Why does a decrease in cloud amount increase terrestrial evapotranspiration in a monsoon transition zone?

Terrestrial evapotranspiration plays a critical role in drought monitoring and water resource management. Changes in evapotranspiration are significantly influenced by cloud-related precipitation and radiation effects. However, the impact of cloud amount (CA) on evapotranspiration through its influence on precipitation remains uncertain, especially in the transition zone affected by the East Asian summer monsoon (EASM), which limits the understanding of the water cycle. Therefore, this study deeply explores the impact of CA on evapotranspiration and its potential physical mechanisms in Northwest China. The results show that the correlation between 31-year average evapotranspiration and CA is negative only in the semi-arid region and is positive in other climatic regions of Northwest China. This unique negative correlation is related to the change of precipitation pattern in the semi-arid region caused by the weak EASM. Smaller CA in weak monsoons results in more short-wave radiation reaching the surface, larger sensible heat, and weaker convective inhibition. Consequently, the proportion of convective clouds (CCs) increases and precipitation from these CCs enhances evapotranspiration. Less CA increases evapotranspiration and potentially exacerbates aridity in the semi-arid region of Northwest China. These results emphasize the role of cloud type in evapotranspiration. It is well known that global warming can change cloud type with more CCs. Therefore, this study sheds new light on evapotranspiration change under global warming.


Introduction
Evapotranspiration is the sum of all water entering the atmosphere, including the evaporation of water, land, and the transpiration of vegetation (Qiu et al 1999, Lian et al 2018, Wasti et al 2020).It supplies water vapor for the precipitation process (Mintz and Walker 1993, Jung et al 2010, Wei and Dirmeyer 2019), facilitating the formation of clouds and rainfall.And some studies suggest that rainfall also returns to the atmosphere in the form of evapotranspiration (De Dios et al 2015, Cui et al 2022).Therefore, evapotranspiration plays a crucial role in the water cycle, which is a fundamental and integral process in the Earth's climate system.In addition, evapotranspiration has a significant regulating effect on surface temperature (Sun et al 2016).As water evaporates, it absorbs heat from the surrounding environment, contributing to the maintenance of Earth's surface temperature balance (Carlson et al 1981, Chen et al 2021).Therefore, evapotranspiration also plays a crucial role in regulating regional climate characteristics.
Global and regional evapotranspiration have significantly changed in recent years owing to climate change (Valipour et al 2020).In comparison with other climatic regions, the semi-arid region exhibits a notable downward trend in evapotranspiration (Yang et al 2016, 2019, Jin et al 2019), indicating the uniqueness of evapotranspiration variation in semi-arid regions.Northwest China contains a typically semi-arid region, which serves as a transition zone influenced by the East Asian summer monsoon (EASM) and experiences a shift from a monsoon climate to a non-monsoon climate (Zhang et al 2016(Zhang et al , 2020)).Consequently, meteorological factors such as precipitation and temperature exhibit substantial spatial variations in this region (Zhang et al 2016).Additionally, surface characteristics, including vegetation and soil moisture, also experience significant spatial variability due to the impact of precipitation and temperature (Hua et al 2017, Wang et al 2022b).All these factors may lead to spatial differences in evapotranspiration.The changes in evapotranspiration and its influencing factors in the semiarid region of Northwest China are complex, necessitating a detailed study.
The factors influencing evapotranspiration can be divided into two types: water condition and energy condition.Firstly, precipitation has a significant effect on evapotranspiration by adjusting local moisture conditions.For example, Miralles et al (2014) pointed out that the decrease in evapotranspiration in central Australia can be attributed to the scarcity of land water supply resulting from decreased precipitation during El Niño.Similarly, Moura et al (2019) found that in years with lower precipitation, evapotranspiration would decrease due to insufficient water vapor supply caused by the presence of El Niño.Secondly, solar radiation and temperature, as energy conditions, drive surface evaporation and vegetation transpiration (Wang et al 2007, Zhang et al 2007, Herath et al 2018).Cloud amount (CA) represents the fraction or percentage of the sky that is covered by clouds at a particular location and time.The variability of CA impacts water conditions through its influence on precipitation by precipitating clouds and alters energy conditions; therefore, understanding CA is vital for connecting the factors influencing evapotranspiration.For example, global warming increases temperature, atmospheric moisture and precipitation intensity (Zhang et al 2021, Wang et al 2022a).This elevates soil moisture levels, ultimately promoting an increase in evapotranspiration (Jung et al 2010, Wang et al 2016, Helbig et al 2020, ISCCP 2023).However, a lot of observations have revealed a decrease in potential evaporation in numerous regions, known as the 'evaporation paradox' (Irmak et al 2012, Miralles et al 2014, Yang et al 2019, Bian et al 2020).Notably, in regions where global warming has led to an increase in CA (Abe et al 2016, Lohmann et al 2020), the increased CA reduces solar radiation and subsequently lowers potential evapotranspiration (PET) (Peterson et al 1995, Roderick andFarquhar 2002).In these regions, the impact of CA on radiation is a key factor in explaining the 'evaporation paradox' (Teuling et al 2013), and indirectly influencing actual evapotranspiration (Xing et al 2016).Therefore, the radiation effect of CA has been widely studied as an important factor to explain the 'evaporation paradox.'Some studies have shown that in monsoon areas, a strong monsoon increases water vapor, thereby raising CA, which can lead to increased precipitation and promote evapotranspiration.Conversely, a weak monsoon is not conducive to cloud formation, reducing precipitation and evapotranspiration (Zhang et al 2019, Yang et al 2022b).However, it remains uncertain how changes in CA affect evapotranspiration in monsoon transition areas, such as the semi-arid region of Northwest China.Hence, we chose the semi-arid region of Northwest China, a representative transition area, to investigate the connection between evapotranspiration and CA under the influence of varying monsoon intensities and explore the potential underlying mechanisms.This study aims to determine the (1) relationship between evapotranspiration and CA in the semi-arid region of Northwest China and how this differs from other climatic regions of Northwest China; (2) physical mechanisms that underlie this relationship; and (3) effects of cloud type and EASM strength on this relationship.These results have a potential to enhance our understanding of the water cycle related to evapotranspiration in the monsoon transition zone.(Rossow andSchiffer 1991, 1999).CA and convective clouds (CCs) are obtained from the ISCCP-Basic H-Series monthly dataset from 1984 to 2014, with a spatial resolution of 1 • × 1 • .Data are obtained from https://isccp.giss.nasa.gov/.

