Global vegetation productivity responses to the West Pacific Warm Pool

Science of the Total Environment 655 (2019) 641–651

Contents lists available at ScienceDirect

Science of the Total Environment journal homepage: www.elsevier.com/locate/scitotenv

Global vegetation productivity responses to the West Pacific Warm Pool Mei Huang a,⁎,1, Zhaosheng Wang a,1, Shaoqiang Wang a, Fengxue Gu b, He Gong a, Man Hao a, Yaping Shao c a

Key Laboratory of Ecosystem Network Observation and Modeling, Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, Beijing 100101, China Key Laboratory of Dryland Agriculture, Ministry of Agriculture, Institute of Environment and Sustainable Development in Agriculture, Chinese Academy of Agricultural Sciences, Beijing 100081, China c Institute for Geophysics and Meteorology, University of Cologne, 50923, Germany b

H I G H L I G H T S

G R A P H I C A L

A B S T R A C T

• WPWP SST is associated with global vegetation productivity. • A warm (cool) WPWP promotes (suppresses) vegetation growth in the north of 70°N. • A warm (cool) WPWP suppresses (enhances) vegetation growth in 10–30°S. • WPI serves as a useful climate index for studying ocean-vegetation teleconnections.

a r t i c l e

i n f o

Article history: Received 27 July 2018 Received in revised form 21 October 2018 Accepted 11 November 2018 Available online 12 November 2018 Editor: SCOTT SHERIDAN Keywords: West Pacific Warm pool Climate index NPP NDVI Teleconnection

a b s t r a c t Sea surface temperatures (SSTs) strongly influence atmospheric circulation and the Earth's climate, which in turn significantly affects vegetation productivity. Most of the previous studies on the subject have focused on links between the El Niño-Southern Oscillation (ENSO) and vegetation productivity, but few studies have addressed the effects of West Pacific Warm Pool (WPWP) on that although the early stages of the ENSO phenomenon may first develop there. In this paper, we use the mean SST values in the WPWP to construct a climate index, known as the WPWP index (WPI), and study the impacts of the WPWP on global vegetation productivity. We provide evidence for a robust link among the alternating warm and cool WPI pattern, terrestrial vegetation productivity and carbon balance. The analysis is based on both satellite observations and model simulations. The results of this study show that the warm and cool WPWP phases have inverse effects on land surface temperature and precipitation. A warm (cool) WPWP is associated with a warmer (cooler) climate on global land surfaces as well as a drier (wetter) climate in southern hemisphere, and hence enhances (suppresses) vegetation productivity in the latitudes of approximately 10–70°N and suppresses (enhances) vegetation growth in the latitudes of approximately 10–30°S. The underlying mechanism is also discussed. The WPI serves as a meaningful climate index for studying the ocean-vegetation teleconnections. © 2018 Elsevier B.V. All rights reserved.

⁎ Corresponding author at: Room 3423, Datun Road A11, Chaoyang District, Beijing 100101, China. E-mail address: [email protected] (M. Huang). 1 The first two authors contributed equally to this work and should be considered co-first authors.

https://doi.org/10.1016/j.scitotenv.2018.11.170 0048-9697/© 2018 Elsevier B.V. All rights reserved.

642

M. Huang et al. / Science of the Total Environment 655 (2019) 641–651

1. Introduction Global climate change has significantly impacted vegetation productivity over the past century (de Jong et al., 2013; Wu et al., 2015; Zhao and Running, 2010). It is important to understand the responses of vegetation to climate change since vegetation plays a key role in the global energy balance as well as in hydrological and biogeochemical cycles (Buermann et al., 2003). Temperature (T) and precipitation (P) are the two main factors that affect vegetation productivity (Bonan, 2008; Nemani et al., 2003). Previous studies have shown that the variations in land T and P may be predictable in terms of sea surface T (SST) and sea surface pressure (Hoerling et al., 2001; Schubert et al., 2009; Ting et al., 1996). Therefore, the variations in SST and sea surface pressure may also influence vegetation productivity. The influence of ocean surface patterns on the continental climate is realized through teleconnections (Glantz et al., 1991). The global ocean-atmospheric system has several important internal variability patterns, such as those associated with El Niño-Southern Oscillation (ENSO), which affects the atmospheric circulation and climatic patterns worldwide. Understanding these teleconnection patterns is crucial for studying the influence of oceanic processes on vegetation productivity on land. Previous studies have identified the teleconnection of SSTs to land surface processes at global and regional scales (Buermann et al., 2003; Hashimoto et al., 2004; Huber and Fensholt, 2011; Iguchi, 2011; Keeling et al., 1995; Los et al., 2001; Potter et al., 2008, 2004, 2003; Wharton et al., 2009). Climate indices are originally developed to depict ocean-atmosphere mechanism, and they are currently widely used to study oceanatmosphere-vegetation teleconnections (Gonsamo et al., 2016; Li et al., 2016; Zhu et al., 2017). These climate indices can be derived simply from SSTs or a combination of several climate variables, such as air T, pressure or surface winds (Huber and Fensholt, 2011). Climate indices associated with ENSO and their relationships with vegetation production and terrestrial carbon balance have been intensively studied (Battle et al., 2000; Cadson et al., 1996; Cobb et al., 2003; Fang et al., 2017; Foley et al., 2002; McPhaden et al., 2006; Oba et al., 2001; Ropelewski and Halpert, 1986, 1987; Tian et al., 1998; Woodward et al., 2008). Other climate indices, such as the North Atlantic Oscillation (NAO), Pacific Decadal Oscillation (PDO) and Indian Ocean Dipole (IOD) are used either independently or combined with ENSO indices in the studies of vegetation and ecosystem carbon cycle response to climate change (Huber and Fensholt, 2011; Potter et al., 2003; Williams and Hanan, 2011). Although several studies revealed the teleconnections between climate indices and regional NDVI variability (Gong and Shi, 2003; Woodward et al., 2008), less attention has been given to the globalscale ocean, climate and NDVI teleconnections. Gonsamo et al. (2016) reported that ENSO has weak control on global-scale vegetation productivity, although its continental scale responses are substantial. Zhu et al. (2017) analyzed the effects of 15 major teleconnections on terrestrial ecosystem carbon fluxes and concluded that these climate indices had only regional influences on the vegetation productivity in the Northern Hemisphere. Therefore, it is necessary to develop new climate indices which could connect ocean, climate and vegetation productivity on global-scale. ENSO is a leading mode of tropical climate variability at interannual timescales. Many studies have been made to understand the influence of ENSO on global scale vegetation productivity. However, less efforts have been made to explore the effects of the West Pacific Warm Pool (WPWP) on global vegetation productivity, though the early stages of the ENSO phenomenon may first develop there (Cane, 1983; Yan et al., 1992, 1997). The WPWP is the largest warm water air mass dominating the western tropical Pacific, which is in the global maximum centre of vertically integrated diabatic heating in the troposphere (Masuda, 1984). The WPWP is a monolithic heat source to the atmosphere and experiences the greatest rainfall of any oceanic region on Earth. Previous studies demonstrated that small changes in the

