Niño Australia region to consider the whole continent. The precipitation deficit in MERRA-2 (using the observation corrections), the anomalous downward motion, and reduction in cloudiness all begin as much as 6 months before the peak in Niño-3.4 SSTs (Fig. 12c). The surface responds to the cloud and precipitation reductions with reduced evaporation (Fig. 13) and more sensible heating. Surface warming follows these forcings, but somewhat abruptly stops just after the peak in El Niño SSTs (M = 0), when continental anomalous upward vertical motion and precipitation positive anomalies return and temperature anomalies drop to near zero. Note that downward longwave radiation decreases sharply until M = −4 and remains anomalously low until M = 0, when T2m is increasing. The strong correlation between −omega (Fig. 12c) and downward surface LW radiation (Fig. 12g) is evidence of how strongly the El Niño–altered circulation controls atmospheric temperature and moisture changes. In contrast to South America where T2m and downward LW are tightly correlated, for Australia we see the T2m maximum (M = −1) leading downward LW (M = 3) by 4 months. Also, in contrast to the sequence over South America, here T500 is decreasing with time before M = 0 as T2m is increasing. T2m and T500 variations are anticorrelated. This tropospheric cooling is in response to the eastward shift of convection from the Indo-Pacific region into the central Pacific Ocean and, in conjunction with the loss of lower tropospheric moisture, controls the anomalous decrease of downward longwave radiation.
As in Fig. 11, but for Australia (land only). Scales are color coded for each line.
Citation: Journal of Climate 33, 3; 10.1175/JCLI-D-19-0231.1
As in Fig. 12, but for for evaporation (black; mm day−1) and surface soil wetness (red; m3 m−3).
Citation: Journal of Climate 33, 3; 10.1175/JCLI-D-19-0231.1
Previously, Figs. 6 and 7 showed significant disagreement between the M2AMIP and observations in Australia with warming onset too late, and maintaining the anomalously warm temperature beyond when MERRA-2 has returned to normal. In Figs. 12a and 12b, the M2AMIP precipitation continues reduced for some time after the Niño-3.4 peak SST and recovers more slowly than does the MERRA-2 observation-corrected precipitation. SWgCRE and vertical motions agree with the precipitation anomaly and the reduction of surface soil water anomalies (Fig. 13) is contributing to the increase in sensible heat flux and extension of the surface air temperature warm anomalies far beyond M = 0. We suspect that the poor M2AMIP simulation over Australia is linked to the fact that biases in precipitation and air–sea interaction over the Indo-Pacific region remain persistent both in AGCMs and in coupled models (Wu and Kirtman 2005; Bollasina and Nigam 2009). The cold T500 anomalies in the Southern Hemisphere subtropics (Fig. 6) that are weakly reproduced in M2AMIP at M = −4 and completely gone at M = 0 west of the date line are evidence of this deficient Indo-Pacific circulation response in M2AMIP.
Our analysis here has highlighted two regions with distinct responses to El Niño. Responses in other localities may have additional complexity. For example, the disparate climatic regimes over Africa are linked to the semiannual cycle and the complicating influence of Atlantic SSTs and the African monsoonal system. The seasonality of the Asian monsoon and the influence of IOZM mean that the response of South Asia and the Maritime Continent to ENSO likely have even more complex regime structure. A more detailed treatment of these local contributions to the global signal is warranted.
5. High frequency and individual El Niño variability
To this point the discussion has focused on the monthly composite El Niño, in which relationships are inherently related to seasonal and longer time scale variability. The composite average minimizes internal atmospheric variability as well as random differences among El Niño events, retaining any consistent response to evolving far-field tropical SST anomalies. In addition, this analysis allowed comparison to AMIP experiments and comparison across several long duration observational and reanalysis data. To refine the global tropical lag relationships and connections, we now evaluate the composite in MERRA-2 with pentad time series (M2AMIP ensemble data were not saved at such high frequency). We compute pentads from MERRA-2’s hourly data collections and composite the same El Niño events averaging 5 days of data for 73 pentads per year (with leap days averaged into the 12th pentad of each year). As in the monthly composite, a climatological annual cycle (1981–2010) is removed from the pentad time series, and the variables are detrended before making the composite mean. The pentad composite time series contains more variability. In the next section, we present the pentad composite evolution, but first we test the higher-frequency relationships of the physical quantities that contribute to the tropical land surface warming and drying.
a. High-frequency variability
Figure 14 shows the lead/lag correlations for several quantities as related to the land surface temperature and precipitation/evaporation. The time correlations are computed for 101 pentads (±8 months) relative to the pentad with peak Niño-3.4 SST. Lead/lag correlations are determined by shifting the first variable listed, relative to the second. In this way, we highlight the modes of variation between the relevant quantities during the evolution of El Niño. For example, 500-hPa vertical velocity and SWgCRE correlate most strongly at zero lag, indicating a mode of rapid response of one to the other. Given the large-scale change over the composite El Niño, this represents the cloud responding to the dynamical variations represented in vertical velocity.