Data, methods and study area
Meteorological data including wind speed, sunshine duration, maximum temperature, and minimum temperature are collected from the China surface climatological data daily values dataset.

PET
As recommended by the Food and Agriculture Organization of the United Nations (FAO), the Penman-Monteith equation is used to calculate the PET (mm) (Allen et al 1998): where R n is the net radiation (MJ m −2 d −1 ), G is the soil heat flux (MJ m −2 d −1 ), γ is the dry and wet constant (Pa K −1 ), T is the average temperature ( • C), u 2 is the wind speed at a height of 2 m (m s −1 ), e s is the saturated and actual vapor pressure (kPa), e a is the actual vapor pressure (KPa), ∆ is the slope of the curve relating saturation water vapor pressure to the temperature (Pa K −1 ).

Evapotranspiration-precipitation coupling strength
The evapotranspiration-precipitation coupling index proposed by Zeng et al ( 2010) is used to diagnose the important influence of precipitation on evapotranspiration in Northwest China: Alternatively, it can be written as: where Γ is evapotranspiration-precipitation coupling, N is the total month or year; P ′ i and E ′ i are the outliers of precipitation and evapotranspiration, respectively, r P, E is the correlation coefficient between precipitation and evapotranspiration, and σ E and σ P are the standard deviation of evapotranspiration and precipitation, respectively.The index uses the relative magnitude of the covariance of precipitation and evapotranspiration and the variance of precipitation to reflect the influence of precipitation on evapotranspiration.The more consistent the rate of change between these two variables, the greater the range of change, and the stronger the land-atmosphere coupling.Positive and negative values also reflected the coupling relationship between evapotranspiration and precipitation.When Γ is negative, evapotranspiration is negatively correlated with precipitation, and an increase in precipitation inhibits evapotranspiration.Conversely, a positive Γ indicates a positive coupling between evapotranspiration and precipitation, with increased precipitation fostering greater evapotranspiration.When the coupling is very weak, Γ tends toward 0, indicating that precipitation exerts minimal influence on evapotranspiration.