magnitude of the SST in the WPWP can result in great variations in global atmospheric circulation (Karnauskas and Busalacchi, 2009; Palmer and Mansfield, 1984). Therefore, SST anomalies in WPWP may have a great influence on global vegetation. However, the teleconnections between the WPWP and land climate, as well as the associated impacts on vegetation productivity, have not been well studied. In this study, we introduce the WPWP index (WPI) and investigate the global teleconnections between the WPI and normalized difference vegetation index (NDVI), as well as explore the responsible mechanisms. Since NDVI is often used as a surrogate for primary productivity of the terrestrial biosphere (Myneni et al., 1997), we use the growing season aggregated NDVI to represent vegetation productivity in this study. NDVI has been widely used for studying the links between climate indices and regional vegetation variations (Anyamba et al., 2002; Gong and Shi, 2003; Jarlan et al., 2005; Li et al., 2016; Li and Kafatos, 2000; Nicholson et al., 1998). Further, we use model simulated net primary productivity (NPP) and net ecosystem productivity (NEP) for comparison to the vegetation responses to different WPI phase. These carbon cycle components are obtained from Earth system model (ESM) simulations, which are part of the Coupled Model Intercomparison Project Phase 5 (CMIP5). We use ESM simulations 1) to evaluate the responses of NPP and NEP to different WPI phase and 2) to compare the responses of NPP with NDVI for different WPI phase and evaluate whether the NPP responses are consistent with that of NDVI. The objectives of this study are 1) to evaluate the value of the WPI as a climate index for studying WPWP SSTs and global vegetation teleconnections, 2) to gain a better understanding of the mechanisms underlying the teleconnections and 3) to verify to what degree the WPWP SSTs influence the spatial patterns of global vegetation productivity as well as the carbon balance of terrestrial ecosystems. 2. Data sources and methods 2.1. Data sources The NDVI3g datasets, which are produced by the Global Inventory Monitoring and Modeling Studies (GIMMS) group using AVHRR/NOAA series satellites, (https://ecocast.arc.nasa.gov/data/pub/gimms/3) are used for the analysis. The data, which is in 15-day intervals, covers the period from 1982 to 2015 at a spatial resolution of 8 km2. The maximum value composite (MVC) method is used to derive the monthly NDVI based on two images per month. Pixels with monthly NDVI values of less than 0.1 were considered non-vegetated areas and removed. In the vegetated areas, monthly NDVI values of less than 0.1 were considered to occur in the nongrowing season and were removed. The annual sum of NDVI values was used to study the NDVI responses to WPI. The monthly T and P data, which have a spatial resolution of 0.5°, are obtained from the Climate Research Unit (CRU) (version TS4.0) (Harris et al., 2014). The monthly mean land surface pressure, vertically integrated water vapour and 850 hPa vector wind are obtained from the European Centre for Medium Range Weather Forecasts Interim reanalysis dataset (ERA-Interim), which has a spatial resolution of approximately 80 km (T255 spectral). The monthly mean SSTs are obtained from the NOAA Earth System Research Laboratory (ESRL) Physical Sciences Division (PSD), Boulder, Colorado, USA ((Hirahara et al., 2014); http://www.esrl.noaa.gov/psd/). The monthly precipitation is accumulated as annual total precipitation, P, and other monthly climate factors are averaged over a year as annual climate factors for our analysis. The Extended Multivariate ENSO Index (MEI) was obtained from the NOAA/OAR/ESRL PSD, Boulder, CO, USA (https://www.esrl.noaa.gov/ psd/enso/past_events.html). 2.2. ESMs The main features of the six ESMs used in this study are summarized in Table 1. BCC-CSM1-1 is a global, fully coupled climate-carbon cycle

M. Huang et al. / Science of the Total Environment 655 (2019) 641–651

643

Table 1 Main features of the ESMs used in this study. Model name

Land resolution

Land surface component

Dynamic vegetation cover

Reference

BCC_CSM1-1 BNU-ESM CanESM2 IPSL-CM5A-LR MIROC-ESM MRI-ESM1

2.8 × 2.8° 2.8 × 2.8° 2.8 × 2.8° 3.75 × 1.875° 2.8 × 2.8° 1.125 × 1.125°

BCC-AVIM1.0 CLM CTEM ORCHIDEE SEIB-DGVM Land carbon modules

No Yes No No Yes Yes

Wu et al. (2013) Ji et al. (2014) Christian et al. (2010) Dufresne et al. (2013) Watanabe et al. (2011) Adachi et al. (2013)