Tropical land global monsoon region lead lag correlations for (a) surface temperature variables and (b) water cycle variables. The correlations are developed from the MERRA-2 pentad time series, with leading correlations as negative days and lag correlations as positive days for pairs of variables x:y (see labels), where the x is computed as leading or lagging y. Correlation units are dimensionless, and the values are computed over 101 data points (8 months of pentads), and values of 0.26 are significant at 99% confidence. Here, 0 days refers to contemporaneous correlation of pentads, not the M = 0 El Niño peak.
Citation: Journal of Climate 33, 3; 10.1175/JCLI-D-19-0231.1
On the other hand, SWgCRE leads surface temperature Ts even out to leads of 100 days (~3 months), suggesting the effects of persistent radiative forcing. However, there is a maximum correlation at a lead of 25 days. The surface temperature leads sensible heating, but with a broad peak correlation with an approximate lead of 15–10 days. This contributes to a rapid response of T850 to the surface temperature, which correlate to 0.99 at lag zero, indicating strong mixing within the planetary boundary layer. Surface temperature does have a low-frequency leading correlation with T500, but their correlation is declining at zero lag. These results show the importance of multiple temporal scales (weather to interannual) governing the nature of the surface warming as the circulation reduces clouds over land, increases the shortwave, and warms the surface and the lower troposphere.
The water cycle is strongly coupled to these processes. In Fig. 14b, as could be expected, precipitation (model generated) is highly correlated with vertical motion at zero lag. Precipitation that forces soil moisture (note for MERRA-2 this is the observation-corrected precipitation) leads evaporation, primarily at longer lead times. This reflects the influence of the soil water storage on the evaporation and the time it takes to deplete the soil water and warm the surface, as the soil wetness also has a peak correlation leading the evaporation (by 15 days). Surface temperature is already increasing by the time evaporation is decreasing.
b. 2015/16 El Niño
The 2015/16 El Niño produced tropical ocean temperatures as high as the warmest previous events in recent records (Santoso et al. 2017). A significant portion of the globe (12%) experienced drought conditions, some of the highest in modern records, related to El Niño influence (Dunn et al. 2017). Here, pentad resolution data permit closer examination of how one particular event, in this case the extreme 2015/16 El Niño, evolves compared to the composite (Fig. 15) and the contributions from various regions (Fig. 16). The 2015/16 El Niño is approximately 1 K warmer (its peak occurs 9 November 2015 using the MERRA-2 SST pentad anomalies) than the composite Niño-3.4 temperature, while it is warmer than the composite throughout the entire 2-yr period of the composite (Fig. 15a). Tropical land 2-m air temperature in 2015 is warm for the 6 months leading up to the Niño-3.4 peak, but then drops dramatically before warming again 3 months after the peak. When the land is warm, it is much warmer than the composite El Niño (more than 1 standard deviation). Early rises in T2m are driven by South America and, after M = −1, by Australia and tropical South Asia (Fig. 16a). Negative precipitation anomalies in the months leading up to the peak of El Niño (Fig. 15b) are followed by a recovery punctuated by episodic spikes of positive anomalies near M = 2 and M = 4. Regionally, South America experiences the largest reduction in precipitation on average during the 2015/16 El Niño (Fig. 16b). Record-breaking temperatures and drought occurred in Amazonia, but also a significant wet/dry dipole in precipitation occurred (Jiménez-Muñoz et al. 2016). Australia and tropical South Asia have much weaker, but predominantly negative, precipitation anomalies in the six months prior to M = 0, but little T2m response. Conversely, Africa in 2016 becomes sporadically wetter and cooler than average just before M = 0 and is the only region with T2m cooling and anomalously low sensible heat flux during the peak of El Niño, from M = −3 to M = 3 (Fig. 16c). It is the temporary peak in precipitation over all regions centered near M = 2, along with a brief spike in South American precipitation, that drives the drop in T2m and T850 during this period. Although the 2015/16 sequence of events is “noisy” compared to the composite mean (Fig. 15), the relationships are similar over tropical land as a whole; precipitation correlates well with vertical velocity and −SWgCRE variations, at least on seasonal time scales, and is consistent with changes in latent and sensible heat flux.
(a) The Niño-3.4 region SST composite with the 2015/16 event. (b)–(f) Tropical land global monsoon regions composite El Niño for the MERRA-2 pentad time series (solid lines with shaded ensemble standard deviation) in comparison with the 2015/16 El Niño (dashed line) for several key quantities (where LE is surface latent heat, SfcNet is the net flux of energy into the surface, and TPW is total precipitable water). Black-colored variables are scaled on the left axis, and red-colored variables are scaled on the right axis.
Citation: Journal of Climate 33, 3; 10.1175/JCLI-D-19-0231.1
Tropical land regions composite El Niño for the MERRA-2 pentad time series (dashed lines) compared with the 2015/16 El Niño (solid lines) for (a) 2-m air temperature (T2m), (b) modeled precipitation (Prec), (c) surface sensible heat (Hs), and (d) midtropospheric temperature at 500 hPa (T500) over the main tropical regions (colors match the those in the figure title). The Australia region is for the whole continent, and the other regions are for the global monsoon areas of the respective regions.