Northernmost margin of the EASM
This study uses the monsoon marginal index, which combines air mass, precipitation, and wind (Hu and Qian 2007).The three criteria are as follows: first, the southwest pentad wind (wind averaged for a fiveday mean) at 850 hPa is larger than 0 m s −1 ; second, the average pentad pseudo-equivalent potential temperature at 850 hPa is larger than or equal to 335 K; finally, the average pentad precipitation is larger than or equal to 4 mm d −1 .The northernmost latitude that meets these three criteria is considered the northernmost margin of the EASM.The pseudo-equivalent potential temperature is calculated using the method proposed by Bolton (1980).

Study area and climatic regions
Northwest China includes five provinces: Shaanxi Province, Gansu Province, Qinghai Province, Ningxia Hui Autonomous Region, and Xinjiang Uyghur Autonomous Region, with latitudes and longitudes ranging from 31.32 to 49.10 • N and 73.15-111.15• E, respectively (figure S1).
Figure S1(a) shows the climatic regions of Northwest China according to the AI as shown in equation ( 2).Northwest China has five climate types: hyper-arid, arid, semi-arid, sub-humid, and humid.The hyper-arid and arid regions are the Gobi and Desert, respectively, with precipitation below 200 mm.These two climate regions account for 63.1% of the land cover of Northwest China.The semi-arid and sub-humid regions account for 28.4% and 4.8% of the land cover of Northwest China, respectively, with precipitation values ranging from 200-400 mm and 400-600 mm.The humid region is an area with precipitation greater than 600 mm, mainly located in the south of Shaanxi Province, accounting for 3.7% of the total area.Consequently, the study area is mainly arid and semi-arid, with only a small part being semihumid and humid.