model developed by the Beijing Climate Center and its land surface component BCC-AVIM1.0, which is an integration of the AtmosphereVegetation Interaction Model (AVIM) with the biogeophysical frame of NCAR/CLM3 (Wu et al., 2013). BNU-ESM was developed by Beijing Normal University and its land surface component Common Land Model (CLM) is used to simulate terrestrial ecosystem processes (Ji et al., 2014). CanESM2 is the new version of the Canadian ESM (CanESM1), which was developed by the Canadian Centre for Climate Modelling and Analysis (CCCma). The Canadian Terrestrial Ecosystem Model (CTEM) is used in the land surface component of CanESM2 to model terrestrial ecosystem processes. IPSL-CM5A-LR was developed by the Institute Pierre Simon Laplace (IPSL) in France, which is used to study the long-term response of the climate system to natural and anthropogenic forcing. The land surface component of the IPSL-CM5A-LR is the Organizing Carbon and Hydrology in Dynamic Ecosystems (ORCHIDEE), which simulates the energy and water cycles of soil and vegetation, the terrestrial carbon cycle, and the vegetation composition and distribution (Dufresne et al., 2013). MIROC-ESM was cooperatively developed by the University of Tokyo's National Institute for Environmental Studies (NIES) and the Japan Agency for Marine-Earth Science and Technology (JAMSTEC). The land surface component of MIROCESM is a terrestrial ecosystem component that deals with dynamic vegetation (SEIB-DGVM) (Watanabe et al., 2011). MRI-ESM1 was developed at the Meteorological Research Institute of Japan and its land carbon cycle module simulates the energy, water cycle and terrestrial ecosystem carbon cycle as well as dynamically representing the global vegetation system (Adachi et al., 2013). We obtained the CMIP5 multi-model dataset through the Program for Climate Model Diagnosis and Intercomparison (PCMDI) at http:// pcmdi9.llnl.gov/esgf-web-fe/. ESMs were selected based on the availability of monthly NPP and NEP outputs for the historical simulation. The historical simulation is referred to as the twentieth-century simulations and is suitable for comparison with observations. The monthly mean NPP and NEP from 1982 to 2005 are first accumulated over an annual interval to obtain the annual NPP and NEP values, and then they are spatially interpolated to a 0.5° resolution. 2.3. Definition of WPI WPWP is usually defined as a warm water body with SSTs warmer than 28–29 °C (Hu et al., 2017). In this study, we first calculate the annual mean SST averaged over 1982–2015 in West Pacific Ocean, and then use the 29° isotherm of the annual mean SST as the threshold to define the domain of WPWP. The monthly mean SST averaged over WPWP is used to construct WPI. Using the monthly mean WPWP SST anomaly, a WPI time series from 1982 to 2015 is created by removing the annual cycle and linear trend, standardizing (by dividing by its standard deviation), and smoothing with a 5-month running mean. In analogy to the definition of ENSO and Indo-Pacific warm pool indices (Trenberth, 1997; Zhou et al., 2017), we propose a threshold of 0.5, which is equal to half the standard deviation of the WPI series, to be used as the indictor of the warm phase (WP) for WPWP. A WP year is classified as a year with a 5-month running mean of monthly WPI anomalies that exceed the threshold for at least 6 consecutive months. A complementary definition, which uses a threshold of −0.5, is used for the WPWP cold phase (CP). The WP years detected include 1987,

1990, 1991, 1994, 1995, 1996, 2001, 2003, 2007, 2010, 2013 and 2014, and the CP years detected include 1984, 1989, 1992, 1993, 1997, 1999, 2000, 2004, 2008, 2009, 2011 and 2012. The WPI time series is displayed in Fig. 1a. The WPWP events are more like Indo-Pacific warm pool events, which can last from one to three years based on the WPI anomalies; this is in contrast to ENSO events, which peak in boreal winter and decay in boreal spring. Fig. 1b and c show the location of the defined WPWP and the WP and CP composite SST anomalies, respectively. 2.4. Statistical analysis Composite analysis is capable of reducing case-to-case variability and allows for a highly realistic representation of specific characteristics, and it helps to effectively characterize climatic conditions for particular cases (Rudeva and Gulev, 2011). In this study, a composite analysis is conducted to timely average variables, including NDVI, NPP, NEP, T, P, 500 hPa geopotential height, 850 hPa vector wind and water vapour fluxes, for the WP and CP years, respectively. Correlation analysis is used to investigate the relationships between WPI and other variables, such as NDVI, T and P. 3. Results 3.1. Impacts of WPI phase on circulation Generally, a warm SST is associated with stronger ascending motion, which results in negative surface pressure and positive rainfall anomalies. The global spatial distribution of composite surface pressure anomalies for the WP is consistent with this understanding (Fig. 2a). In the WP, the surface pressure over the Pacific Ocean is decreased, while positive surface pressure anomalies occur over the southern high latitudes and most areas of the northern high latitudes, which indicates a weakening of the polar vortex. The weakness of the polar vortex is associated with the positive air T anomalies. In the CP, positive surface pressure anomalies occupy the central and eastern Pacific Ocean, and negative anomalies dominate the Antarctic and most areas of the Arctic (Fig. 2b). The strengthening of the polar vortex implies meridian circulation enhancement, which results in decreased T. In the western Pacific and Indian Oceans, the sign for the spatial pattern of composited integrated water vapour anomalies for the WP is opposite to that for the CP. In the WP, the vertically integrated water vapour flux anomalies are positive over the western Pacific and Indian Oceans, with maximum anomalies present in the domain of the WPWP (Fig. 2c). In the CP, negative vertically integrated water vapour flux anomalies dominate most areas of the intertropical convergence zone (ITCZ), and the positive anomalies dominate Australia and its surrounding ocean (Fig. 2d). ITCZ is the area where winds originating from the Northern Hemisphere (NH) and Southern Hemisphere (SH) combine. The variation in ITCZ greatly influences P in tropical areas as well as in the middle and high latitudes because tropical moisture export, which transports moisture from the tropical oceans poleward through a variety of dynamic processes, is an important part of the global water and energy budgets (Knippertz et al., 2013). Opposite WP and CP water vapour flux spatial distributions indicate opposing P patterns for both phases on global land surfaces.

644

M. Huang et al. / Science of the Total Environment 655 (2019) 641–651

Fig. 1. (a) The WPI time series from 1982 to 2005; composite SST anomalies (°C) for (b) WP and (c) CP. The 29° isotherm for the annual mean SST averaged over 1982–2015 is used to define the WPWP domain.