Citation: Journal of Climate 33, 3; 10.1175/JCLI-D-19-0231.1
Interestingly, while T2m and T850 warm along with the surface and the turbulent heating increases following precipitation declines in the 2015 event, T500 only rises after M = 0 (Fig. 15d), completely at odds with the low-level warming in the developing months. Australia is the dominant contributor to this midlevel cool anomaly (Fig. 16d) and this coolness, while extreme, follows the evolution of the composite prior to M = 0. This behavior contrasts in general with that of South America, which warms at midlevels beginning at M = −6 and cools after M = 3. The exceptional warming at midlevels afterward from M = 3 to M = 5 is driven initially by the tropical South Asian and Africa regions and is sustained by changes over Australia. We suspect that this warming is due to the relaxation of cold SSTs in the Indo-Pacific region and return toward climatological precipitation there. A strongly positive IOZM event (cold eastern, warm western equatorial Indian Ocean SST) develops during late 2015 (https://www.esrl.noaa.gov/psd/gcos_wgsp/Timeseries/DMI/). Saji and Yamagata (2003) and Cai et al. (2011) make note of equivalent barotropic wave structures over the Indo-Pacific subtropics during El Niño that get augmented by positive IOZM events. These flow patterns are consistent with the late 2015 midlevel coolness over Australia and tropical South Asia seen in Fig. 16d.
These results emphasize that the composite provides general guidance on the dominant circulation, flux processes, and tropical land temperature response during typical El Niño events, at least on a tropical average basis. However, as evidenced by the high-frequency variations and intermittent spikes in the time series (e.g., the tropical mean precipitation increase near T = 2) even one of the strongest El Niño events in the last 35 years can manifest land and atmosphere processes through the life cycle that differ significantly from the composite El Niño. Weather-scale events, intraseasonal variability, and especially internal atmospheric variability can add significant variance to flux processes.
6. Summary and conclusions
In this work we have sought to better understand how various components of the tropical land surface energy balance adjust during the evolution of typical El Niño events. Using a compositing technique, and centering the composite members on the time of the Niño-3.4 SST maxima, allowed us to identify the role of various processes leading to predominant drought and warming. Preliminary observational assessment justified the use of MERRA-2 reanalysis and AMIP fluxes and related variables over the ±12-month period.
In MERRA-2, we find that dynamical effects of atmospheric teleconnections embodied in Walker/Hadley dislocations during El Niño couple via tropospheric vertical motion changes to SWgCRE, precipitation anomalies, and eventually other components of the surface energy balance (Fig. 9). These changes are consistent with reduced moisture import from ocean to land and with weakened DSE+KE export (but with a partial recovery near M = 0). Individually these fluxes are consistent with arguments outlined in Lambert et al. (2011) for SST-driven changes, but the for El Niño events the present results indicate little evidence of increased net energy export. However, it should be noted that the interevent spread of time variations of the tropical continental energy fluxes across the range of El Niño events is large (as evidenced by low statistical significance). We suspect the partial recovery in moisture transport near M = 0 is tied to the relaxation of the downward component of the Walker circulation and increased precipitation anomalies over much of East Africa (Fig. 7). A complicating factor in this partial recovery of is the tendency for positive phasing of the IOZM with El Niño, which, although important for increasing East African rainfall, also tends to produce drought over southern Australia.
From a global tropic perspective the quadrupole pattern of warming over the eastern Pacific and cooling over the Indo-Pacific (dominantly in the SH) during the developing and mature stages of El Niño is accompanied by distinctive regional surface energy flux changes. Over South America Ts, T2m, and downward LW are in phase, whereas for Australia a reduction in downward LW maximizes at M = 3, some 4 months after T2m and Ts anomalies reach their peak. The anomalous downward motion, reduced precipitation, and increased Ts, T2m, Hs, and SWgCRE over Australia during growth stages of El Niño (Fig. 12) occur with a cold troposphere (Fig. 6) linked to reduced precipitation and associated dynamical heating over the Indo-Pacific region (Fig. 7). In contrast, South America has similar upper-level anomalous downward motion and surface flux changes that occur at and after El Niño maturity, but in the presence of tropospheric warmth that has now spread throughout the equatorial region between M = 0 and M = 4. This sequence of events and processes is more in line with the dynamical picture argued by earlier papers (e.g., Lambert and Chiang 2007; Joshi et al. 2008; Tyrrell et al. 2015). We find that, globally, Ts and T2m lag relationships to midtropospheric warmth are ambiguous; T500 lags Ts increases in MERRA-2 (Figs. 4b and 14) but the timing is ill defined. On the other hand, the CRU T2m and ERA5 lag relationship is reversed (Fig. 4a), but also ill defined. As noted earlier though, T500 increases prior to the time of T2m maximum are larger than those near the surface. So midtropospheric warming, anomalous descent, and increased SW absorption are all linked.
