Ambiguous Variations in Tropical Latent Heat Flux since the Years around 1998

Rongwang Zhang aState Key Laboratory of Tropical Oceanography, South China Sea Institute of Oceanology, Chinese Academy of Sciences, Guangzhou, China
bGlobal Ocean and Climate Research Center, South China Sea Institute of Oceanology, Chinese Academy of Sciences, Guangzhou, China
cInnovation Academy of South China Sea Ecology and Environmental Engineering, Chinese Academy of Sciences, Guangzhou, China

Search for other papers by Rongwang Zhang in
Current site
Google Scholar
PubMed
Close
https://orcid.org/0000-0001-6870-474X
,
Weihao Guo aState Key Laboratory of Tropical Oceanography, South China Sea Institute of Oceanology, Chinese Academy of Sciences, Guangzhou, China
dUniversity of Chinese Academy of Sciences, Beijing, China

Search for other papers by Weihao Guo in
Current site
Google Scholar
PubMed
Close
,
Xin Wang aState Key Laboratory of Tropical Oceanography, South China Sea Institute of Oceanology, Chinese Academy of Sciences, Guangzhou, China
bGlobal Ocean and Climate Research Center, South China Sea Institute of Oceanology, Chinese Academy of Sciences, Guangzhou, China
cInnovation Academy of South China Sea Ecology and Environmental Engineering, Chinese Academy of Sciences, Guangzhou, China

Search for other papers by Xin Wang in
Current site
Google Scholar
PubMed
Close
, and
Chunzai Wang aState Key Laboratory of Tropical Oceanography, South China Sea Institute of Oceanology, Chinese Academy of Sciences, Guangzhou, China
bGlobal Ocean and Climate Research Center, South China Sea Institute of Oceanology, Chinese Academy of Sciences, Guangzhou, China
cInnovation Academy of South China Sea Ecology and Environmental Engineering, Chinese Academy of Sciences, Guangzhou, China

Search for other papers by Chunzai Wang in
Current site
Google Scholar
PubMed
Close
Free access

Abstract

The tropical latent heat flux (LHF) has experienced a significant increase under the background of global warming in the past four decades. However, since the years around 1998, the long-term LHF variations in the tropics have been found to be quite different in various flux products. Three different trends in the LHF, climbing, near zero, and declining, are suggested by five widely used flux products, which hinders our knowledge of the actual LHF variations. Although there are buoy observations in the tropics, these observations are hard to use to evaluate flux products as they have been assimilated and/or used as benchmarks in the flux data production. This study aims to identify credible long-term LHF variations since 1998. A linear model decomposing the LHF variations into contributions from sea surface wind U and air–sea humidity differences is first applied. The linear model results show that the LHF variations have been more positively connected to U variations since 1998. Evidence from in situ and remote sensing observations is subsequently employed to identify how U has varied recently. Both Global Tropical Moored Buoy Array (GTMBA) buoy observations (from 82 buoys) and a multisensor merged satellite product support a slightly downward trend in U in the last two decades. Such a weakening of U is not conducive to oceanic evaporation and leads to a reduced LHF. Consequently, a declining LHF under a weakening U since the emergence of the global warming “hiatus” in approximately 1998 might be more convincing in the sense of data accuracy and physical consistency.

Significance Statement

The latent heat flux acts as the language of air–sea interactions. This study aims to examine how the tropical latent heat flux has changed since the emergence of the global warming slowdown in approximately 1998. The most striking finding is that the long-term variations in the tropical latent heat flux are fairly inconsistent in several widely used flux products in the last two decades. The sea surface wind variation is found to be the primary contributor to the latent heat flux variation after 1998. Observational evidence from buoy and remote sensing data is hence employed to clarify the actual sea surface wind and the latent heat flux variations.

© 2023 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: Xin Wang, wangxin@scsio.ac.cn

Abstract

The tropical latent heat flux (LHF) has experienced a significant increase under the background of global warming in the past four decades. However, since the years around 1998, the long-term LHF variations in the tropics have been found to be quite different in various flux products. Three different trends in the LHF, climbing, near zero, and declining, are suggested by five widely used flux products, which hinders our knowledge of the actual LHF variations. Although there are buoy observations in the tropics, these observations are hard to use to evaluate flux products as they have been assimilated and/or used as benchmarks in the flux data production. This study aims to identify credible long-term LHF variations since 1998. A linear model decomposing the LHF variations into contributions from sea surface wind U and air–sea humidity differences is first applied. The linear model results show that the LHF variations have been more positively connected to U variations since 1998. Evidence from in situ and remote sensing observations is subsequently employed to identify how U has varied recently. Both Global Tropical Moored Buoy Array (GTMBA) buoy observations (from 82 buoys) and a multisensor merged satellite product support a slightly downward trend in U in the last two decades. Such a weakening of U is not conducive to oceanic evaporation and leads to a reduced LHF. Consequently, a declining LHF under a weakening U since the emergence of the global warming “hiatus” in approximately 1998 might be more convincing in the sense of data accuracy and physical consistency.

Significance Statement

The latent heat flux acts as the language of air–sea interactions. This study aims to examine how the tropical latent heat flux has changed since the emergence of the global warming slowdown in approximately 1998. The most striking finding is that the long-term variations in the tropical latent heat flux are fairly inconsistent in several widely used flux products in the last two decades. The sea surface wind variation is found to be the primary contributor to the latent heat flux variation after 1998. Observational evidence from buoy and remote sensing data is hence employed to clarify the actual sea surface wind and the latent heat flux variations.

© 2023 American Meteorological Society. For information regarding reuse of this content and general copyright information, consult the AMS Copyright Policy (www.ametsoc.org/PUBSReuseLicenses).

Corresponding author: Xin Wang, wangxin@scsio.ac.cn

1. Introduction

Oceanic evaporation can modulate and influence the freshwater flux (Schmitt et al. 1989), sea surface salinity (Curry et al. 2003; Boyer et al. 2005; Helm et al. 2010), and continental precipitation (Findell et al. 2019). According to similarity theory and bulk parameterization (Liu et al. 1979; Fairall et al. 2003), oceanic evaporation is approximately proportional to the latent heat flux (LHF) and can be represented by the latter. The LHF, depicting the moisture and heat exchanges between the ocean and atmosphere, is the second-largest term in the net surface heat flux (Qnet) but makes a leading contribution to Qnet variations on time scales larger than the diurnal cycle (Liu and Curry 2006).

The past two decades have witnessed the emergence of a number of studies focusing on the long-term LHF variations. Many of them are in regard to the period ending around the beginning of the twenty-first century, such as 1958–2005 in Yu (2007), 1989–2000 in Liu and Curry (2006), and the second half of the twentieth century in Roderick and Farquhar (2002) and Durack et al. (2012). As evidence of the intensification of the hydrological cycle in response to global warming, the LHF has been demonstrated to have increased since the late 1970s (Curry et al. 2003; Boyer et al. 2005; Held and Soden 2006; Yu 2007; Durack et al. 2012; Zhang et al. 2018; Yu et al. 2020) and is likely to continue to intensify in future climate simulations (Liu et al. 2017).

In contrast, the LHF variations since the beginning of the twenty-first century are not yet clear. Nonnegligible uncertainties exist in current surface heat flux products (Chou et al. 2004; Smith et al. 2011; Bentamy et al. 2017; Zhang et al. 2018; Cronin et al. 2019). Utilizing different flux products may lead to different conclusions. For instance, Liang and Yu (2016) investigated the Qnet variations during the “hiatus” from 2001 to 2010. However, three kinds of long-term variations were found among the four surface heat flux products used in their study. Note that from the late 1990s to the beginning decade of the twenty-first century, the global mean surface temperature was found to show no trend or even a slight cooling (Easterling and Wehner 2009), which is often referred to as the global warming “hiatus” (Guemas et al. 2013; Kosaka and Xie 2013; Boykoff 2014; Trenberth 2015). The different performances of those flux products might be associated with the global warming hiatus.

Generally, the decadal shift of climate variability combined with the uncertainty in current flux products hinders our in-depth understanding of the long-term LHF variations. This paper aims to revisit the long-term LHF variations with a focus on the tropical ocean and examine the LHF performance of current flux products since the emergence of the global warming hiatus. The rest of this paper is organized as follows. Section 2 describes the data and methods employed. Section 3 presents the main results, including the inconsistent LHF variations since 1998 and the possible reasons. A brief buoy validation is conducted in section 4. Conclusions and discussion are presented in section 5.

2. Data and methods

a. Flux products

Five widely used flux products are employed in this study. They are the European Centre for Medium-Range Weather Forecasts (ECWMF) interim reanalysis (ERA-Interim; Dee et al. 2011) and the fifth major global reanalysis produced by ECMWF (ERA5; Hersbach et al. 2020), the objectively analyzed air–sea fluxes (OAFlux; Yu and Weller 2007), the National Centers for Environmental Prediction–Department of Energy AMIP-II reanalysis (NCEP-2; Kanamitsu et al. 2002), and the Tropical Flux dataset (TropFlux; Praveen Kumar et al. 2012). General information about these flux products can be found in Table 1.