3.2. Impacts of circulation on T and P The composite T pattern is generally consistent with that of the surface pressure, and the composite P pattern is consistent with that of the vertically integrated water vapour flux for both WP and CP (Fig. 3). The combination of T and P in the two phases are nearly opposite. Most of the global land surface is covered by positive T anomalies in the WP and negative T anomalies in the CP. North Africa and Asia were dominated by positive P and T anomalies during WP years, and these areas were occupied by negative P and T anomalies during CP years, which indicates a warmer, wetter climate during the WP and a cooler, drier climate during the CP. In the SH, most areas were covered by positive T and negative P anomalies during the WP and negative T and positive P anomalies during the CP, which indicates a warmer, drier climate during the WP and a cooler, wetter climate during the CP. The T anomalies in Europe, part of North America, the southern part of South Africa, Peru and Argentina were different from those of other global land surface area. These areas were covered by negative T anomalies during the WP and positive T anomalies during the CP. 3.3. Impacts of WPI phase on NDVI A composite analysis of NDVI reveals that positive NDVI anomalies occurred in most areas of Russia, United States, eastern China, and

northern South Africa during the WP; however, negative NDVI anomalies occupied the above stated areas during the CP. Meanwhile, negative NDVI anomalies occupied southern South Africa, most of Australia and northern South America during the WP, but positive NDVI anomalies dominated these areas during the CP (Fig. 4). Correlation analysis of association between NDVI and WPI is generally consistent with the results of composite analysis. For example, in Russia, eastern China and United States, where positive and negative NDVI anomalies occupied in WP and CP, respectively, WPI is positively correlated with NDVI. In most areas of Australia and southern South Africa, where negative and positive NDVI anomalies occupied in WP and CP, respectively, WPI is negatively correlated with NDVI (Fig. 5). Both composite and correlation analysis reveal a strong association between the NDVI and WPI phases, which implies the teleconnections between WPI and vegetation productivity could be observed by satellite observations. To assess the impacts of WPI on vegetation type, such as forests, herbaceous, shrubs and crops, we first define climate zones and then composite analysis the NDVI anomalies averaged over vegetation types in each climate zone. The climate zones are defined as boreal (BO) between 66.5°N to 90°N, northern temperate (NT) between 23.5°N to 66.5°N, northern tropic (NTO) between 0° to 23.5°N, southern tropic (STO) between 0°to 23.5°S and southern temperate (ST) between 23.5°S to 66.5°S.

M. Huang et al. / Science of the Total Environment 655 (2019) 641–651

645

Fig. 2. Spatial distribution of composite annual mean surface pressure anomalies (in hPa) for (a) WP and (b) CP. Spatial distribution of composite annual mean vertically integrated water flux anomalies (in kg m−2) and annual mean vector wind (in m s−1) anomalies at 850 hPa for (c) WP and (d) CP.

Fig. 3. Signal composites of annual mean T and annual total P anomalies for (a) WP and (b) CP. For example, “T−P−” corresponds to negative anomalies in both T and P.

646

M. Huang et al. / Science of the Total Environment 655 (2019) 641–651

Fig. 4. Composite spatial distribution of annual mean NDVI anomalies for (a) WP and (b) CP.

Fig. 5. Spatial distribution of correlation coefficients between WPI and NDVI. The thresholds of 0.34 and 0.44 denote 0.05 and 0.01 significance levels, respectively.

M. Huang et al. / Science of the Total Environment 655 (2019) 641–651 Table 2 Composite analysis of NDVI anomaly (%) averaged over forests, shrubs, herbaceous and crops in climate zones. NDVI (%) Forests Shrubs Herbaceous Crops

WP CP WP CP WP CP WP CP

BO

NT

NTO

STO

ST

0.71 0.56 0.79 0.72 0.69 0.55 0.38 0.24

0.59 −0.04 0.68 −0.36 0.58 −0.15 0.54 0.17

0.28 0.15 0.58 0.49 0.32 −0.01 0.44 0.18

−0.06 0.17 −0.45 1.14 −0.39 1.02 −0.29 0.34

0.28 0.21 −0.21 1.06 −0.24 1.04 0.41 0.23

Table 2 shows the NT is the most sensitive climate zone to WPI variations in the Northern Hemisphere, where NDVI anomalies averaged over forests, shrubs and herbaceous are 0.59%, 0.68% and 0.58% in WP, and −0.04%, −0.36% and −0.15% in CP, respectively. NDVI anomalies averaged over crops are 0.54% and 0.17% for WP and CP, respectively. Unlike the negative anomalies for other vegetation types in CP, crops' NDVI anomalies in CP are positive, which indicates the influence of human activities on crops. The STO is the most sensitive climate zone to WPI variations in the Southern Hemisphere, where NDVI anomalies averaged over forests, shrubs, herbaceous and crops are −0.06%, −0.45%, −0.39% and −0.29% in WP and 0.17%, 1.14%, 1.02% and 0.34% in CP, respectively. Shrubs is the most sensitive vegetation to WPI variations, and its averaged NDVI anomalies are the highest among other vegetation type means in a given climate zone. 3.4. Impacts of WPI phase on NPP and NEP We analyzed the composite NPP and NEP from the CMIP5 ESMs. The ensemble mean NPP is the output average of 6 models, which include