Effect of CA on evapotranspiration
To explore the relationship between evapotranspiration and CA, figure 1(a) illustrates how evapotranspiration changes with increasing CA across various climatic regions, which are determined based on the GLDAS data.Except for the semi-arid region, a positive correlation between evapotranspiration and CA is observed in all the other climatic regions.This positive correlation is consistent with the results of many previous studies (Jiang et al 2018, Yang et al 2022b), and is supported by the common understanding that increasing CA is typically accompanied by higher precipitation and lower boundary layer height (Shi et al 2017, Khanna et al 2018).A higher precipitation and lower boundary layer height indicates a larger amount of water vapor for evapotranspiration (Meng et al 2014, Zhang et al 2020).Conversely, the negative correlation between evapotranspiration and CA in the semi-arid region is unique.Figure 1(b) shows the variation in evapotranspiration with the increase of CA in different climate regions based on ERA5 data.In the semi-arid region, a unique negative correlation between evapotranspiration and CA is also observed.Table S1 and figure S2 further shows the relationship between evapotranspiration and CA for each year.And table S2 presents correlation coefficient (r) and significant level (p) between evapotranspiration and CA from 1984 to 2014.A positive correlation between evapotranspiration and CA is evident in 9 years, while a negative correlation prevails over 22 years.During weak EASM, a robust negative correlation between evapotranspiration and CA is notable (e.g.1997, r = − 0.58, p < 0.01).Conversely, during strong EASM, evapotranspiration and CA are positively correlated (e.g.1988, r = 0.32, p < 0.01).Because the negative correlation dominates, the combination of these negative and positive correlations ultimately yields a negative overall correlation for 1984-2014.
The Bootstrap method is used to provide systematic uncertainty analysis and estimates, and to show the robustness of this negative correlation.First, and introduction to the Bootstrap method is given in text S1.Second, we apply the Bootstrap method to sample evapotranspiration and CA data in the semiarid region, with 200 samples each time.We conduct extraction 10, 100, and 1000 times to obtain corresponding Bootstrap samples.When we increase the number of resampling to 1000, the Bootstrap samples can get an accurate interval estimation.Figure S3 shows the Bootstrap probability density distribution of r after 1000 extraction times.The distribution of r values shows that 99.8% are less than 0, with only 0.2% greater than 0. The majority of r values are concentrated in the range of −0.3 to −0.1.Additionally, the percentage of Bootstrap sample r values estimated to be less than 0 reaches 99%, indicating the robustness of the negative correlation between evapotranspiration and CA in the semi-arid region.Table S3 also provides the average value of the Bootstrap sample r, p-value, 95% confidence interval of the Bootstrap sample r for 10, 100, and 1000 extractions.Regardless of the number of extractions, the average value of the Bootstrap sample r remains close to −0.20, with p less than 0.05.Additionally, the 95% confidence interval is less than 0.
Third, to ensure the accuracy of the negative correlation between evapotranspiration and CA in the semi-arid region, the ERA5 data are also used instead of the GLDAS data.We then conduct the same assessment using the Bootstrap method.Based on the Bootstrap probability density distribution of r after 1000 times of resampling, all sample r values are less than 0. The concentration area of r values in the ERA5 data is −0.40 to −0.24 (figure S4), indicating a more significant negative correlation between evapotranspiration and CA compared to that in the GLDAS data (figure S3).The specific instability results assessed with ERA5 data are presented in table S4.After 10, 100, and 1000 extractions, the mean value of r is always around −0.3, with p less than 0.05.Furthermore, the 95% confidence interval is consistently below 0. Now, the question is why the relationship between evapotranspiration and CA in the semi-arid region varies between years and why the negative correlation prevails.Numerous studies have emphasized that the semi-arid region of Northwest China represents a transition zone significantly influenced by the EASM (Yue et al 2022, Zhang et al 2022).Variations in the northernmost margin of the EASM can induce changes in water vapor, clouds and precipitation patterns in Northwest China, subsequently influencing the relationship between evapotranspiration and CA (Ren et al 2021).To elucidate this relationship, figure 2 illustrates the temporal variation in the northernmost margin of the EASM at 110 • E, which  By comparing table S1 and figure 2, it is evident that years displaying a positive CAevapotranspiration relationship coincide with the northward shift of the northernmost margin of the EASM, whereas the negative CA-evapotranspiration relationship corresponds to the southward shift.Therefore, the EASM intensity plays an important role in adjusting the influence of CA on evapotranspiration.The years 1988, 1999and 2011(strong EASM), and 1986, 1997and 2013 (weak EASM) are selected for an in-depth exploration to clarify the reasons behind the different effects of CA on evapotranspiration in the semi-arid region.