In terms of the surface energy budget fluxes, we also find a significant role for sensible heat flux increases, which act to offset SWgCRE and downward longwave effects in the face of reduced precipitation and LE. Because sensible heat flux increases warm the overlying near-surface air (and ultimately the PBL), we interpret the increases in downward LW as responding more to this turbulent energy transport than being coupled to the midtropospheric temperature by virtue of changes in heat transport. Figure 9 suggests that the tropical land surface integrates sustained SW anomalous forcing and that the sensible heat flux anomalies couple Ts, T2m, and downward LW. Correlations of Ts with T850 are stronger than with T500 (Fig. 14), supporting this view. This interpretation is consistent with the results of surface flux process sensitivity tests in Chiang and Lintner (2005) that showed that constraining sensible heating drastically distorted the surface temperature response. The significant increases in SH seen in our study also echo the results of Sutton et al. (2007), who found reduced moisture transport to land and significant increases in sensible heat flux in the face of SST increases in a global warming scenario.
Compo and Sardeshmukh (2009), on the other hand, find a dominant role for Ts forcing by downward longwave. In trying to reconcile our results with theirs we first note that their study involves changes on decadal time scales which greatly diminishes the average El Niño signal and precludes resolving life cycle effects. Furthermore, their domain encompasses global land whereas ours is a tropical subset. But as in the present study Compo and Sardeshmukh (2009) also find a significant role for SW forcing of Ts that relates to midtropospheric vertical motion changes. As we argue above, since turbulent processes typically couple surface and PBL temperature strongly, a significant component of the correlation between Ts and downward LW radiation may mask the intrinsic importance of the SW and cloud effects on Ts.
The M2AMIP ensemble reproduces many of the changes in fluxes in a tropical mean sense (Fig. 9). While the nature of the teleconnection changes and the effects on flux changes over South America can be reproduced approximately, the Indo-Pacific basin and the adjacent continents are much more of a challenge because of strong two-way coupling of the ocean–atmosphere system there. Consequently, the equatorially trapped atmospheric response to deep convective heating variations over the SST warm pool (and associated equivalent barotropic waves) is problematic in AMIP experiments. The model responds more closely to SSTA observations closer to the peak of El Niño SSTs. This likely results from AMIP experiment ensembles averaging out internal atmospheric variability, enhancing SST-forced signals.
Compositing approaches have both merits and drawbacks—one prominent shortcoming being the associated blurring of different El Niño flavors into one mean event. In comparing the 2015/16 event to the composite mean, significant weather variations were also noted, including the significant effect of a significant temporary spike in precipitation a few months after M = 0. However, the distinct difference in response to El Niño between Australia and South America was still in evidence. To the extent that lagged, far-field ocean warming effects are present (Klein et al. 1999; Alexander et al. 2002) relating systematically to Niño 3.4 at M = 0, they are captured by the compositing method and implicit in lag relationships between flux quantities.
In this composite, we have not attempted to separate so-called eastern Pacific versus central Pacific events (Kao and Yu 2009; Hu et al. 2016). Nor have we attempted to control for the variable presence of the IOZM or cross-equatorial SSTA in the Atlantic widely known to affect Brazilian hydrometeorology. This may explain to some extent the ambiguities in some of the regional time series shown here. Longer time series of reanalyses may allow more events that would permit adequate sampling and stratification to address these factors. Nevertheless, the present study identifies the sequence of processes important to global tropical land response to El Niño forcing. Regional and local response to ENSO events is a significant challenge in subseasonal to seasonal forecasting as well and continued efforts to understand the hydrometeorological responses to this continuum of events remain important work.
Acknowledgments
MERRA-2 and M2AMIP were developed and published with support from the NASA Modeling and Analysis Program. The first two authors were supported by the NASA Energy and Water Cycle Studies program. The third author is supported by the NASA Radiation Sciences program and CERES Mission. We thank Young-Kwon Lim and Siegfried D. Schubert for discussions on ENSO and King-Sheng Tai and Allison Collow for efforts in processing the M2AMIP experiment. Useful contributions to the paper from three anonymous reviewers are greatly appreciated.
APPENDIX
Acronyms
| AIRS | Atmospheric Infrared Sounder |
| AMIP | Atmospheric Model Intercomparison Project |
| AMSU | Advanced Microwave Sounding Unit |
| CERES | Clouds and the Earth’s Radiant Energy System |
| CPCU | Climate Prediction Center Unified precipitation data |
| CRU | Climate Research Unit |
| DSE | Dry static energy |
| EBAF | Energy Balanced and Filled |
| ECMWF | European Centre for Medium-Range Weather Forecasts |
| ERA5 | ECWMF Reanalysis 5 |
| GPCP | Global Precipitation Climatology Project |
| HadISST | Hadley Centre Sea Ice and Sea Surface Temperature |
| ISCCP | International Satellite Cloud Comparison Project |
| IOZM | Indian Ocean zonal mode |
| MERRA | Modern-Era Retrospective Analysis for Research and Applications |
| M2AMIP | MERRA-2 Atmospheric Model Intercomparison Project ensemble experiment |
| MSU | Microwave Sounding Unit |
| NASA | National Aeronautics and Space Administration |
| Niño-3.4 | refers to the index based on SST and region of the tropical Pacific Ocean defined in the text |
| SRB | Surface radiation budget |
| SSM/I | Special Sensor Microwave Imager |
| SWgCRE | Shortwave ground cloud radiative effect (all sky minus clear sky) |
| Ts, T2m, T850, and T500 | Temperature at the surface, 2 m AGL, 850 hPa, and 500 hPa, respectively |
REFERENCES
Adler, R. F., G. Gu, M. Sapiano, J.-J. Wang, and G. J. Huffman, 2017: Global precipitation: Means, variations and trends during the satellite era (1979–2014). Int. J. Climatol., 38, 679–699, https://doi.org/10.1007/s10712-017-9416-4.