647

BCC_CSM1_1, BNU-ESM, CanESM2, IPSL-CM5A-LR, MIROC-ESM and MRI-ESM1. The ensemble mean NEP is the output average of 4 models, which include BNU-ESM, CanESM2, IPSL-CM5A-LR and MRI-ESM1. Since NPP and NEP from the CMIP5 ESMs are not available after 2005, the detected WP and CP years after 2005 are excluded from the composite analysis. The composite spatial distribution of NPP shows that most global areas are covered by positive NPP anomalies during the WP and negative NPP anomalies during the CP (Fig. 6a and b). NEP is the difference between NPP and heterotrophic respiration (Rh), and a positive (negative) NEP corresponds to a sink of CO2 (source) to the atmosphere. The spatial pattern of the ensemble mean NEP anomalies for the WP and CP are of opposite signs (Fig. 6c and d). For example, Australia, northern South America and parts of South Africa are occupied by positive NEP anomalies during the WP and negative NEP anomalies during the CP. Europe and most areas of North America are dominated by positive NEP anomalies during the WP and negative NEP anomalies during the CP. 4. Discussion 4.1. Linking NDVI variations with WPI The latitudinal distribution of composite NDVI anomalies for WP and CP are shown in Fig. 7a. The signs of NDVI anomalies for WP and CP are positive and negative at most latitudes of the NH, and the averaged NDVI anomalies over 10–70°N are 0.71% and −0.02% for WP and CP, respectively. The NDVI anomalies in the latitudes between 10 and 70°N are generally explained by the corresponding T variations in the WP and CP. As both NDVI and T anomalies are positive in WP and negative in CP (Fig. 7a), which implies higher T corresponding to higher NDVI, and the correlation analysis also reveals a positive correlation between NDVI and T (Fig. 8a), which is consistent with the results of composite analysis.

Fig. 6. Composite spatial distribution of ensemble mean NPP (g C m−2 a−1) anomalies from 6 CMIP5 ESMs for (a) WP and (b) CP. Composite spatial distribution of ensemble mean NEP (g C m−2 a−1) from 4 CMIP5 ESMs for (c) WP and (d) CP.

648

M. Huang et al. / Science of the Total Environment 655 (2019) 641–651

70

(b)

(a)

60

WP CP

50 40

Latitude (o)

30 20 10 0 -10 -20 -30 -40

-4

0

NDVI (%)

70

4

-0.25

0.25

0

70

(d)

(C)

60

0.00

T(oC)

El La

50 40

Latitude (o)

30 20 10 0 -10 -20 -30 -40

-4

0

NDVI (%)

4

-70

P (mm)

Fig. 7. Comparison of the latitudinal distributions of composite variables. (a) NDVI anomalies for WP and CP; (b) T anomalies for WP and CP; (c) NDVI anomalies for El Niño and La Niña and (d) P anomalies for WP and CP. El and Na represent El Niño and La Niña, respectively. The composite analysis is based on data from 1982 to 2015.

In terms of the effects of P on NDVI, the composite analysis shows P anomalies are positive in WP and negative in CP in most latitudes of 10–70°N (Fig. 7d). As NDVI anomalies are positive in WP and negative in CP, which implies higher P corresponding to higher NDVI. However, correlation analysis reveals P is negatively correlated with NDVI in most areas of 10–70°N (Fig. 8b), which is contrast to the results of composite analysis. The small P differences between WP and CP and the insignificant correlations between P and NDVI were anticipated for the sources of the contradiction. The mean P anomalies averaged over 10–70°N are

3.6 mm and −3.2 mm for WP and CP, respectively, and that averaged over 10–30°S are −18 mm and 36 mm for WP and CP, respectively, which indicates the P difference between WP and CP in 10–70°N is small, and the uncertainties in P observations may cause the contradiction. As the samples used for correlation analysis are 34, which are form 1982–2015, the correlation coefficients of 0.34 and 0.44 denote 0.05 and 0.01 significance levels, respectively. As shown in Fig. 8a, parts of the correlation coefficients are insignificant in the latitudes between 10 and 70°N, and hence the weak correlations between P and NDVI may cause the contradiction.

Fig. 8. Spatial distribution of correlation coefficient between (a) NDVI and T and (b) NDVI and P.

M. Huang et al. / Science of the Total Environment 655 (2019) 641–651

Although the effects of P on NDVI are complex and need further studying, the effects of T on NDVI are significant, which implies the NDVI are mainly control by T in the latitudes of 10–70°N. The NDVI anomalies in the latitudes of 10–30°S are negative and positive for WP and CP, respectively (Fig. 7a), and the average NDVI anomalies for WP and CP are −0.38% and 0.9%, respectively. The NDVI variations are generally explained by the corresponding T and P variations in the WP and CP. Correlation analysis shows NDVI are negatively correlated with T and positively correlate with P in this zone (Fig. 8), which implies positive NDVI anomalies corresponding to negative T and positive P anomalies, and negative NDVI anomalies with positive T and negative P anomalies. Composite analysis shows T and P anomalies are positive and negative for WP, and negative and positive for CP, respectively, which is consistent with the results of correlation analysis. Therefore, both positive P and negative T anomalies in CP are in favor of vegetation growth, while negative P and positive T anomalies in WP suppress vegetation growth in this zone. The variation of NDVI is controlled by both T and P.

4.2. Comparison of NPP and NDVI responses to WPI phase Although NDVI is not measured NPP, the growing season aggregated NDVI indicates the rate of organic biomass growth and accumulation by plants, and represents vegetation productivity. Therefore, the responses of NDVI and NPP to WPI phase should have similar characteristics. The latitudinal distributions of the composite NPP anomalies for WP and CP simulated by the six ESMs are displayed in Fig. 9. Although there are differences between model simulated NPP, all model simulated NPP anomalies for WP are opposite to CP in all latitudes. All NPP averaged over 10–70°N are 1.22% and −1.0% for WP and CP, respectively, which are consistent with NDVI anomalies for the two phases. This further confirms the impacts of WPI phase on NDVI. In the latitudes between 10 and 30°S, the NDVI anomalies are negative for WP and positive for CP, which are reverse to that in the NH. The NPP anomalies simulated by BCC_CSM1-1 and MIROC-ESM between 20 and 40°S, and that simulated by MRI-ESM1 between 10 and 20°S are coherent with the variations of NDVI anomalies. Other NPP anomalies simulated by BNU-ESM, CanESM2 and IPSL-CM5A-LR are positive for WP and negative for CP between the latitudes of 10–30°S, which are opposite to the NDVI anomalies. Since the NDVI anomalies are coherent with the P anomalies in these latitudes, the discrepancy was anticipated for the NPP uncertainties for these ESMs. 70 60 50