Physical mechanisms of CA affecting evapotranspiration in the semi-arid region 3.2.1. Effects of CA on precipitation during strong/weak EASM periods
To determine the significance of precipitation on evapotranspiration in Northwest China, the spatial variation characteristics of Γ are presented in figure S1(b).In Northwest China, the spatial pattern of Γ exhibits distinct transitional characteristics, with values decreasing from the northwest to the southeast.This value is consistently greater than zero throughout the entire region.This demonstrates a close relationship between evapotranspiration and precipitation in Northwest China, where increased precipitation significantly boosts ET.The conclusion is consistent with the results of previous studies (e.g.Yang et al 2017, 2022a).
As mentioned above, evapotranspiration increases with the rise of CA in the semi-arid region during strong EASM (figure 1(c)); however, the correlation between evapotranspiration and CA is negative during weak EASM (figure 1(d)).As the variation in evapotranspiration in Northwest China is primarily influenced by precipitation (figure S1(b)), the effects of precipitation on evapotranspiration at different EASM intensities should be considered.Notably, evapotranspiration in all climatic regions shows an increasing trend with precipitation for both strong and weak EASM (figures 3(a) and (b)).
Moreover, CA, as a significant determinant of precipitation, exerts a crucial effect on precipitation and consequently influences evapotranspiration.
Regardless of the strength of the EASM, a remarkable positive correlation between precipitation and CA is evident in all climatic regions, except for the semiarid region.Furthermore, the correlation between precipitation and CA is positive for the strong EASM and negative for the weak EASM in the semiarid region (figures 3(c) and (d)).Comparison of figures 3(c), (d) and 1(c), (d) indicates that the unique evapotranspiration-CA relationship in the semi-arid region stems from the impact of CA on precipitation.Previous studies have traditionally indicated that an increase in CA typically accompanies an increase in precipitation (Stephens et al 2008, Fu et al 2020, Li et al 2024).Nonetheless, the above results reveal a negative correlation between precipitation and CA.The specific reasons are explained below.

More convective precipitation during weak EASM period
Figure S5 illustrates the differences in the annual averages of various physical quantities between strong and weak EASM in different climatic regions.As expected, the CA is smaller in the weak EASM than in the strong EASM (figure S5(a)) and the net short-wave radiation flux (SW) is larger.The difference in SW between weak and strong EASM (12.81 W m −2 ) is most pronounced in the semi-arid region (figure S5(b)).Higher SW levels generally contribute to increased SH exchange (Ban-Weiss et al 2011, Chen et al 2020, Schwartz et al 2020).Therefore, the most difference in SH is observed in the semi-arid region in both of the datasets although the difference in SH in the GLDAS data is notably greater than that in the ERA5 (figures S5(c) and (d)).The thermal condition is an important factor in regulating precipitation (Wang et al 2023). Ford et al (2015) found that significant increases in SH weakened convective inhibition (CIN) and promoted convective precipitation under drought conditions.Similar conclusions have been reached in other studies (Petrova et al 2018, Yang et al 2022b).Therefore, SH is a key factor affecting CIN; the largest difference in SH also corresponds to the largest negative difference of CIN in the semi-arid region, i.e. −73.69 J kg −1 (figure S5(e)).This suggests that exceptionally low CIN levels in the semiarid region of the weak EASM create favorable conditions for convective processes.
During the weak EASM period, the negative correlation between the proportion of CCs and CA in the semi-arid region indicates that CCs is more likely to occur when CA is lower (figure S6(b)), owing to its lower CIN, as discussed above.Figure S6 (d) shows a positive correlation between the proportion of CCs and precipitation in the semi-arid region, signifying that a larger CCs induced by intensified thermal conditions dominantly drives the increase of precipitation.Therefore, CA and precipitation are negatively correlated in the semi-arid region during the weak EASM period (figure 3(d)).In contrast, during the strong EASM period, the correlations between the proportion of CCs and CA, and between the proportion of CCs and precipitation are both positive (figures S6(a) and (c)), resulting in a positive correlation between CA and precipitation (figure 3(c)).Because the negative correlation between the proportion of CCs and CA dominates in the semi-arid region from 1984 to 2014 (table S1), the correlation between CA and precipitation is also dominantly negative (table S1).Because of the positive correlation between evapotranspiration and precipitation (figure S1(b)), evapotranspiration and CA are negatively correlated in the semi-arid region (figure 1(a)).
Previous studies have shown that precipitation fluctuations are important factors affecting evapotranspiration in the semi-arid region (Zhang et al 2019, 2020, Wang et al 2022b).Nevertheless, it is crucial to emphasize that changes in precipitation are not solely attributable to variations in CA. the modification of cloud type also contributes significantly to this phenomenon.Cloud types are also important factors affecting precipitation because various cloud types have distinct physical mechanisms for precipitation (Lu et al 2023, Zhu et al 2024).For instance, Myoung and Nielsen-Gammon (2010) found that dry conditions may promote convective precipitation and augment local precipitation despite exhibiting unfavorable CA trends.Moreover, several studies have suggested that CCs contribute substantially to the local precipitation under arid conditions (Guillod et al 2015, Petrova et al 2018).Consequently, the influence of cloud type on evapotranspiration is very important.In this study, we shed light on cloudtype alterations as the principal driver of precipitation fluctuation in the semi-arid region, highlighting its pivotal role as an intermediary variable influencing evapotranspiration.