Alexander, L. V., P. Uotila, and N. Nicholls, 2009: Influence of sea surface temperature variability on global temperature and precipitation extremes. J. Geophys. Res., 114, D18116, https://doi.org/10.1029/2009JD012301.
Alexander, M. A., I. Bladé, M. Newman, J. R. Lanzante, N. Lau, and J. D. Scott, 2002: The atmospheric bridge: The influence of ENSO teleconnections on air–sea interaction over the global oceans. J. Climate, 15, 2205–2231, https://doi.org/10.1175/1520-0442(2002)015<2205:TABTIO>2.0.CO;2.
Bollasina, M., and S. Nigam, 2009: Indian Ocean SST, evaporation, and precipitation during the South Asian summer monsoon in IPCC-AR4 coupled simulations. Climate Dyn., 33, 1017–1032, https://doi.org/10.1007/s00382-008-0477-4.
Bosilovich, M. G., J. Chen, F. R. Robertson, and R. F. Adler, 2008: Evaluation of global precipitation in reanalyses. J. Appl. Meteor. Climatol., 47, 2279–2299, https://doi.org/10.1175/2008JAMC1921.1.
Bosilovich, M. G., F. R. Robertson, and J. Chen, 2011: Global energy and water budgets in MERRA. J. Climate, 24, 5721–5739, https://doi.org/10.1175/2011JCLI4175.1.
Bosilovich, M. G., and Coauthors, 2015: MERRA-2: Initial evaluation of the climate. NASA Technical Report Series on Global Modeling and Data Assimilation NASA/TM-2015-104606, Vol. 43, 136 pp.
Bosilovich, M. G., F. R. Robertson, L. Takacs, A. Molod, and D. Mocko, 2017: Atmospheric water balance and variability in the MERRA-2 reanalysis. J. Climate, 30, 1177–1196, https://doi.org/10.1175/JCLI-D-16-0338.1.
Cai, W., P. van Rensch, T. Cowan, and H. H. Hendon, 2011: Teleconnection pathways of ENSO and the IOD and the mechanisms for impacts on Australian rainfall. J. Climate, 24, 3910–3923, https://doi.org/10.1175/2011JCLI4129.1.
CERES Science Team, 2018: EBAF-Surface Ed4.0. NASA Atmospheric Science Data Center (ASDC), accessed 14 February 2018, https://doi.org/10.5067/Terra+Aqua/CERES/EBAF-Surface_L3B004.0.
Charney, J. G., 1963: A note on large-scale motions in the tropics. J. Atmos. Sci., 20, 607–609, https://doi.org/10.1175/1520-0469(1963)020<0607:ANOLSM>2.0.CO;2.
Chiang, J. C., and B. R. Lintner, 2005: Mechanisms of remote tropical surface warming during El Niño. J. Climate, 18, 4130–4149, https://doi.org/10.1175/JCLI3529.1.
Collow, A. B. M., S. P. Mahanama, M. G. Bosilovich, R. D. Koster, and S. D. Schubert, 2017: An evaluation of teleconnections over the United States in an ensemble of AMIP simulations with the MERRA-2 configuration of the GEOS atmospheric model. NASA Rep. NASA/TM-2017-104606, Vol. 47, 68 pp., https://gmao.gsfc.nasa.gov/pubs/docs/Collow963.pdf.
Compo, G. P., and P. D. Sardeshmukh, 2009: Oceanic influences on recent continental warming. Climate Dyn., 32, 333–342, https://doi.org/10.1007/s00382-008-0448-9.
Curtis, S., and R. F. Adler, 2003: Evolution of El Niño–precipitation relationships from satellites and gauges. J. Geophys. Res., 108, 4153, https://doi.org/10.1029/2002JD002690.
Dai, A., and T. Wigley, 2000: Global patterns of ENSO-induced precipitation. Geophys. Res. Lett., 27, 1283–1286, https://doi.org/10.1029/1999GL011140.
Dommenget, D., 2009: The ocean’s role in continental climate variability and change. J. Climate, 22, 4939–4952, https://doi.org/10.1175/2009JCLI2778.1.
Dunn, R. J. H., D. F. Hurst, N. Gobron, and K. M. Willett, Eds., 2017: Global climate [in “State of the Climate in 2016”]. Bull. Amer. Meteor. Soc., 98, S93–S128., https://doi.org/10.1175/2017BAMSStateoftheClimate.1.