Latitude (o)

40

(a)

4.3. Comparison of WPI impacts with ENSO impacts Since the extreme SST variations in WPWP usually indicate the early stage of the ENSO phenomenon, it is necessary to compare the WPI impacts with ENSO impacts. Based on the definition of MEI, there are 10 El Niño years and 6 La Niña years during the study period. The El Niño years are 1983, 1987, 1988, 1992, 1995, 1998, 2003, 2007, 2010 and 2015, and the La Niña years are 1989, 1991, 2000, 2008, 2011 and 2012. As described in Section 2.3, there are 12 WP and CP years during the study period. The years of 1987, 1995, 2003, 2007 and 2010 belong to both El Niño and WP years, and the years of 1989, 2008, 2011 and 2012 belong to both La Niña and CP years. Therefore, WP years contain most of the El Niño years and CP years contain most of the La Niña years, and WPI events occurred more frequently than ENSO events in the study period. By comparison of the composite NDVI for El Niño and La Niña (Fig. 7c) with that for WP and CP years (Fig. 7a), it is found that the major difference between WPI and ENSO impacts on NDVI mainly occurred in the north of 10°N, where all NDVI anomalies in WP are higher than that in CP, but that are not in the case of El Niño and La Niña. In the Southern Hemisphere, the latitudinal distribution of composite NDVI anomalies for WP are similar as that for El Niño, and for CP as for La Niña, though the minimum NDVI anomalies for WP, −0.8%, are lower than that for El Niño, −1.84%, and the maximum NDVI anomalies for CP, 1.5%, are also lower than that for La Niña, 3.16%. This is reasonable because WP and CP years contain most of the El Niño and La Niña years during the study period. ENSO is the largest climate oscillation in the world and its effects on global terrestrial ecosystems have been demonstrated to be the most significant among all climate indices (Zhu et al., 2017). However, we demonstrate here the impacts of ENSO on NDVI mainly occurred in the Southern Hemisphere. Previous studies investigated the effects of teleconnections on vegetation productivity found little spatial correlation between climate indices and terrestrial ecosystem variables across large areas (Buermann et al., 2003; Gonsamo et al., 2016; Stenseth et al., 2003), especially in the Northern Hemisphere where each climate index explains relatively small fractions of atmospheric variability in a given region (Comas-Bru and McDermott, 2014). The WPI developed in this study, which explains the relationship between climatic patterns and vegetation productivity, could fill in gaps in the studying of teleconections between SSTs and vegetation productivity in the Northern Hemisphere. However, it should be noted that our present annual time scale analyses would limit the interpretation of the effects of teleconnections on seasonal vegetation growth. Future multi-time scale analyses are necessary to improve our knowledge of such effects on vegetation productivity.

(b) BCCBNUCan-

BCC+ BNU+ Can+

IPSLMIROCMRI-

IPSL+ MIROC+ MRI+

30 20 10 0

-10 -20 -30 -40 -15

649

0

NPP凚 %凛

15 -15

0

15

NPP (%)

Fig. 9. Comparison of the latitudinal distributions of composite NPP anomalies between WP and CP (a) for BCC_CSM1-1, BNU-ESM and CanESM2 and (b) for IPSL-CM5A-LR, MIROC-ESM and MRI-ESM1. The symbols + and − denote WP and CP, respectively. The composite NPP is based on data from 1982 to 2005.

5. Conclusion In this study, we defined the WPI climate index to describe the SST variations in the warmest ocean on Earth. We also investigated the teleconnections between the WPWP SSTs and globe vegetation productivity as well as the terrestrial ecosystem carbon balance. We found the NDVI values to be higher in the WP and lower in the CP for most of the land surface in the NH, and to be lower in the WP and higher in the CP in latitudes between 10 and 30°S in the SH. The mechanism for the teleconnections is as follows. On the one hand, warm WPWP SSTs are associated with strong ascending motion, which decreases tropical surface pressure and increases surface pressure in the high latitudes; conversely, cool SSTs are associated with weakened ascending motion, which increases tropical surface pressure and decreases surface pressure in the high latitudes. Air pressure changes in the WP and CP result in increasing and decreasing global land surface T values, respectively. On the other hand, warm and cool SSTs have different water vapour flux spatial patterns over the ITCZ, which causes different P patterns for the continents in the WP and CP. Warmer SSTs cause a warmer

650

M. Huang et al. / Science of the Total Environment 655 (2019) 641–651

climate on the Earth and a drier climate in the SH; conversely, cooler SSTs cause a cooler climate on the Earth and a wetter climate in SH. The warmer climate in the middle and high northern latitudes (approximately 10–70°N) of the WP promotes vegetation growth, whereas a cooler climate in the CP suppresses the vegetation growth in that area. In the lower and middle latitudes (approximately 10–30°S) of the SH, a warmer and drier climate in the WP suppress vegetation growth, whereas a cooler and wetter climate in the CP promotes vegetation growth. Almost all model simulated NPP pattern for the two phases are consistent with the NDVI pattern in the NH, and three ESMs simulated NPP in the southern lower latitudes (approximately 10–30°S) are generally consistent with the NDVI pattern. The discrepancy was anticipant from the uncertainties of NPP. The ensemble mean NEP anomalies indicate opposing spatial distributions for the WP and CP, which suggests WPWP SSTs also influence the global carbon balance. In summary, the WP corresponding to a greener land surface in the north of 10°N, and a browner land surface between 10 and 30°S, and the CP corresponding to a browner land surface in the north of 10°N, and a greener land surface between 10 and 30°S. The WPI serves as a meaningful climate index for studying the teleconnections between WPWP and global vegetation productivity.