Table 1

Information on flux products employed in this study.

Table 1

ERA-Interim is an interim-generation ECWMF product and improvements such as revised model physics and background error constraints have been added. ERA5 is the latest generation of the ECWMF reanalysis products; it replaces ERA-Interim. It combines vast amounts of historical observations into global estimates using advanced modeling and data assimilation systems. NCEP-2 is the revised version of the National Centers for Environmental Prediction–National Center for Atmospheric Research reanalysis 1 (NCEP-1; Kalnay et al. 1996), with corrections for known errors mainly in satellite measurements. TropFlux was developed mainly with input variables from NCEP-1, NCEP-2, and ERA-Interim and some improvements have been made, such as to surface wind in the tropics. OAFlux is a synthesized dataset applying an objective analysis approach to take into account errors from various sources such as reanalysis and satellite data.

Except for ERA-Interim, which stopped updating in August 2019, all of the flux products start from 1979 and extend to the present, so they can provide up-to-date information on global and/or regional air–sea flux variations. Monthly data from 1979 to 2018 are employed here. Note that to balance the length of data before and after 1998, which are both 20 years, the data since 2018 have not been used. This does not change the analyses and main conclusions drawn in this study. All of the data from those flux products are interpolated onto 1° × 1° grid maps. The study area is focused on 30°S–30°N of the global ocean. Reasons why only the data in the tropics are analyzed are given in section 3.

b. GTMBA data

The Global Tropical Moored Buoy Array (GTMBA) data are used to evaluate the credibility of the flux products. The GTMBA is a multinational effort aimed at developing a sustained buoy observation system and providing real-time in situ data in the tropics for weather and climate research. Flux-related variables such as sea surface wind, sea surface temperature (SST), and air humidity from buoy stations across the three tropical oceans are used. A basic data quality-control process has been performed for all buoy observations. Those stations with too many missing values (>20%) are excluded because an intermittent time series of data is not conducive to capturing the long-term variation of a certain variable. Finally, 82 out of 97 buoy stations are extracted and analyzed. By using the COARE 3.0 algorithm (Fairall et al. 2003), the surface wind and air humidity of buoy observations are adjusted to values at 10 and 2 m, respectively, to remain consistent with the nominal height of corresponding variables from flux products. An inverse-distance-weighted average approach is employed to match the flux product data and buoy observational data. For each buoy station, first, the grid box containing the buoy station is identified, and the four values at the box corners are extracted. Second, the distance between the buoy station and each box corner is determined. Third, the average value of the four corner values is calculated using their inverse distance as weights. This value is regarded as a matchup to the buoy measurement. The monthly data from January 1999 to December 2018 are used.

c. Multisensor merged remote sensing observations

The Cross-Calibrated Multi-Platform (CCMP), a level 4 product that provides global sea surface wind estimates, is used here. It was mainly constructed by merging multisensor remote sensing data from the scatterometers QuikScat and ASCAT-A, as well as microwave radiometers such as SSM/I, SSMIS, TMI, GMI, ASMR-E, AMSR2, and WindSat. Here, the CCMP V3.0 monthly data on a 0.25° × 0.25° grid map are used. Detailed information on the development of the CCMP data can be found in Atlas et al. (2011) and Mears et al. (2022) (and online at https://www.remss.com/measurements/ccmp/).

d. Decomposition of the LHF variations

This study focuses on the long-term LHF variations in the past four decades. The bulk formula adopted by those flux products to compute the LHF is expressed as follows (Liu et al. 1979):
QLHF=ρLυCEUΔQ,
where ρ is the air density, Lυ (= 2.46 × 10−6 J kg−1) is the constant of latent heat of evaporation, CE is the turbulent exchange coefficient for latent heat, U is the sea surface wind speed relative to the sea surface (equal to the wind speed at 10 m above the sea surface, if the ocean current is not considered), and ΔQ = qsqa is the air–sea specific humidity difference. The term qa is the surface specific humidity (usually 2 m above the sea surface), and qs is 98% of the saturated specific humidity at a temperature equal to the SST.
Following the linear diagnosis model for interannual LHF variations proposed by Tanimoto et al. (2003), which is based on the Reynolds decomposition x=x¯+x (where x denotes the LHF, U, or ΔQ), the anomalous LHF can be expressed as follows:
QLHF=ρLυCEΔQ¯U+ρLυCE(ΔQ)U¯+R,
where the prime denotes the anomaly and the overbar denotes the time average. The term R is the difference between the left-hand side term and the sum of the first two terms on the right-hand side, which accounts for the total residual caused by second-order terms and/or other uncertainty sources. Note that to simplify the analysis, constant values of ρ (=1.25 kg m−3) and CE (=1.15 × 10−3) are used. The impact of the nonlinearity of ρ on Eq. (2) has been reported to be quite insignificant (Gulev 1994, 1997). The variation in CE is also nonlinear and highly dependent on surface wind conditions (Fairall et al. 2003). However, accurate estimation of CE requires high-frequency eddy covariance measurements from synoptic to seasonal scales. For interannual and larger scales, constant CE values are usually employed and have provided reasonable results in previous studies (Tanimoto et al. 2003; Zhou et al. 2015; Zhang et al. 2018). As we will see later, applying a constant CE leads to a trivial residual in Eq. (2), indicating that the ignoring of CE variations is acceptable here. Equation (2) thus provides us with the opportunity to focus on the different contributions of sea surface wind and air–sea humidity differences to the LHF variations. The first and second terms on the right-hand side of Eq. (2) are defined as term U and term ΔQ, respectively. The anomalous values are extracted as follows: first, the area mean of LHF is calculated over the tropical ocean (30°S–30°N), and the monthly LHF time series is obtained; second, the climatology mean of each month from January to December is calculated; third, the corresponding climatology mean is subtracted from each month to obtain the anomalous value. The anomalous values of other variables are processed in the same way.

3. Results

a. Inconsistent LHF trends since 1998

Figure 1a depicts the long-term LHF variations in the tropics over the past four decades. During the whole period, despite the different rates of variation, upward trends can be observed in all five flux products. This is in agreement with the global warming signal, as a warmer SST leads to a larger LHF release. The coherence between global warming and the LHF increase continued until 1998. Between 1979 and 1998, the long-term trends in the LHF of ERA-Interim, ERA5, OAFlux, NCEP-2, and TropFlux are 0.35, 0.30, 0.23, 0.43, and 0.46 W m−2 yr−1, respectively. All these trend values pass the 95% confidence level. A consistent enhancement of the LHF during this period was supported by these flux products.

Fig. 1.
Fig. 1.

Time series of the anomalous (a) LHF, (b) U, and (c) ΔQ over tropical oceans (30°S–30°N, 0°–360°). The thin and thick straight lines denote the linear trends in 1979–98 and 1999–2018, respectively. The anomalies in each flux product are computed by removing their climatology mean values from 1979 to 2018. The two values in parentheses denote the linear trends before and after 1998, respectively. The bold italics indicate that the value passes the 95% confidence level. The units for the linear trends in (a), (b), and (c) are W m−2 yr−1, m s−1 per decade, and g kg−1 per decade, respectively. The green filled area denotes the period 1979–98. During this period, all the products show consistent signs of long-term trends in the LHF, U, and ΔQ.

Citation: Journal of Climate 36, 10; 10.1175/JCLI-D-22-0381.1

After the emergence of the global warming “hiatus” in approximately 1998, however, the five flux products exhibited different LHF trends with different signs. It becomes difficult to determine the actual LHF variations. The five flux products can be classified into three types according to their LHF variations since 1998. Between 1999 and 2018, the LHF in ERA-Interim and TropFlux shows an increase with a rate of 0.22 and 0.26 W m−2 yr−1, respectively. The LHF in ERA5 and NCEP-2 varies at a rate of 0.05 and −0.06 W m−2 yr−1, respectively. The LHF in OAFlux exhibits a unique declining trend of −0.25 W m−2 yr−1, which, in terms of magnitude, is comparable to its upward trend before 1998. Among these trend values, those of ERA5 and NCEP-2 are near zero and do not exceed the 95% confidence level.

The three types of LHF trends shown in this period hinder our understanding of the LHF response to the global warming hiatus. Note that the three types of LHF variations after 1998 can be found for global oceans as well. Further inspections suggest that, during the period 1999–2018, the differences in global LHF and U performances of these flux products are mainly caused by the differences in their tropical LHF and U performances. In addition, the GTMBA validation data are available only in the tropical oceans. Therefore, the study area of this paper is focused on the tropical oceans within 30°S and 30°N.