Fasullo, J. T., 2010: Robust land–ocean contrasts in energy and water cycle feedbacks. J. Climate, 23, 4677–4693, https://doi.org/10.1175/2010JCLI3451.1.
Gelaro, R., and Coauthors, 2017: The Modern-Era Retrospective Analysis for Research and Applications, version 2 (MERRA-2). J. Climate, 30, 5419–5454, https://doi.org/10.1175/JCLI-D-16-0758.1.
Grimm, A. M., 2003: The El Niño impact on the summer monsoon in Brazil: Regional processes versus remote influences. J. Climate, 16, 263–280, https://doi.org/10.1175/1520-0442(2003)016<0263:TENIOT>2.0.CO;2.
Gu, G., R. F. Adler, G. J. Huffman, and S. Curtis, 2007: Tropical rainfall variability on interannual-to-interdecadal and longer time scales derived from the GPCP monthly product. J. Climate, 20, 4033–4046, https://doi.org/10.1175/JCLI4227.1.
Harris, I., and P. D. Jones, 2017: CRU TS4.01: Climatic Research Unit (CRU) Time-Series (TS) version 4.01 of high-resolution gridded data of month-by-month variation in climate (Jan. 1901–Dec. 2016). Centre for Environmental Data Analysis, accessed 8 December 2017, https://catalogue.ceda.ac.uk/uuid/58a8802721c94c66ae45c3baa4d814d0.
Harris, I., P. D. Jones, T. J. Osborn, and D. H. Lister, 2014: Updated high-resolution grids of monthly climatic observations—The CRU TS3.10 dataset. Int. J. Climatol., 34, 623–642, https://doi.org/10.1002/joc.3711.
Hersbach, H., and Coauthors, 2018: Operational global reanalysis: Progress, future directions and synergies with NWP. ECMWF ERA Report Series No. 27, 63 pp., https://doi.org/10.21957/tkic6g3wm.
Hoskins, B. J., and D. J. Karoly, 1981: The steady linear response of a spherical atmosphere to thermal and orographic forcing. J. Atmos. Sci., 38, 1179–1196, https://doi.org/10.1175/1520-0469(1981)038<1179:TSLROA>2.0.CO;2.
Hu, X., S. Yang, and M. Cai, 2016: Contrasting the eastern Pacific El Niño and the central Pacific El Niño: Process-based feedback attribution. Climate Dyn., 47, 2413–2424, https://doi.org/10.1007/s00382-015-2971-9.
Jiménez-Muñoz, J. C., C. Mattar, J. Barichivich, A. Santamaría-Artigas, K. Takahashi, Y. Malhi, J. A. Sobrino, and G. van der Schrier, 2016: Record-breaking warming and extreme drought in the Amazon rainforest during the course of El Niño 2015–2016. Sci. Rep., 6, 33130, https://doi.org/10.1038/srep33130.
Joshi, M. M., J. M. Gregory, M. J. Webb, D. M. H. Sexton, and T. C. Johns, 2008: Mechanisms for the land/sea warming contrast exhibited by simulations of climate change. Climate Dyn., 30, 455–465, https://doi.org/10.1007/s00382-007-0306-1.
Kao, H. Y., and J.-Y. Yu, 2009: Contrasting eastern-Pacific and central-Pacific types of ENSO. J. Climate, 22, 615–632, https://doi.org/10.1175/2008JCLI2309.1.
Kato, S., and Coauthors, 2018: Surface irradiances of Edition 4.0 Clouds and the Earth’s Radiant Energy System (CERES) Energy Balanced and Filled (EBAF) data product. J. Climate, 31, 4501–4527, https://doi.org/10.1175/JCLI-D-17-0523.1.
Klein, S. A., B. J. Soden, and N.-C. Lau, 1999: Remote sea surface temperature variations during ENSO: Evidence for a tropical atmospheric bridge. J. Climate, 12, 917–932, https://doi.org/10.1175/1520-0442(1999)012<0917:RSSTVD>2.0.CO;2.
Kousky, V. E., M. T. Kagano, and I. F. A. Cavalcanti, 1984: A review of the Southern Oscillation: Oceanic–atmospheric circulation changes and related rainfall anomalies. Tellus, 36A, 490–504, https://doi.org/10.1111/j.1600-0870.1984.tb00264.x.
Lambert, F. H., and J. C. H. Chiang, 2007: Control of land–ocean temperature contrast by ocean heat uptake. Geophys. Res. Lett., 34, L13704, https://doi.org/10.1029/2007GL029755.
Lambert, F. H., M. J. Webb, and M. M. Joshi, 2011: The relationship between land–ocean surface temperature contrast and radiative forcing. J. Climate, 24, 3239–3256, https://doi.org/10.1175/2011JCLI3893.1.
Lim, Y., R. M. Kovach, S. Pawson, and G. Vernieres, 2017: The 2015/16 El Niño event in context of the MERRA-2 reanalysis: A comparison of the tropical Pacific with 1982/83 and 1997/98. J. Climate, 30, 4819–4842, https://doi.org/10.1175/JCLI-D-16-0800.1.