Acknowledgements This study is supported by the National Key Research and Development Program of China (2017YFC0503905) and the National Natural Science Foundation of China (No. 41671101). All of the authors have no conflict of interest to declare. References Adachi, Y., et al., 2013. Basic performance of a new Earth system model of the Meteorological Research Institute (MRI-ESM1). Pap. Meteorol. Geophys. 64, 1–19. Anyamba, A., Tucker, C.J., Mahoney, R., 2002. From El Niño to La Niña: vegetation response patterns over east and southern Africa during the 1997–2000 period. J. Clim. 15, 3096–3103. Battle, M., Bender, M., Tans, P.P., White, J., Ellis, J., Conway, T., Francey, R., 2000. Global carbon sinks and their variability inferred from atmospheric O2 and δ13C. Science 287, 2467–2470. Bonan, G.B., 2008. Forests and climate change: forcings, feedbacks, and the climate benefits of forests. Science 320, 1444–1449. Buermann, W., Anderson, B., Tucker, C.J., Dickinson, R.E., Lucht, W., Potter, C.S., Myneni, R.B., 2003. Interannual covariability in Northern Hemisphere air temperatures and greenness associated with El Niño-Southern Oscillation and the Arctic Oscillation. J. Geophys. Res. Atmos. 108. Cadson, R., Todey, D.P., Taylor, S.E., 1996. Midwestern corn yield and weather in relation to extremes of the Southern Oscillation. J. Prod. Agric. 9, 347–352. Cane, M.A., 1983. Oceanographic events during El Niño. Science 222, 1189–1195. Christian, J.R., Arora, V.K., Boer, G.J., Curry, C.L., Zahariev, K., Denman, K.L., Flato, G., Lee, W. G., Merryfield, W.J., Roulet, N., Scinocca, J.F., 2010. The global carbon cycle in the Canadian Earth system model (CanESM1): preindustrial control simulation. J. Geophys. Res. 115, G03014. https://doi.org/10.1029/2008JG000920. Cobb, K.M., Charles, C.D., Cheng, H., Edwards, R.L., 2003. El Niño/Southern Oscillation and tropical Pacific climate during the last millennium. Nature 424, 271. Comas-Bru, L., McDermott, F., 2014. Impacts of the EA and SCA patterns on the European twentieth century NAO-winter climate relationship. Q. J. R. Meteorol. Soc. 140, 354–363. de Jong, R., Schaepman, M.E., Furrer, R., De Bruin, S., Verburg, P.H., 2013. Spatial relationship between climatologies and changes in global vegetation activity. Glob. Chang. Biol. 19, 1953–1964. Dufresne, J.-L., et al., 2013. Climate change projections using the IPSL-CM5 Earth system model: from CMIP3 to CMIP5. Clim. Dyn. 40, 2123–2165. Fang, Y., et al., 2017. Global land carbon sink response to temperature and precipitation varies with ENSO phase. Environ. Res. Lett. 12 (064007). Foley, J.A., Botta, A., Coe, M.T., Costa, M.H., 2002. El Niño–southern oscillation and the climate, ecosystems and rivers of Amazonia. Glob. Biogeochem. Cycles 16. Glantz, M.H., Katz, R.W., Nicholls, N., 1991. Teleconnections Linking Worldwide Climate Anomalies. vol. 535. Cambridge University Press, Cambridge. Gong, D.-Y., Shi, P.-J., 2003. Northern hemispheric NDVI variations associated with largescale climate indices in spring. Int. J. Remote Sens. 24, 2559–2566. Gonsamo, A., Chen, J.M., Lombardozzi, D., 2016. Global vegetation productivity response to climatic oscillations during the satellite era. Glob. Chang. Biol. 22, 3414–3426. Harris, I., Jones, P., Osborn, T., Lister, D., 2014. Updated high-resolution grids of monthly climatic observations–the CRU TS3. 10 Dataset. Int. J. Climatol. 34, 623–642.