Due to the multiple time-scale variations, the results of the trend calculation depend on the selected time period. The reason why 1998 is used to separate the data into two time periods is mainly that the years around 1998 are typically regarded as the time of the emergence of the global warming hiatus. It is known that a super El Niño event occurred in 1997/98. The long-term variations in many parameters of Earth’s climate system have experienced obvious shifts since then. Taking such a specific time as the dividing point is relatively fair for the trend studies of the two time periods. The trend calculation for each time period may be greatly affected if the signals around 1998 are placed in either of the two time periods alone. In addition, the year 1998 exactly balances the lengths of the two time periods, guaranteeing that the trend calculation for each time period is based on the same length of data.

b. Possible causes for the LHF uncertainties

1) Residual analysis of the LHF decomposition

Equation (2) provides a simplified way to attribute the LHF variations to the impacts of anomalous U and ΔQ. However, both assumptions of ignoring the second-order terms and setting the coefficients ρ and CE as constants can affect the magnitude of residuals and thus the accuracy of LHF diagnosis. Therefore, the performance of Eq. (2) is examined first. Figure 2 shows the time series of the anomalous LHF, the sum of term U and term ΔQ, and the residual computed in Eq. (2). The sum of term U and term ΔQ (red curve) can well reproduce the LHF variations (black curve) with quite limited residuals (green bar) in all five flux products. The two curves exhibit quite similar trends and the correlation coefficients (CCs) between term U + term ΔQ and the LHF vary from 0.947 to 0.997 among the five flux products. In particular, for ERA-Interim, ERA5, OAFlux, and TropFlux, the two curves nearly overlap (Figs. 2a–c,e). The mean absolute residuals in ERA-Interim, ERA5, OAFlux, NCEP-2, and TropFlux are 0.64, 0.31, 0.42, 1.85, and 0.36 W m−2, respectively. It is noted that the residual in NCEP-2 is relatively larger than those in the others. Previous studies have reported that the performance of the LHF in NCEP-2 is problematic to some extent (Wang et al. 2013, 2017). The relatively high residual found here indicates that the uncertainties in the LHF itself can probably affect the residual and thereby the performance of Eq. (2).

Fig. 2.
Fig. 2.

Time series of the anomalous LHF (black) and the sum of term U and term ΔQ (red). Term U and term ΔQ denote the first and second terms on the right-hand side of Eq. (2) and represent the contributions to the LHF variations from anomalous U and ΔQ, respectively. The bar denotes the difference between the black and red curves. The vertical gray line denotes the year 1998.

Citation: Journal of Climate 36, 10; 10.1175/JCLI-D-22-0381.1

The above analysis shows that term U and term ΔQ together have sufficient representativeness to reproduce the LHF variations on the time scale discussed in this study, indicating that anomalous variations in the LHF can be well explained by those in U and ΔQ. The impacts of the residuals derived from Eq. (2) on the LHF variations are thus ignored in the following analysis.

2) Spatial distributions of long-term LHF variations

Figures 3 and 4 show the spatial distributions of long-term variations in the LHF, U, and ΔQ before and after 1998, respectively. For the period 1979–98, these three variables exhibited positive trends in most of the tropics (Fig. 3). Although there were some areas with negative trends, they did not change the overall positive trends in these three variables (Fig. 1). Note that the positive trends in U were relatively weak, and areas with negative trends were relatively large, resulting in an overall weaker positive trend in U than those in the LHF and ΔQ (Fig. 1b). For NCEP-2, an overall near-zero trend in U occurred because of the profound declining U in the tropical Pacific.

Fig. 3.
Fig. 3.

Spatial distributions of the long-term trends in the LHF, U, and ΔQ in the period 1979–98. The results of the LHF from (a) ERA-Interim, (b) ERA5, (c) OAFlux, (d) NCEP-2, and (e) TropFlux. (f)–(o) As in (a)–(e), but for U and ΔQ, respectively. Contours denote values that pass the 95% confidence level.

Citation: Journal of Climate 36, 10; 10.1175/JCLI-D-22-0381.1

Fig. 4.
Fig. 4.

As in Fig. 3, but for the period 1999–2018.

Citation: Journal of Climate 36, 10; 10.1175/JCLI-D-22-0381.1

For the period 1999–2018, however, the trends in these three variables became relatively complicated in these five flux products. For the LHF in ERA-Interim and TropFlux (Figs. 4a,e), strong positive trends in the tropical Pacific led to their overall positive trends, and the climbing rates of the LHF were comparable to those before 1998. The LHF in ERA5 and NCEP-2 exhibited overall near-zero trends (Fig. 1a), but the reasons were different. In ERA5, both positive and negative trends were relatively weak (Fig. 4b). In NCEP-2, however, positive and negative trends were larger in magnitude, which offset each other and resulted in an overall near-zero trend but a greater amplitude of the LHF variations (Fig. 4b). In OAFlux, negative trends existed in all three tropical oceans (Fig. 4c) and undoubtedly led to an overall downward trend in the LHF (Fig. 1a).

Similar to the LHF, the variations in U exhibited three types of long-term trends after 1998 (Fig. 1b). ERA-Interim and TropFlux had the strongest upward trends in U, mainly due to their significant increase in U in the southeastern tropical Pacific (Figs. 4f,j). For U in ERA5 and NCEP-2 (Figs. 4g,i), the areas and intensities of positive and negative trends were comparable, leading to their overall near-zero trends (Fig. 1b). For U in OAFlux (Fig. 4h), negative trends were prominent in the tropical Indian Ocean and Pacific, leading to an overall negative trend in U (Fig. 1b).

The long-term trends in ΔQ slowed down significantly after 1998 in all five flux products (Fig. 1c). The areas and intensities of positive trends decreased, accompanied by a marked increase in areas with negative trends (Figs. 4k–o). Consequently, none of the trends in ΔQ in the five flux products exceeded the 95% confidence level (Fig. 1c). In the northern tropical Indian Ocean in ERA-Interim, OAFlux, and TropFlux, the southeast tropical Pacific in OAFlux and NCEP-2, and the equatorial Atlantic in all five flux products, even positive trends before 1998 shifted to negative ones after 1998.

3) Relative contributions from U and ΔQ

Both the tropical mean variations in Fig. 1 and the detailed information in Figs. 3 and 4 suggest that long-term variations in the LHF, U, and ΔQ have changed significantly since 1998. According to Eq. (2), the inconsistent LHF variations after 1998 revealed in the five flux products were probably related to their synchronous variations in U and ΔQ. The four-decade time series in Fig. 1 was divided into two periods: the one before 1998 (1979–98; P1) and the other after 1998 (1999–2018; P2). Figures 5 and 6 show the point-to-point correlation distributions between the LHF and term U and term ΔQ in P1 and P2, respectively. For the relationship between the LHF and term U, high CCs (>0.7) dominated the entire tropics, while low CCs (<0.3) were in quite limited areas (Fig. 5). From P1 to P2, CC distributions exhibited little change, indicating a relatively steady relationship between the LHF and term U for the whole period in all five flux products. For the relationship between the LHF and term ΔQ, high CCs (>0.7) were primarily located in the middle and east equatorial Pacific, the east equatorial Atlantic, the south tropical Indian Ocean, and the south and north borders of the tropics. On the other hand, low CCs (<0.3) dominated the Indo-Pacific warm pool region (Fig. 6). From P1 to P2, areas with low CCs (<0.3) increased significantly, which extended to the southeastern tropical Pacific. This suggests that the relationship between the LHF and term ΔQ has weakened since 1998.

Fig. 5.
Fig. 5.

Distributions of the point-to-point CC between the LHF and term U in the periods (a)–(e) 1979–98 and (f)–(j) 1999–2018. Red and blue contours highlight the values of 0.7 and 0.3, respectively.

Citation: Journal of Climate 36, 10; 10.1175/JCLI-D-22-0381.1

Fig. 6.
Fig. 6.

As in Fig. 5, but for the results of term ΔQ.

Citation: Journal of Climate 36, 10; 10.1175/JCLI-D-22-0381.1

Figure 7 depicts the overall relationships between the tropical mean LHF and term U and term ΔQ. Despite the high correlation between the LHF and term U + term ΔQ shown in Fig. 2, term U and term ΔQ individually exhibited lower and different correlations with the LHF. CCs between term U and the LHF increased slightly in P2 in all flux products except for ERA5, which experienced a slight drop from 0.69 in P1 to 0.64 in P2. Note that NCEP-2 exhibited relatively lower CCs than did those in other flux products. Its CC in P1 did not pass the 95% (but exceeded the 90%) confidence level. In general, the relationships between term U and the LHF varied little from P1 to P2 in all five flux products. A contrasting result for term ΔQ, however, is that its correlation with the LHF experienced a prominent weakening from P1 to P2. Before 1998, the CCs in all five flux products passed the 95% confidence level. However, the CCs after 1998 decreased substantially and did not pass the 95% confidence level in all five flux products. The largest correlation weakening is found in TropFlux, which dropped from 0.73 in P1 to 0.28 in P2. The reduction of correlation between the LHF and ΔQ after 1998 might be associated with changes of large-scale air–sea interactions in a warming climate. In addition, the reduction in the correlation between the LHF and ΔQ after 1998 might be caused by data uncertainties in both the LHF and ΔQ. Further investigations will be conducted to determine why the relationship between the LHF and ΔQ has weakened since 1998.

Fig. 7.
Fig. 7.

CCs between the LHF and both term U (effect of sea surface wind) and term ΔQ (effect of air–sea humidity difference). The light and dark green lines denote the 90% and 95% confidence levels, respectively.