Manabe, S., R. J. Stouffer, M. J. Spelman, and K. Bryan, 1991: Transient responses of a coupled ocean–atmosphere model to gradual changes of atmospheric CO2. Part I: Annual mean response. J. Climate, 4, 785–818, https://doi.org/10.1175/1520-0442(1991)004<0785:TROACO>2.0.CO;2.
Mayer, M., and L. Haimberger, 2012: Poleward atmospheric energy transports and their variability as evaluated from ECMWF reanalysis data. J. Climate, 25, 734–752, https://doi.org/10.1175/JCLI-D-11-00202.1.
Mayer, M., K. E. Trenberth, L. Haimberger, and J. T. Fasullo, 2013: The response of tropical atmospheric energy budgets to ENSO. J. Climate, 26, 4710–4724, https://doi.org/10.1175/JCLI-D-12-00681.1.
McCarty, W., L. Coy, R. Gelaro, A. Huang, D. Merkova, E. B. Smith, M. Sienkiewicz, and K. Wargan, 2016: MERRA-2 input observations: Summary and assessment. NASA Tech. Rep. Series on Global Modeling and Data Assimilation NASA/TM-2016-104606, Vol. 46, 44 pp., https://gmao.gsfc.nasa.gov/pubs/docs/McCarty885.pdf.
Miralles, D. G., and Coauthors, 2014: El Niño–La Niña cycle and recent trends in continental evaporation. Nat. Climate Change, 4, 122–126, https://doi.org/10.1038/nclimate2068.
Molod, A., L. Takacs, M. Suarez, and J. Bacmeister, 2015: Development of the GEOS-5 atmospheric general circulation model: Evolution from MERRA to MERRA-2. Geosci. Model Dev., 8, 1339–1356, https://doi.org/10.5194/gmd-8-1339-2015.
Oort, A. H., and J. J. Yienger, 1996: Observed interannual variability in the Hadley circulation and its connection to ENSO. J. Climate, 9, 2751–2767, https://doi.org/10.1175/1520-0442(1996)009<2751:OIVITH>2.0.CO;2.
Pillai, P. A., S. A. Rao, G. George, D. N. Rao, S. Mahapatra, M. Rajeevan, A. Dhakate, and K. Salunke, 2017: How distinct are the two flavors of El Niño in retrospective forecasts of Climate Forecast System version 2 (CFSv2)? Climate Dyn., 48, 3829–3854, https://doi.org/10.1007/s00382-016-3305-2.
Rayner, N. A., D. E. Parker, E. B. Horton, C. K. Folland, L. V. Alexander, D. P. Rowell, E. C. Kent, and A. Kaplan, 2003: Global analyses of sea surface temperature, sea ice, and night marine air temperature since the late nineteenth century. J. Geophys. Res., 108, 4407, https://doi.org/10.1029/2002JD002670.
Reichle, R., Q. Liu, R. Koster, C. Draper, S. Mahanama, and G. Partyka, 2017: Land surface precipitation in MERRA-2. J. Climate, 30, 1643–1664, https://doi.org/10.1175/JCLI-D-16-0570.1.
Robertson, F. R., M. G. Bosilovich, and J. B. Roberts, 2016: Reconciling land–ocean moisture transport variability in reanalyses with P-ET in observationally driven land surface models. J. Climate, 29, 8625–8646, https://doi.org/10.1175/JCLI-D-16-0379.1.
Ropelewski, C. F., and M. S. Halpert, 1987: Global and regional scale precipitation patterns associated with the El Niño/Southern Oscillation. Mon. Wea. Rev., 115, 1606–1626, https://doi.org/10.1175/1520-0493(1987)115<1606:GARSPP>2.0.CO;2.
Rossow, W. B., and R. A. Schiffer, 1999: Advances in understanding clouds from ISCCP. Bull. Amer. Meteor. Soc., 80, 2261–2288, https://doi.org/10.1175/1520-0477(1999)080<2261:AIUCFI>2.0.CO;2.
Saji, N. H., and T. Yamagata, 2003: Possible impacts of Indian Ocean dipole mode events on global climate. Climate Res., 25, 151–169, https://doi.org/10.3354/cr025151.
Santoso, A., M. J. McPhaden, and W. Cai, 2017: The defining characteristics of ENSO extremes and the strong 2015/2016 El Niño. Rev. Geophys., 55, 1079–1129, https://doi.org/10.1002/2017RG000560.
Schubert, S. D., and Coauthors, 2016: Global meteorological drought: A synthesis of current understanding with a focus on SST drivers of precipitation deficits. J. Climate, 29, 3989–4019, https://doi.org/10.1175/JCLI-D-15-0452.1.
Simmons, A. J., P. Berrisford, D. P. Dee, H. Hersbach, S. Hirahara, and J.-N. Thépaut, 2017: A reassessment of temperature variations and trends from global reanalyses and monthly surface climatological datasets. Quart. J. Roy. Meteor. Soc., 143, 101–119, https://doi.org/10.1002/qj.2949.