Hashimoto, H., et al., 2004. El Niño–Southern Oscillation–induced variability in terrestrial carbon cycling. J. Geophys. Res. Atmos. 109. Hirahara, S., Ishii, M., Fukuda, Y., 2014. Centennial-scale sea surface temperature analysis and its uncertainty. J. Clim. 27, 57–75. Hoerling, M.P., Hurrell, J.W., Xu, T., 2001. Tropical origins for recent North Atlantic climate change. Science 292, 90–92. Hu, S., Hu, D., Guan, C., Xing, N., Li, J., Feng, J., 2017. Variability of the western Pacific warm pool structure associated with El Niño. Clim. Dyn. 49, 2431–2449. Huber, S., Fensholt, R., 2011. Analysis of teleconnections between AVHRR-based sea surface temperature and vegetation productivity in the semi-arid Sahel. Remote Sens. Environ. 115, 3276–3285. Iguchi, T., 2011. Correlations between interannual variations of simulated global and regional CO2 fluxes from terrestrial ecosystems and El Niño Southern Oscillation. Tellus Ser. B Chem. Phys. Meteorol. 63, 196–204. Jarlan, L., Tourre, Y., Mougin, E., Philippon, N., Mazzega, P., 2005. Dominant patterns of AVHRR NDVI interannual variability over the Sahel and linkages with key climate signals (1982–2003). Geophys. Res. Lett. 32. Ji, D., et al., 2014. Description and basic evaluation of Beijing Normal University Earth system model (BNU-ESM) version 1. Geosci. Model Dev. 7, 2039–2064. Karnauskas, K.B., Busalacchi, A.J., 2009. Mechanisms for the interannual variability of SST in the East Pacific warm pool. J. Clim. 22, 1375–1392. Keeling, C.D., Whorf, T.P., Wahlen, M., Van der Plichtt, J., 1995. Interannual extremes in the rate of rise of atmospheric carbon dioxide since 1980. Nature 375, 666. Knippertz, P., Wernli, H., Gläser, G., 2013. A global climatology of tropical moisture exports. J. Clim. 26, 3031–3045. Li, Z., Kafatos, M., 2000. Interannual variability of vegetation in the United States and its relation to El Niño/Southern Oscillation. Remote Sens. Environ. 71, 239–247. Li, J., Fan, K., Xu, Z., 2016. Links between the late wintertime North Atlantic Oscillation and springtime vegetation growth over Eurasia. Clim. Dyn. 46, 987–1000. Los, S.O., Collatz, G.J., Bounoua, L., Sellers, P.J., Tucker, C.J., 2001. Global interannual variations in sea surface temperature and land surface vegetation, air temperature, and precipitation. J. Clim. 14, 1535–1549. Masuda, K., 1984. Diabatic heating during the FGGE: a preliminary report. J. Meteorol. Soc. Jpn. 62, 702–708 (Ser. II). McPhaden, M.J., Zebiak, S.E., Glantz, M.H., 2006. ENSO as an integrating concept in earth science. Science 314, 1740–1745. Myneni, R.B., Keeling, C., Tucker, C.J., Asrar, G., Nemani, R.R., 1997. Increased plant growth in the northern high latitudes from 1981 to 1991. Nature 386, 698. Nemani, R.R., et al., 2003. Climate-driven increases in global terrestrial net primary production from 1982 to 1999. Science 300, 1560–1563. Nicholson, S.E., Tucker, C.J., Ba, M., 1998. Desertification, drought, and surface vegetation: an example from the West African Sahel. Bull. Am. Meteorol. Soc. 79, 815–830. Oba, G., Post, E., Stenseth, N.C., 2001. Sub-saharan desertification and productivity are linked to hemispheric climate variability. Glob. Chang. Biol. 7, 241–246. Palmer, T., Mansfield, D., 1984. Response of two atmospheric general circulation models to sea-surface temperature anomalies in the tropical East and West Pacific. Nature 310, 483. Potter, C., et al., 2003. Global teleconnections of climate to terrestrial carbon flux. J. Geophys. Res. Atmos. 108. Potter, C., Klooster, S., Steinbach, M., Tan, P.N., Kumar, V., Shekhar, S., Carvalho, C.R.d., 2004. Understanding global teleconnections of climate to regional model estimates of Amazon ecosystem carbon fluxes. Glob. Chang. Biol. 10, 693–703. Potter, C., Boriah, S., Steinbach, M., Kumar, V., Klooster, S., 2008. Terrestrial vegetation dynamics and global climate controls. Clim. Dyn. 31, 67–78. Ropelewski, C.F., Halpert, M.S., 1986. North American precipitation and temperature patterns associated with the El Niño/Southern Oscillation (ENSO). Mon. Weather Rev. 114, 2352–2362. Ropelewski, C.F., Halpert, M.S., 1987. Global and regional scale precipitation patterns associated with the El Niño/Southern Oscillation. Mon. Weather Rev. 115, 1606–1626. Rudeva, I., Gulev, S.K., 2011. Composite analysis of North Atlantic extratropical cyclones in NCEP-NCAR reanalysis data. Mon. Weather Rev. 139, 1419–1446. Schubert, S., et al., 2009. A US CLIVAR project to assess and compare the responses of global climate models to drought-related SST forcing patterns: overview and results. J. Clim. 22, 5251–5272. Stenseth, N.C., et al., 2003. Studying climate effects on ecology through the use of climate indices: the North Atlantic Oscillation, El Niño Southern Oscillation and beyond. Proc. R. Soc. B Biol. Sci. 270, 2087–2096. Tian, H., Melillo, J.M., Kicklighter, D.W., McGuire, A.D., Helfrich III, J.V., Moore III, B., VoÈroÈsmarty, C.J., 1998. Effect of interannual climate variability on carbon storage in Amazonian ecosystems. Nature 396, 664. Ting, M., Hoerling, M.P., Xu, T., Kumar, A., 1996. Northern Hemisphere teleconnection patterns during extreme phases of the zonal-mean circulation. J. Clim. 9, 2614–2633. Trenberth, K.E., 1997. The definition of El Niño. Bull. Am. Meteorol. Soc. 78, 2771–2777. Watanabe, S., et al., 2011. MIROC-ESM 2010: model description and basic results of CMIP5-20c3m experiments. Geosci. Model Dev. 4, 845. Wharton, S., Chasmer, L., Falk, M., Paw U, K.T., 2009. Strong links between teleconnections and ecosystem exchange found at a Pacific Northwest old-growth forest from flux tower and MODIS EVI data. Glob. Chang. Biol. 15, 2187–2205. Williams, C., Hanan, N., 2011. ENSO and IOD teleconnections for African ecosystems: evidence of destructive interference between climate oscillations. Biogeosciences 8, 27–40. Woodward, F., Lomas, M., Quaife, T., 2008. Global responses of terrestrial productivity to contemporary climatic oscillations. Philos. Trans. R. Soc. Lond. Ser. B Biol. Sci. 363, 2779–2785. Wu, T., et al., 2013. Global carbon budgets simulated by the Beijing Climate Center Climate System Model for the last century. J. Geophys. Res. Atmos. 118, 4326–4347.

M. Huang et al. / Science of the Total Environment 655 (2019) 641–651 Wu, D., Zhao, X., Liang, S., Zhou, T., Huang, K., Tang, B., Zhao, W., 2015. Time-lag effects of global vegetation responses to climate change. Glob. Chang. Biol. 21, 3520–3531. Yan, X.-H., Ho, C.-R., Zheng, Q., Klemas, V., 1992. Temperature and size variabilities of the Western Pacific Warm Pool. Science 258, 1643–1645. Yan, X.-H., He, Y., Liu, W.T., Zheng, Q., Ho, C.-R., 1997. Centroid motion of the western Pacific warm pool during three recent El Niño–Southern Oscillation events. J. Phys. Oceanogr. 27, 837–845.

651

Zhao, M., Running, S.W., 2010. Drought-induced reduction in global terrestrial net primary production from 2000 through 2009. Science 329, 940–943. Zhou, X., et al., 2017. The effects of the Indo-Pacific warm pool on the stratosphere. Climate Dynamics. Zhu, Z., Piao, S., Xu, Y., Bastos, A., Ciais, P., Peng, S., 2017. The effects of teleconnections on carbon fluxes of global terrestrial ecosystems. Geophys. Res. Lett. 44, 3209–3218.