Citation: Journal of Climate 36, 10; 10.1175/JCLI-D-22-0381.1

From the notable weakening in the relationship between term ΔQ and the LHF from P1 to P2, it can be inferred that the LHF variations were more related to the contributions of anomalous U after 1998, which is well illustrated in Figs. 1b and 1c. The long-term trends in U after 1998 were in agreement with those in the LHF after 1998 in all five flux products. Flux products with upward (ERA-Interim and TropFlux), near-zero (ERA5 and NCEP-2), and declining (OAFlux) trends in U were exactly in agreement with corresponding types of trends in the LHF. However, the long-term variations in ΔQ after 1998 did not exceed the 95% confidence level in any of the five flux products (Fig. 1c).

Although the LHF variations in different flux products were inconsistent in the period after 1998, all of their LHF variations were primarily affected by their corresponding U variations. Therefore, to distinguish the different types of long-term LHF trends since 1998, careful attention should be paid to the credibility of U data in those flux products.

4. Validation with in situ and remote sensing observations

As a unique, long-term, and stable suite of in situ observations in global tropical oceans, the GTMBA data are utilized as a benchmark to validate the performances of the five flux products. Figure 8 depicts the comparisons between the flux product data and GTMBA data from 1999 to 2018. For the slopes of the linear fitting lines of the LHF, U, and ΔQ, the variational ranges are 0.41–0.65, 0.73–0.97, and 0.48–0.60, respectively. This suggests that the flux product U data have a much better accuracy, while biases in the LHF and ΔQ are relatively large and the corresponding slopes have nonnegligible disparities compared with a reference slope of one. Here, the relatively large deviations in the LHF might be more attributable to deviations in ΔQ, which is similar to previously reported results (Wang et al. 2017; Zhang et al. 2018).

Fig. 8.
Fig. 8.

Comparisons between the flux products and buoy observations from 1999 to 2018. All data have been processed into anomalies by subtracting their corresponding climatological mean values. (a)–(e) LHF, (f)–(j) U, and (k)–(o) ΔQ. The green line denotes the reference line with slope 1, and the red line denotes the least squares fitting line.

Citation: Journal of Climate 36, 10; 10.1175/JCLI-D-22-0381.1

Table 2 lists the statistics of comparisons between the flux product data and the GTMBA data. For the LHF, all of the flux product data have slight underestimates, with mean biases ranging from −0.83 to −0.07 W m−2. The root-mean-square errors (RMSEs) range from 12.68 to 20.85 W m−2. The mean biases plus RMSE are −0.13 ± 12.95, −0.08 ± 12.68, −0.07 ± 14.97, −0.83 ± 20.85, and −0.59 ± 13.17 W m−2 for ERA-Interim, ERA5, OAFlux, NCEP-2, and TropFlux, respectively. The LHF in ERA5 (NCPE2) has the lowest (highest) RMSE and highest (lowest) CC with buoy observations. For U, all flux product data have near-zero biases on the order of 0.001 m s−1. The RMSEs vary from 0.36 m s−1 in ERA5 to 0.76 m s−1 in NCEP-2. ERA5 (NCEP-2) has the smallest (largest) sum of mean bias and RMSE and the highest (lowest) CC with buoy observations. For ΔQ, all flux product data show slight underestimates, with mean biases ranging from −0.006 g kg−1 in ERA5 to −0.032 g kg−1 in TropFlux. The RMSEs vary from 0.49 g kg−1 in ERA5 to 0.62 g kg−1 in NCEP-2. ERA5 performs slightly better than the others, as evidenced by its having the lowest RMSE and highest CC with buoy observations.

Table 2

Comparisons between the five flux products and the GTMBA data for the LHF (W m−2), U (m s−1), and ΔQ (g kg−1) in the period 1999–2018. The climatological mean values in each product and buoy data have been removed to focus on the anomalous values only. The abbreviations CORR, BIAS, and RMSE denote the correlation coefficient, mean bias, CC, and root-mean-square error, respectively.

Table 2

It should be noted that despite the contrasting differences between ERA-Interim and ERA5 for the long-term variations in the LHF and U, their data performances of the LHF and U with respect to buoy observations are quite close to each other. This phenomenon has implications regarding two aspects. On the one hand, for a certain dataset, its mean bias and trend error can be unrelated, as the former is independent of the chronological order of each error. On the other hand, existing limited buoy observations are less effective for adequately evaluating them.

To avoid the trend error being sheltered by the bias averaging process, the long-term trend in U at each buoy station is depicted in Fig. 9. For buoy observations, positive and negative trends were mainly distributed in the southeastern and western tropical Pacific, respectively. In the tropical Atlantic and Indian Oceans, the numbers of stations with positive and negative trends were almost comparable. Most of the flux products exhibited distributions of positive and negative trends similar to those of the buoy observations, except for their different intensities. By counting the number of stations with the same sign of the U trend for both flux products and buoy data, it is found that the numbers of eligible buoy stations for the flux products are quite close to each other (ranging from 64 to 68). ERA5 has the highest number of eligible buoy stations, 68 out of 82.

Fig. 9.
Fig. 9.

Long-term trend in U calculated at each buoy station for the (a) GTMBA data and (b)–(f) five flux products. There are 82 buoy stations employed in total. The numbers at the top left of (b)–(f) denote the counting number of buoy stations with the same sign of the long-term trend in U for both the flux product and buoy data.

Citation: Journal of Climate 36, 10; 10.1175/JCLI-D-22-0381.1

Figure 10 shows the time series of U variations averaged over all buoy stations. As remote sensing observations, the CCMP sea surface wind product is employed as auxiliary data here. The spread in these flux products is quite small, and their ensemble mean shows overall high consistency with buoy observations. From 1999 to 2018, the flux products and buoy observations exhibited similar peaks, troughs, and long-term trends. For instance, the troughs in 2004, 2009, and 2015 and the peaks in 2007, 2010, and 2011 corresponded precisely to the occurrence of El Niño events in 2004/05, 2009/10, and 2015/16 and La Niña events in 2007/08, 2010/11, and 2011/12, respectively. Similar features can also be depicted by the satellite data. This indicates that, on the one hand, existing buoy observations can well reflect the corresponding weakening and strengthening signals of trade winds during the onset of El Niño and La Niña, respectively. On the other hand, both flux products and the satellite data coincide well with the GTMBA data.

Fig. 10.
Fig. 10.

Time series of U variations from 1999 to 2018. The dashed–dotted curve denotes the anomalous values, and the solid curve denotes the least squares fitting line. The bar denotes the result from satellite data averaged over the entire tropics. The pink filled area denotes the ensemble mean uncertainty defined by one standard deviation among those five flux products. The value in parentheses denotes the linear trend (m s−1 per decade). The green value denotes the trend passing the 95% confidence level.

Citation: Journal of Climate 36, 10; 10.1175/JCLI-D-22-0381.1

The long-term trends in U in the ensemble mean of flux products, merged satellite data, and buoy observations are −0.03, −0.12, and −0.08 m s−1 per decade, respectively, consistently supporting a downward trend in U. However, these results are derived from the data averaged over existing buoy stations, which cannot represent U variations over the entire tropics. In addition, these trends do not pass the 95% confidence level. Considering the spatiotemporal heterogeneity of buoy observations, the entire tropical mean U was computed with the merged satellite data. The trend in the entire tropical mean U is 0.07 m s−1 per decade and passes the 95% confidence level.

5. Conclusions and discussion

The five flux products investigated in this study consistently reveal the enhancement of tropical oceanic evaporation over the past four decades under the influence of global warming. However, accompanied by the emergence of the global warming “hiatus” in approximately 1998, the long-term LHF variations since 1998 have exhibited large diversity among five widely used flux products. This diversity is not only in regard to differences in magnitude but also in the sign of the trend. The ambiguous LHF variations can be typically classified into three types: increasing (ERA-Interim and TropFlux), near zero (ERA5 and NCEP-2), and declining (OAFlux). Uncertainty in the LHF variation after 1998 makes it difficult to determine the oceanic evaporation responses to the global warming “hiatus.”

A simplified linear decomposition method is utilized to differentiate the anomalous LHF variations from contributions from anomalous U and ΔQ variations. Although some assumptions have been made, the residual of this method is found to be within acceptable ranges on the time scale discussed in this study. This method provides an efficient way to explore the origins of uncertainties in the LHF since 1998. The correlation between the LHF and term U plus term ΔQ exhibited no obvious difference before and after 1998, indicating that the sum of contributions from anomalous U and ΔQ variations can adequately represent the anomalous LHF variations throughout the past four decades. However, a contrasting feature is that the contribution from term ΔQ to the anomalous LHF variations apparently weakened after 1998, while the correlation between term U and the LHF remained at a relatively high and stable level. Taking the two aspects together, the contribution from term U to the anomalous LHF variations became more important after 1998. Therefore, it is expected that a flux product with more reliable U variations would probably have more reliable LHF variations since 1998.