Sobel, A. H., I. M. Held, and C. S. Bretherton, 2002: The ENSO signal in tropical tropospheric temperature. J. Climate, 15, 2702–2706, https://doi.org/10.1175/1520-0442(2002)015<2702:TESITT>2.0.CO;2.
Stackhouse, P. W., Jr., S. K. Gupta, S. J. Cox. T. Zhang, J. C. Mikovitz, and L. M. Hinkelman, 2011: The NASA/GEWEX Surface Radiation Budget Release 3.0: 24.5-year dataset. GEWEX News, No. 21 (1), International GEWEX Project Office, Silver Spring, MD, 10–12.
Stephens, G. L., and Coauthors, 2018: Regional intensification of the tropical hydrological cycle during ENSO. Geophys. Res. Lett., 45, 4361–4370, https://doi.org/10.1029/2018GL077598.
Sutton, R. T., B. Dong, and J. M. Gregory, 2007: Land/sea warming ratio in response to climate change: IPCC AR4 model results and comparison with observations. Geophys. Res. Lett., 34, L02701, https://doi.org/10.1029/2006GL028164.
Takacs, L. L., M. Suárez, and R. Todling, 2016: Maintaining atmospheric mass and water balance in reanalyses. Quart. J. Roy. Meteor. Soc., 142, 1565–1573, https://doi.org/10.1002/qj.2763.
Trenberth, K. E., and L. Smith, 2006: The vertical structure of temperature in the tropics: Different flavors of El Niño. J. Climate, 19, 4956–4973, https://doi.org/10.1175/JCLI3891.1.
Trenberth, K. E., and J. T. Fasullo, 2013: Regional energy and water cycles: Transports from ocean to land. J. Climate, 26, 7837–7851, https://doi.org/10.1175/JCLI-D-13-00008.1.
Trenberth, K. E., J. M. Caron, D. P. Stepaniak, and S. Worley, 2002: Evolution of El Niño–Southern Oscillation and global atmospheric surface temperatures. J. Geophys. Res., 107, 4065, https://doi.org/10.1029/2000JD000298.
Trenberth, K. E., J. T. Fasullo, and J. Mackaro, 2011: Atmospheric moisture transports from ocean to land and global energy flows in reanalyses. J. Climate, 24, 4907–4924, https://doi.org/10.1175/2011JCLI4171.1.
Tyrrell, N. L., D. Dommenget, C. Frauen, S. Wales, and M. Rezny, 2015: The influence of global sea surface temperature variability on the large-scale land surface temperature. Climate Dyn., 44, 2159–2176, https://doi.org/10.1007/s00382-014-2332-0.
Walker, G. T., 1924: Correlation in seasonal variations of weather, IX. A further study of world weather. Mem. India Meteor. Dept., 24, 275–333.
Wang, B., J. Liu, H. Kim, P. J. Webster, and S. Yim, 2012: Recent change of the global monsoon precipitation (1979–2008). Climate Dyn., 39, 1123–1135, https://doi.org/10.1007/s00382-011-1266-z.
Webster, P., A. Moore, J. P. Loschnigg, and R. R. Leben., 1999: Coupled ocean–atmosphere dynamics in the Indian Ocean during 1997–98. Nature, 401, 356–360, https://doi.org/10.1038/43848.
Wu, R., and B. P. Kirtman, 2005: Roles of Indian and Pacific Ocean air–sea coupling in tropical atmospheric variability. Climate Dyn., 25, 155–170, https://doi.org/10.1007/s00382-005-0003-x.
Young, A. H., K. R. Knapp, A. Inamdar, W. Hankins, and W. B. Rossow, 2018: The International Satellite Cloud Climatology Project H-Series climate data record product. Earth Syst. Sci. Data, 10, 583–593, https://doi.org/10.5194/essd-10-583-2018.
Yu, L., and M. M. Rienecker, 1999: Mechanisms for the Indian Ocean warming during the 1997–98 El Niño. Geophys. Res. Lett., 26, 735–738, https://doi.org/10.1029/1999GL900072.
Yulaeva, E., and J. M. Wallace, 1994: The signature of ENSO in global temperature and precipitation fields derived from the Microwave Sounding Unit. J. Climate, 7, 1719–1736, https://doi.org/10.1175/1520-0442(1994)007<1719:TSOEIG>2.0.CO;2.
Zhang, T., P. W. Stackhouse Jr., S. K. Gupta, S. J. Cox, and J. C. Mikovitz, 2015: The validation of the GEWEX SRB surface longwave flux data products using BSRN measurements. J. Quant. Spectrosc. Radiat. Transfer, 150, 134–147, https://doi.org/10.1016/j.jqsrt.2014.07.013.
Zhang, T., P. W. Stackhouse Jr., S. J. Cox, M. J. Mikovitz, and C. N. Long, 2019: Clear-sky shortwave downward flux at the Earth’s surface: Ground-based data vs. satellite-based data. J. Quant. Spectrosc. Radiative Transfer, 224, 247–260, https://doi.org/10.1016/j.jqsrt.2018.11.015.