Concluding remarks
Terrestrial evapotranspiration is very important for water cycle.Clouds play an important role in evapotranspiration by affecting precipitation and radiation.However, the relationship between evapotranspiration and CA remains uncertain, especially in the semiarid region of Northwest China, which is the transition zone of the EASM.Thus, we investigate the influence of CA on evapotranspiration from 1984 to 2014 in Northwest China and discuss the physical mechanisms underlying the unique relationship between CA and evapotranspiration in the semi-arid region.The main conclusions are summarized as follows.
The 31 year average evapotranspiration and CA exhibits a positive correlation in the hyper-arid, arid, sub-humid and humid regions of Northwest China and a unique negative correlation in the semi-arid region owing to the dominance of negative correlations in 22 years over the positive correlations in 9 years.The 22 years of negative correlation and 9 years of positive correlation correspond to weak and strong EASM, respectively.This reflects the unique drought transition mechanism in the semiarid region of Northwest China under different monsoon intensities.When the EASM is weak, the reduction in CA enhances SW and SH.Concurrently, intensified thermal conditions reduce CIN, promoting the formation of CCs.This phenomenon is particularly pronounced in areas with low CA, leading to a shift in cloud type toward CCs and an increase in precipitation.Therefore, despite the reduction in CA, the increasing ratio of CCs cover to total CA enhances the influence of convective precipitation on evapotranspiration, thereby giving rise to a unique negative correlation between evapotranspiration and CA.When the EASM is strong, the thermal conditions are weak, and the cloud type cannot easily change.Therefore, the correlations between the proportion of CCs and CA, and between the proportion of CCs and precipitation are both positive, leading to a positive correlation between evapotranspiration and CA.To summarize these physical mechanisms, we have presented a conceptual diagram that elucidates the influence of CA on evapotranspiration in various climatic regions, with an emphasis on the mechanisms under different EASM intensities in the semi-arid region (figure 4).
Our study provides a deeper understanding of how CA affects evapotranspiration and reveals a unique water cycling mechanism in the semi-arid region, where less CA may promote water cycling.This study emphasizes the importance of considering the combined effects of CA, radiation, and precipitation when analyzing evapotranspiration.Furthermore, it highlights the need for future research on the impact of changes in cloud type on evapotranspiration.Many previous studies (Mitchell and Finnegan 2009, Ceppi et al 2017, Su et al 2017, Noda et al 2023) found that there will be more CCs under global warming.Therefore, the results in this study provides a good reference on how evapotranspiration is affected by CA and cloud type.

Figure 1 .
Figure 1.Variation in evapotranspiration (ET) with the increase of cloud amount (CA) in different climatic regions in Northwest China.(a) GLDAS data from 1984 to 2014, (b) ERA5 data from 1984 to 2014, (c) GLDAS data in 1988, 1999 and 2011 (strong East Asian summer monsoon, EASM), and (d) GLDAS data in 1986, 1997 and 2013 (weak EASM).r and p represent correlation coefficient and significant level, respectively.

Figure 2 .
Figure 2. Annual latitude variation in the northernmost margin of the East Asian summer monsoon (EASM) at 110 • E from 1982 to 2014.

Figure 4 .
Figure 4. Conceptual image of the influencing mechanism of cloud amount (CA) on evapotranspiration (ET) in different climatic regions in Northwest China.SH, SW, Pre, CIN and CC represent sensible heat flux, net short-wave radiation flux, convective inhibition, and proportion of convective clouds, respectively.