As a key parameter in climate change studies, the sea surface wind and its long-term variation have drawn wide attention in recent years. Using satellite observations from the SSM/I, Wentz et al. (2007) reported that U exhibited growing rates of 5% ± 3.5% K−1 over the global ocean and 3% ± 3.5% K−1 in the tropics from 1987 to 2006. Using a wave- and anemometer-based sea surface wind dataset constructed from ship observations, Tokinaga and Xie (2011) reported a slow upward trend in U from 1950 to 2008. Using corrected anemometer-based observations, Ma et al. (2016) reported a slight increase in U from 1970 to 1995, regardless of the concurrent weakening of the Pacific Walker circulation. An upward trend in U since the late 1970s was supported by Yu (2007) using objectively blended data sources from satellite and reanalysis data. Based on six wind products, Hu et al. (2020) reported a planetary intensification in U from 1985 to 2010. Using an extensive satellite database consisting of 13 altimeters, 11 radiometers, and seven scatterometers, Young and Ribal (2019) identified a small increase in U from 1985 to 2018, a period quite close to the one focused on in this study. Similar weak increases in U in all five flux products can be observed by considering the U variations from 1979 to 2018 in Fig. 1b.

These previously reported results derived from multisource datasets consistently suggest that the global- and/or tropical mean U has increased since the second half of the twentieth century. However, these studies basically focused on U variations either in a relatively long period or before the global warming “hiatus.” Little attention has been specifically paid to this century’s U variations. During this period, different flux products exhibited inconsistent trends in U, leading to correspondingly ambiguous variations in the LHF. Recently, Robertson et al. (2020) reported that U increased from the early 1990s to nearly 2010 but declined after 2010, according to both satellite-based estimates and reduced observation reanalyses. Their findings regarding the decrease in U after 2010 can be well supported by this study.

The validation based on in situ observations from the GTMBA data suggests that ERA5 U data exhibited the overall lowest mean biases and RMSEs after 1998. However, ERA5 did not have an overwhelming advantage. Except for NCEP-2, which exhibited a relatively large bias and RMSE, there was little significant difference among the other flux products. It is known that the GTMBA data have been assimilated and/or used as validation data in the production of these flux products (Josey et al. 2014), leading to their highly consistent data performance with respect to existing buoy observations (Fig. 10). However, the fact that they differ in the entire tropical mean U variations and thus the LHF variations (Figs. 1a,b) suggests that it is difficult to determine which flux product has a more reliable data performance according to existing limited and not independent buoy observations.

Recent years have witnessed the explosive growth of remote sensing observations (Wentz et al. 2007; Young and Ribal 2019). The CCMP, a multisensor merged satellite sea surface wind product, exhibited high data reliability with respect to the GTMBA data in the period 1999–2018. From the view of this satellite product, the sea surface wind has probably weakened in the tropics in the last two decades. Benefitting from the deep incorporation of multisensor satellite data, OAFlux exhibited a quite similar declining rate of U compared with this satellite data after 1998. Therefore, as suggested by OAFlux, the tropical LHF has probably weakened since 1998. Along with the emergence of the global warming “hiatus” in approximately 1998, the SST warming slowed down as well, which led to the LHF declining from then on. The global warming “hiatus” was thought to have ended in approximately 2013 (Xie and Kosaka 2017). However, as indicated by the LHF variations after 2013, the impact from the reboot of SST warming might not have approximately counterbalanced the impact from the persistent U weakening. As a result, the LHF continued to decline after 2013 against the background of U weakening.

It is worth noting that the defects of buoy observations themselves should be acknowledged. The buoy observations utilized in this study have at least two limitations. On the one hand, instrument error and sampling error may introduce uncertainties into these buoy observations. However, no uncertainty estimate has been provided with these buoy observations. On the other hand, these buoy stations are not quite continuous in time and are relatively sparse in space. Such spatiotemporal heterogeneity of buoy observations may cause uncertainty to some extent when using them in climate research.

Because of the incorporation of ERA-Interim data, the phasing out of ERA-Interim after August 2019 may have impacts on the updating and/or recomputation of historical data for flux products such as TropFlux and OAFlux. It is not clear whether it is better to replace ERA-Interim with ERA5 in part or in whole. Although ERA-Interim and ERA-5 are different versions from the same project, the differences in their LHF variations after 1998 were significant. For a certain flux product that incorporates ERA-Interim data, in the case of keeping ERA-Interim historical data before 2019 and using ERA5 after 2019, its long-term LHF variations may be more inconclusive due to data splicing. If all input ERA-Interim data are completely replaced with ERA5, its historical LHF variations may be changed to some extent. Thus, it is suggested that future updating of flux products that incorporate ERA-Interim data should pay attention to the different performances of the LHF and U in ERA-Interim and ERA5 since 1998.

Acknowledgments.

This work was supported by the National Natural Science Foundation of China (41925024, 41906178, and 41731173), the Strategic Priority Research Program of Chinese Academy of Sciences (XDB42000000), the National Key R&D Program of China (2022YFF0801400, 2019YFA0606701), the Innovation Academy of South China Sea Ecology and Environmental Engineering, Chinese Academy of Sciences (ISEE2018PY06), and the Leading Talents of Guangdong Province Program. Rongwang Zhang was supported by the Independent Research Project Program of State Key Laboratory of Tropical Oceanography (LTOZZ2004). and the Science and Technology Program of Guangdong Province (2022B1212050003).

Data availability statement.

The LHF and related variables were taken from the following publicly available sources: ERA-Interim (https://www.ecmwf.int/en/forecasts/datasets/reanalysis-datasets/era-interim), ERA5 (https://www.ecmwf.int/en/forecasts/datasets/reanalysis-datasets/era5), OAFlux (http://oaflux.whoi.edu), NCEP-2 (https://psl.noaa.gov/data/gridded/data.ncep.reanalysis2.gaussian.html), and TropFlux (https://incois.gov.in/tropflux/tf_products.jsp). The buoy data GTMBA and satellite data CCMP were obtained via https://www.pmel.noaa.gov/gtmba/ and www.remss.com, respectively.

REFERENCES

  • Atlas, R., R. N. Hoffman, J. Ardizzone, S. M. Leidner, J. C. Jusem, D. K. Smith, and D. Gombos, 2011: A cross-calibrated, multiplatform ocean surface wind velocity product for meteorological and oceanographic applications. Bull. Amer. Meteor. Soc., 92, 157–174, https://doi.org/10.1175/2010BAMS2946.1.

    • Search Google Scholar
    • Export Citation
  • Bentamy, A., and Coauthors, 2017: Review and assessment of latent and sensible heat flux accuracy over the global oceans. Remote Sens. Environ., 201, 196218, https://doi.org/10.1016/j.rse.2017.08.016.

    • Search Google Scholar
    • Export Citation
  • Boyer, T. P., S. Levitus, J. I. Antonov, R. A. Locarnini, and H. E. Garcia, 2005: Linear trends in salinity for the World Ocean, 1955–1998. Geophys. Res. Lett., 32, L01604, https://doi.org/10.1029/2004GL021791.

    • Search Google Scholar
    • Export Citation
  • Boykoff, M. T., 2014: Media discourse on the climate slowdown. Nat. Climate Change, 4, 156158, https://doi.org/10.1038/nclimate2156.

  • Chou, S.-H., E. Nelkin, J. Ardizzone, and R. M. Atlas, 2004: A comparison of latent heat fluxes over global oceans for four flux products. J. Climate, 17, 39733989, https://doi.org/10.1175/1520-0442(2004)017<3973:ACOLHF>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Cronin, M. F., and Coauthors, 2019: Air-sea fluxes with a focus on heat and momentum. Front. Mar. Sci., 6, 430, https://doi.org/10.3389/fmars.2019.00430.

    • Search Google Scholar
    • Export Citation
  • Curry, R., B. Dickson, and I. Yashayaev, 2003: A change in the freshwater balance of the Atlantic Ocean over the past four decades. Nature, 426, 826829, https://doi.org/10.1038/nature02206.

    • Search Google Scholar
    • Export Citation
  • Dee, D. P., and Coauthors, 2011: The ERA-Interim reanalysis: Configuration and performance of the data assimilation system. Quart. J. Roy. Meteor. Soc., 137, 553597, https://doi.org/10.1002/qj.828.

    • Search Google Scholar
    • Export Citation
  • Durack, P. J., S. E. Wijffels, and R. J. Matear, 2012: Ocean salinities reveal strong global water cycle intensification during 1950 to 2000. Science, 336, 455458, https://doi.org/10.1126/science.1212222.

    • Search Google Scholar
    • Export Citation
  • Easterling, D. R., and M. F. Wehner, 2009: Is the climate warming or cooling? Geophys. Res. Lett., 36, L08706, https://doi.org/10.1029/2009GL037810.

    • Search Google Scholar
    • Export Citation
  • Fairall, C. W., E. F. Bradley, J. E. Hare, A. A. Grachev, and J. B. Edson, 2003: Bulk parameterization of air–sea fluxes: Updates and verification for the COARE algorithm. J. Climate, 16, 571591, https://doi.org/10.1175/1520-0442(2003)016<0571:BPOASF>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Findell, K. L., P. W. Keys, R. J. van der Ent, B. R. Lintner, A. Berg, and J. P. Krasting, 2019: Rising temperatures increase importance of oceanic evaporation as a source for continental precipitation. J. Climate, 32, 77137726, https://doi.org/10.1175/JCLI-D-19-0145.1.

    • Search Google Scholar
    • Export Citation
  • Guemas, V., F. J. Doblas-Reyes, I. Andreu-Burillo, and M. Asif, 2013: Retrospective prediction of the global warming slowdown in the past decade. Nat. Climate Change, 3, 649653, https://doi.org/10.1038/nclimate1863.

    • Search Google Scholar
    • Export Citation
  • Gulev, S. K., 1994: Influence of space-time averaging on the ocean-atmosphere exchange estimates in the North Atlantic midlatitudes. J. Phys. Oceanogr., 24, 12361255, https://doi.org/10.1175/1520-0485(1994)024<1236:IOSTAO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Gulev, S. K., 1997: Climatologically significant effects of space–time averaging in the North Atlantic sea–air heat flux fields. J. Climate, 10, 27432763, https://doi.org/10.1175/1520-0442(1997)010<2743:CSEOST>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Held, I. M., and B. J. Soden, 2006: Robust responses of the hydrological cycle to global warming. J. Climate, 19, 56865699, https://doi.org/10.1175/JCLI3990.1.

    • Search Google Scholar
    • Export Citation
  • Helm, K. P., N. L. Bindoff, and J. A. Chuch, 2010: Changes in the global hydrological–cycle inferred from ocean salinity. Geophys. Res. Lett., 37, L18701, https://doi.org/10.1029/2010GL044222.

    • Search Google Scholar
    • Export Citation
  • Hersbach, H., and Coauthors, 2020: The ERA5 global reanalysis. Quart. J. Roy. Meteor. Soc., 146, 19992049, https://doi.org/10.1002/qj.3803.

    • Search Google Scholar
    • Export Citation
  • Hu, S., J. Sprintall, C. Guan, M. J. McPhaden, F. Wang, D. Hu, and W. Cai, 2020: Deep-reaching acceleration of global mean ocean circulation over the past two decades. Sci. Adv., 6, eaax7727, https://doi.org/10.1126/sciadv.aax7727.

    • Search Google Scholar
    • Export Citation
  • Josey, S. A., L. Yu, S. Gulev, X. Jin, N. Tilinina, and B. Barnier, 2014: Unexpected impacts of the tropical Pacific array on reanalysis surface meteorology and heat fluxes. Geophys. Res. Lett., 41, 62136220, https://doi.org/10.1002/2014GL061302.

    • Search Google Scholar
    • Export Citation
  • Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77, 437472, https://doi.org/10.1175/1520-0477(1996)077<0437:TNYRP>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Kanamitsu, M., W. Ebisuzaki, J. Woollen, S.-K. Yang, J. J. Hnilo, M. Fiorino, and G. L. Potter, 2002: NCEP–DOE AMIP-II Reanalysis (R-2). Bull. Amer. Meteor. Soc., 83, 16311643, https://doi.org/10.1175/BAMS-83-11-1631.

    • Search Google Scholar
    • Export Citation
  • Kosaka, Y., and S.-P. Xie, 2013: Recent global-warming hiatus tied to equatorial Pacific surface cooling. Nature, 501, 403407, https://doi.org/10.1038/nature12534.

    • Search Google Scholar
    • Export Citation
  • Liang, X., and L. Yu, 2016: Variations of the global net air–sea heat flux during the “hiatus” period (2001–10). J. Climate, 29, 36473660, https://doi.org/10.1175/JCLI-D-15-0626.1.

    • Search Google Scholar
    • Export Citation
  • Liu, J., and J. A. Curry, 2006: Variability of the tropical and subtropical ocean surface latent heat flux during 1989–2000. Geophys. Res. Lett., 33, L05706, https://doi.org/10.1029/2005GL024809.

    • Search Google Scholar
    • Export Citation
  • Liu, W. T., K. B. Katsaros, and J. A. Businger, 1979: Bulk parameterization of air–sea exchanges of heat and water vapor including the molecular constraints at the interface. J. Atmos. Sci., 36, 17221735, https://doi.org/10.1175/1520-0469(1979)036<1722:BPOASE>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Liu, X., A. Köhl, and D. Stamme, 2017: Dynamical ocean response to projected changes of the global water cycle. J. Geophys. Res. Oceans, 122, 65126532, https://doi.org/10.1002/2017JC013061.

    • Search Google Scholar
    • Export Citation
  • Ma, J., G. R. Foltz, B. J. Soden, G. Huang, J. He, and C. Dong, 2016: Will surface winds weaken in response to global warming? Environ. Res. Lett., 11, 124012, https://doi.org/10.1088/1748-9326/11/12/124012.

    • Search Google Scholar
    • Export Citation
  • Mears, C., T. Lee, L. Ricciardulli, X. Wang, and F. Wentz, 2022: Improving the accuracy of the Cross-Calibrated Multi-Platform (CCMP) ocean vector winds. Remote Sens., 14, 4230, https://doi.org/10.3390/rs14174230.

    • Search Google Scholar
    • Export Citation
  • Praveen Kumar, B., J. Vialard, M. Lengaigne, V. S. N. Murty, and M. J. McPhaden, 2012: TropFlux: Air-sea fluxes for the global tropical oceans–description and evaluation. Climate Dyn., 38, 15211543, https://doi.org/10.1007/s00382-011-1115-0.

    • Search Google Scholar
    • Export Citation
  • Robertson, F. R., and Coauthors, 2020: Uncertainties in ocean latent heat flux variations over recent decades in satellite-based estimates and reduced observation reanalyses. J. Climate, 33, 84158437, https://doi.org/10.1175/JCLI-D-19-0954.1.

    • Search Google Scholar
    • Export Citation
  • Roderick, M. L., and G. D. Farquhar, 2002: The cause of decreased pan evaporation over the past 50 years. Science, 298, 14101411, https://doi.org/10.1126/science.1075390-a.

    • Search Google Scholar
    • Export Citation
  • Schmitt, R. W., P. S. Bogden, and C. E. Dorman, 1989: Evaporation minus precipitation and density fluxes for the North Atlantic. J. Phys. Oceanogr., 19, 12081221, https://doi.org/10.1175/1520-0485(1989)019<1208:EMPADF>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Smith, S. R., P. J. Hughes, and M. A. Bourassa, 2011: A comparison of nine monthly air–sea flux products. Int. J. Climatol., 31, 10021027, https://doi.org/10.1002/joc.2225.

    • Search Google Scholar
    • Export Citation
  • Tanimoto, Y., H. Nakamura, T. Kagimoto, and S. Yamane, 2003: An active role of extratropical sea surface temperature anomalies in determining anomalous turbulent heat flux. J. Geophys. Res., 108, 3304, https://doi.org/10.1029/2002JC001750.

    • Search Google Scholar
    • Export Citation
  • Tokinaga, H., and S.-P. Xie, 2011: Wave and Anemometer-based Sea surface Wind (WASWind) for climate change analysis. J. Climate, 24, 267285, https://doi.org/10.1175/2010JCLI3789.1.

    • Search Google Scholar
    • Export Citation
  • Trenberth, K. E., 2015: Has there been a hiatus? Science, 349, 691692, https://doi.org/10.1126/science.aac9225.

  • Wang, D., L. Zeng, X. Li, and P. Shi, 2013: Validation of satellite-derived daily latent heat flux over the South China Sea, compared with observations and five products. J. Atmos. Oceanic Technol., 30, 18201832, https://doi.org/10.1175/JTECH-D-12-00153.1.

    • Search Google Scholar
    • Export Citation
  • Wang, X., R. Zhang, J. Huang, L. Zeng, and F. Huang, 2017: Biases of five latent heat flux products and their impacts on mixed-layer temperature estimates in the South China Sea. J. Geophys. Res. Oceans, 122, 50885104, https://doi.org/10.1002/2016JC012332.

    • Search Google Scholar
    • Export Citation
  • Wentz, F. J., L. Ricciardulli, K. Hilburn, and C. Mears, 2007: How much more rain will global warming bring? Science, 317, 233235, https://doi.org/10.1126/science.1140746.

    • Search Google Scholar
    • Export Citation
  • Xie, S.-P., and Y. Kosaka, 2017: What caused the global surface warming hiatus of 1998–2013? Curr. Climate Change Rep., 3, 128140, https://doi.org/10.1007/s40641-017-0063-0.

    • Search Google Scholar
    • Export Citation
  • Young, I. R., and A. Ribal, 2019: Multiplatform evaluation of global trends in wind speed and wave height. Science, 364, 548552, https://doi.org/10.1126/science.aav9527.

    • Search Google Scholar
    • Export Citation
  • Yu, L., 2007: Global variations in oceanic evaporation (1958–2005): The role of the changing wind Speed. J. Climate, 20, 53765390, https://doi.org/10.1175/2007JCLI1714.1.

    • Search Google Scholar
    • Export Citation
  • Yu, L., and R. A. Weller, 2007: Objectively analyzed air–sea heat fluxes for the global ice-free oceans (1981–2005). Bull. Amer. Meteor. Soc., 88, 527540, https://doi.org/10.1175/BAMS-88-4-527.

    • Search Google Scholar
    • Export Citation
  • Yu, L., S. A. Josey, F. M. Bingham, and T. Lee, 2020: Intensification of the global water cycle and evidence from ocean salinity: A synthesis review. Ann. N. Y. Acad. Sci., 1472, 7694, https://doi.org/10.1111/nyas.14354.

    • Search Google Scholar
    • Export Citation
  • Zhang, R., X. Wang, and C. Wang, 2018: On the simulations of global oceanic latent heat flux in the CMIP5 multimodel ensemble. J. Climate, 31, 71117128, https://doi.org/10.1175/JCLI-D-17-0713.1.

    • Search Google Scholar
    • Export Citation
  • Zhou, L. T., G. Chen, and R. Wu, 2015: Change in surface latent heat flux and its association with tropical cyclone genesis in the western North Pacific. Theor. Appl. Climatol., 119, 221227, https://doi.org/10.1007/s00704-014-1096-0.

    • Search Google Scholar
    • Export Citation
Save
  • Atlas, R., R. N. Hoffman, J. Ardizzone, S. M. Leidner, J. C. Jusem, D. K. Smith, and D. Gombos, 2011: A cross-calibrated, multiplatform ocean surface wind velocity product for meteorological and oceanographic applications. Bull. Amer. Meteor. Soc., 92, 157–174, https://doi.org/10.1175/2010BAMS2946.1.

    • Search Google Scholar
    • Export Citation
  • Bentamy, A., and Coauthors, 2017: Review and assessment of latent and sensible heat flux accuracy over the global oceans. Remote Sens. Environ., 201, 196218, https://doi.org/10.1016/j.rse.2017.08.016.

    • Search Google Scholar
    • Export Citation
  • Boyer, T. P., S. Levitus, J. I. Antonov, R. A. Locarnini, and H. E. Garcia, 2005: Linear trends in salinity for the World Ocean, 1955–1998. Geophys. Res. Lett., 32, L01604, https://doi.org/10.1029/2004GL021791.

    • Search Google Scholar
    • Export Citation
  • Boykoff, M. T., 2014: Media discourse on the climate slowdown. Nat. Climate Change, 4, 156158, https://doi.org/10.1038/nclimate2156.

  • Chou, S.-H., E. Nelkin, J. Ardizzone, and R. M. Atlas, 2004: A comparison of latent heat fluxes over global oceans for four flux products. J. Climate, 17, 39733989, https://doi.org/10.1175/1520-0442(2004)017<3973:ACOLHF>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Cronin, M. F., and Coauthors, 2019: Air-sea fluxes with a focus on heat and momentum. Front. Mar. Sci., 6, 430, https://doi.org/10.3389/fmars.2019.00430.

    • Search Google Scholar
    • Export Citation
  • Curry, R., B. Dickson, and I. Yashayaev, 2003: A change in the freshwater balance of the Atlantic Ocean over the past four decades. Nature, 426, 826829, https://doi.org/10.1038/nature02206.

    • Search Google Scholar
    • Export Citation
  • Dee, D. P., and Coauthors, 2011: The ERA-Interim reanalysis: Configuration and performance of the data assimilation system. Quart. J. Roy. Meteor. Soc., 137, 553597, https://doi.org/10.1002/qj.828.

    • Search Google Scholar
    • Export Citation
  • Durack, P. J., S. E. Wijffels, and R. J. Matear, 2012: Ocean salinities reveal strong global water cycle intensification during 1950 to 2000. Science, 336, 455458, https://doi.org/10.1126/science.1212222.

    • Search Google Scholar
    • Export Citation
  • Easterling, D. R., and M. F. Wehner, 2009: Is the climate warming or cooling? Geophys. Res. Lett., 36, L08706, https://doi.org/10.1029/2009GL037810.

    • Search Google Scholar
    • Export Citation
  • Fairall, C. W., E. F. Bradley, J. E. Hare, A. A. Grachev, and J. B. Edson, 2003: Bulk parameterization of air–sea fluxes: Updates and verification for the COARE algorithm. J. Climate, 16, 571591, https://doi.org/10.1175/1520-0442(2003)016<0571:BPOASF>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Findell, K. L., P. W. Keys, R. J. van der Ent, B. R. Lintner, A. Berg, and J. P. Krasting, 2019: Rising temperatures increase importance of oceanic evaporation as a source for continental precipitation. J. Climate, 32, 77137726, https://doi.org/10.1175/JCLI-D-19-0145.1.

    • Search Google Scholar
    • Export Citation
  • Guemas, V., F. J. Doblas-Reyes, I. Andreu-Burillo, and M. Asif, 2013: Retrospective prediction of the global warming slowdown in the past decade. Nat. Climate Change, 3, 649653, https://doi.org/10.1038/nclimate1863.

    • Search Google Scholar
    • Export Citation
  • Gulev, S. K., 1994: Influence of space-time averaging on the ocean-atmosphere exchange estimates in the North Atlantic midlatitudes. J. Phys. Oceanogr., 24, 12361255, https://doi.org/10.1175/1520-0485(1994)024<1236:IOSTAO>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Gulev, S. K., 1997: Climatologically significant effects of space–time averaging in the North Atlantic sea–air heat flux fields. J. Climate, 10, 27432763, https://doi.org/10.1175/1520-0442(1997)010<2743:CSEOST>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Held, I. M., and B. J. Soden, 2006: Robust responses of the hydrological cycle to global warming. J. Climate, 19, 56865699, https://doi.org/10.1175/JCLI3990.1.

    • Search Google Scholar
    • Export Citation
  • Helm, K. P., N. L. Bindoff, and J. A. Chuch, 2010: Changes in the global hydrological–cycle inferred from ocean salinity. Geophys. Res. Lett., 37, L18701, https://doi.org/10.1029/2010GL044222.

    • Search Google Scholar
    • Export Citation
  • Hersbach, H., and Coauthors, 2020: The ERA5 global reanalysis. Quart. J. Roy. Meteor. Soc., 146, 19992049, https://doi.org/10.1002/qj.3803.

    • Search Google Scholar
    • Export Citation
  • Hu, S., J. Sprintall, C. Guan, M. J. McPhaden, F. Wang, D. Hu, and W. Cai, 2020: Deep-reaching acceleration of global mean ocean circulation over the past two decades. Sci. Adv., 6, eaax7727, https://doi.org/10.1126/sciadv.aax7727.

    • Search Google Scholar
    • Export Citation
  • Josey, S. A., L. Yu, S. Gulev, X. Jin, N. Tilinina, and B. Barnier, 2014: Unexpected impacts of the tropical Pacific array on reanalysis surface meteorology and heat fluxes. Geophys. Res. Lett., 41, 62136220, https://doi.org/10.1002/2014GL061302.

    • Search Google Scholar
    • Export Citation
  • Kalnay, E., and Coauthors, 1996: The NCEP/NCAR 40-Year Reanalysis Project. Bull. Amer. Meteor. Soc., 77, 437472, https://doi.org/10.1175/1520-0477(1996)077<0437:TNYRP>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Kanamitsu, M., W. Ebisuzaki, J. Woollen, S.-K. Yang, J. J. Hnilo, M. Fiorino, and G. L. Potter, 2002: NCEP–DOE AMIP-II Reanalysis (R-2). Bull. Amer. Meteor. Soc., 83, 16311643, https://doi.org/10.1175/BAMS-83-11-1631.

    • Search Google Scholar
    • Export Citation
  • Kosaka, Y., and S.-P. Xie, 2013: Recent global-warming hiatus tied to equatorial Pacific surface cooling. Nature, 501, 403407, https://doi.org/10.1038/nature12534.

    • Search Google Scholar
    • Export Citation
  • Liang, X., and L. Yu, 2016: Variations of the global net air–sea heat flux during the “hiatus” period (2001–10). J. Climate, 29, 36473660, https://doi.org/10.1175/JCLI-D-15-0626.1.

    • Search Google Scholar
    • Export Citation
  • Liu, J., and J. A. Curry, 2006: Variability of the tropical and subtropical ocean surface latent heat flux during 1989–2000. Geophys. Res. Lett., 33, L05706, https://doi.org/10.1029/2005GL024809.

    • Search Google Scholar
    • Export Citation
  • Liu, W. T., K. B. Katsaros, and J. A. Businger, 1979: Bulk parameterization of air–sea exchanges of heat and water vapor including the molecular constraints at the interface. J. Atmos. Sci., 36, 17221735, https://doi.org/10.1175/1520-0469(1979)036<1722:BPOASE>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Liu, X., A. Köhl, and D. Stamme, 2017: Dynamical ocean response to projected changes of the global water cycle. J. Geophys. Res. Oceans, 122, 65126532, https://doi.org/10.1002/2017JC013061.

    • Search Google Scholar
    • Export Citation
  • Ma, J., G. R. Foltz, B. J. Soden, G. Huang, J. He, and C. Dong, 2016: Will surface winds weaken in response to global warming? Environ. Res. Lett., 11, 124012, https://doi.org/10.1088/1748-9326/11/12/124012.

    • Search Google Scholar
    • Export Citation
  • Mears, C., T. Lee, L. Ricciardulli, X. Wang, and F. Wentz, 2022: Improving the accuracy of the Cross-Calibrated Multi-Platform (CCMP) ocean vector winds. Remote Sens., 14, 4230, https://doi.org/10.3390/rs14174230.

    • Search Google Scholar
    • Export Citation
  • Praveen Kumar, B., J. Vialard, M. Lengaigne, V. S. N. Murty, and M. J. McPhaden, 2012: TropFlux: Air-sea fluxes for the global tropical oceans–description and evaluation. Climate Dyn., 38, 15211543, https://doi.org/10.1007/s00382-011-1115-0.

    • Search Google Scholar
    • Export Citation
  • Robertson, F. R., and Coauthors, 2020: Uncertainties in ocean latent heat flux variations over recent decades in satellite-based estimates and reduced observation reanalyses. J. Climate, 33, 84158437, https://doi.org/10.1175/JCLI-D-19-0954.1.

    • Search Google Scholar
    • Export Citation
  • Roderick, M. L., and G. D. Farquhar, 2002: The cause of decreased pan evaporation over the past 50 years. Science, 298, 14101411, https://doi.org/10.1126/science.1075390-a.

    • Search Google Scholar
    • Export Citation
  • Schmitt, R. W., P. S. Bogden, and C. E. Dorman, 1989: Evaporation minus precipitation and density fluxes for the North Atlantic. J. Phys. Oceanogr., 19, 12081221, https://doi.org/10.1175/1520-0485(1989)019<1208:EMPADF>2.0.CO;2.

    • Search Google Scholar
    • Export Citation
  • Smith, S. R., P. J. Hughes, and M. A. Bourassa, 2011: A comparison of nine monthly air–sea flux products. Int. J. Climatol., 31, 10021027, https://doi.org/10.1002/joc.2225.

    • Search Google Scholar
    • Export Citation
  • Tanimoto, Y., H. Nakamura, T. Kagimoto, and S. Yamane, 2003: An active role of extratropical sea surface temperature anomalies in determining anomalous turbulent heat flux. J. Geophys. Res., 108, 3304, https://doi.org/10.1029/2002JC001750.

    • Search Google Scholar
    • Export Citation
  • Tokinaga, H., and S.-P. Xie, 2011: Wave and Anemometer-based Sea surface Wind (WASWind) for climate change analysis. J. Climate, 24, 267285, https://doi.org/10.1175/2010JCLI3789.1.

    • Search Google Scholar
    • Export Citation
  • Trenberth, K. E., 2015: Has there been a hiatus? Science, 349, 691692, https://doi.org/10.1126/science.aac9225.

  • Wang, D., L. Zeng, X. Li, and P. Shi, 2013: Validation of satellite-derived daily latent heat flux over the South China Sea, compared with observations and five products. J. Atmos. Oceanic Technol., 30, 18201832, https://doi.org/10.1175/JTECH-D-12-00153.1.

    • Search Google Scholar
    • Export Citation
  • Wang, X., R. Zhang, J. Huang, L. Zeng, and F. Huang, 2017: Biases of five latent heat flux products and their impacts on mixed-layer temperature estimates in the South China Sea. J. Geophys. Res. Oceans, 122, 50885104, https://doi.org/10.1002/2016JC012332.

    • Search Google Scholar
    • Export Citation
  • Wentz, F. J., L. Ricciardulli, K. Hilburn, and C. Mears, 2007: How much more rain will global warming bring? Science, 317, 233235, https://doi.org/10.1126/science.1140746.

    • Search Google Scholar
    • Export Citation
  • Xie, S.-P., and Y. Kosaka, 2017: What caused the global surface warming hiatus of 1998–2013? Curr. Climate Change Rep., 3, 128140, https://doi.org/10.1007/s40641-017-0063-0.

    • Search Google Scholar
    • Export Citation
  • Young, I. R., and A. Ribal, 2019: Multiplatform evaluation of global trends in wind speed and wave height. Science, 364, 548552, https://doi.org/10.1126/science.aav9527.

    • Search Google Scholar
    • Export Citation
  • Yu, L., 2007: Global variations in oceanic evaporation (1958–2005): The role of the changing wind Speed. J. Climate, 20, 53765390, https://doi.org/10.1175/2007JCLI1714.1.

    • Search Google Scholar
    • Export Citation
  • Yu, L., and R. A. Weller, 2007: Objectively analyzed air–sea heat fluxes for the global ice-free oceans (1981–2005). Bull. Amer. Meteor. Soc., 88, 527540, https://doi.org/10.1175/BAMS-88-4-527.

    • Search Google Scholar
    • Export Citation
  • Yu, L., S. A. Josey, F. M. Bingham, and T. Lee, 2020: Intensification of the global water cycle and evidence from ocean salinity: A synthesis review. Ann. N. Y. Acad. Sci., 1472, 7694, https://doi.org/10.1111/nyas.14354.

    • Search Google Scholar
    • Export Citation
  • Zhang, R., X. Wang, and C. Wang, 2018: On the simulations of global oceanic latent heat flux in the CMIP5 multimodel ensemble. J. Climate, 31, 71117128, https://doi.org/10.1175/JCLI-D-17-0713.1.

    • Search Google Scholar
    • Export Citation
  • Zhou, L. T., G. Chen, and R. Wu, 2015: Change in surface latent heat flux and its association with tropical cyclone genesis in the western North Pacific. Theor. Appl. Climatol., 119, 221227, https://doi.org/10.1007/s00704-014-1096-0.

    • Search Google Scholar
    • Export Citation
  • Fig. 1.

    Time series of the anomalous (a) LHF, (b) U, and (c) ΔQ over tropical oceans (30°S–30°N, 0°–360°). The thin and thick straight lines denote the linear trends in 1979–98 and 1999–2018, respectively. The anomalies in each flux product are computed by removing their climatology mean values from 1979 to 2018. The two values in parentheses denote the linear trends before and after 1998, respectively. The bold italics indicate that the value passes the 95% confidence level. The units for the linear trends in (a), (b), and (c) are W m−2 yr−1, m s−1 per decade, and g kg−1 per decade, respectively. The green filled area denotes the period 1979–98. During this period, all the products show consistent signs of long-term trends in the LHF, U, and ΔQ.

  • Fig. 2.

    Time series of the anomalous LHF (black) and the sum of term U and term ΔQ (red). Term U and term ΔQ denote the first and second terms on the right-hand side of Eq. (2) and represent the contributions to the LHF variations from anomalous U and ΔQ, respectively. The bar denotes the difference between the black and red curves. The vertical gray line denotes the year 1998.

  • Fig. 3.

    Spatial distributions of the long-term trends in the LHF, U, and ΔQ in the period 1979–98. The results of the LHF from (a) ERA-Interim, (b) ERA5, (c) OAFlux, (d) NCEP-2, and (e) TropFlux. (f)–(o) As in (a)–(e), but for U and ΔQ, respectively. Contours denote values that pass the 95% confidence level.

  • Fig. 4.

    As in Fig. 3, but for the period 1999–2018.

  • Fig. 5.

    Distributions of the point-to-point CC between the LHF and term U in the periods (a)–(e) 1979–98 and (f)–(j) 1999–2018. Red and blue contours highlight the values of 0.7 and 0.3, respectively.

  • Fig. 6.

    As in Fig. 5, but for the results of term ΔQ.

  • Fig. 7.

    CCs between the LHF and both term U (effect of sea surface wind) and term ΔQ (effect of air–sea humidity difference). The light and dark green lines denote the 90% and 95% confidence levels, respectively.

  • Fig. 8.

    Comparisons between the flux products and buoy observations from 1999 to 2018. All data have been processed into anomalies by subtracting their corresponding climatological mean values. (a)–(e) LHF, (f)–(j) U, and (k)–(o) ΔQ. The green line denotes the reference line with slope 1, and the red line denotes the least squares fitting line.

  • Fig. 9.

    Long-term trend in U calculated at each buoy station for the (a) GTMBA data and (b)–(f) five flux products. There are 82 buoy stations employed in total. The numbers at the top left of (b)–(f) denote the counting number of buoy stations with the same sign of the long-term trend in U for both the flux product and buoy data.

  • Fig. 10.

    Time series of U variations from 1999 to 2018. The dashed–dotted curve denotes the anomalous values, and the solid curve denotes the least squares fitting line. The bar denotes the result from satellite data averaged over the entire tropics. The pink filled area denotes the ensemble mean uncertainty defined by one standard deviation among those five flux products. The value in parentheses denotes the linear trend (m s−1 per decade). The green value denotes the trend passing the 95% confidence level.

All Time Past Year Past 30 Days
Abstract Views 691 124 0
Full Text Views 229 85 7
PDF Downloads 283 93 7