1. Introduction
Evaporation plays an important role in the exchange and cycling of energy, water, and heat in the atmosphere, and long-term measurements of its annual rates can serve as an excellent indicator of the intensity of the water cycle (Brutsaert and Parlange 1998; Huntington 2006; Roderick and Farquhar 2002). Significant progress has been made in actual evapotranspiration observation techniques, particularly with eddy covariance measurements, offering advantages of high temporal resolution and broad applicability for advancing evaporation research (Cunliffe et al. 2022; Xue et al. 2023). However, practical adoption may be hindered by factors such as high cost, complex equipment requirements, and dependence on reference flux data. For instance, the ChinaFLUX observation network (http://www.chinaflux.org/enn/index.aspx), with 83 towers, faces challenges in providing long-term and large-scale datasets due to its relatively short construction period. Observing evapotranspiration presents challenges, and the commonly used estimation methods such as Penman’s hypothesis (Allen et al. 1998; Penman 1948), complementary theory (Brutsaert and Parlange 1998), and the coupled water–energy balance (Budyko 1974; Fu 1981; Yang et al. 2006) rely on potential evapotranspiration as a crucial parameter. Pan evaporation (Epan) is widely employed to estimate potential evapotranspiration, wherein the observed evaporation rate is multiplied by an empirically derived Epan coefficient (Abtew et al. 2011; Lugato et al. 2013; McMahon et al. 2013). Meanwhile, notwithstanding the difference between Epan and the evapotranspiration of cropped surfaces, the pan has proved its practical value and has been used successfully to estimate reference crop evapotranspiration (ETo) by observing the evaporation loss from a water surface and applying an empirical pan coefficient to relate Epan to ETo, which is one of the methods recommended by the Food and Agriculture Organization (FAO) (Allen et al. 1998). Moreover, it is essential to acknowledge that the pan coefficient is influenced by local climate and physical conditions (Chiew et al. 1995; Sabziparvar et al. 2010). Observation networks of Epan have been established globally at meteorological stations since the 1950s, owing to the pan’s simplicity, low cost, and wide variety of applications (Lim et al. 2016, 2013; Wang and Dickinson 2012). Earth’s water cycle is currently undergoing modifications, and it is anticipated that these changes will become more pronounced as the climate continues to warm (Elbaum et al. 2022).
Pans provide a measurement of the integrated effect of radiation, wind, temperature, and humidity on the evaporation from an open water surface. As the average global temperature increases, it is generally expected that the air will become drier and that evaporation from terrestrial water bodies will increase (Limjirakan and Limsakul 2012; Roderick and Farquhar 2002). Paradoxically, decreases in Epan have been observed in the past few decades across the world with different climates between the 1950s and 1990s, with a decreasing rate from −1 to −4 mm yr−2 (Brutsaert and Parlange 1998; Limjirakan and Limsakul 2012; Matsoukas et al. 2011; Peterson et al. 1995; Roderick et al. 2009, 2007; Stephens et al. 2018; Wang and Dickinson 2012). Several key physical drivers of evaporation include temperature, radiation, vapor pressure deficit (VPD), and wind speed (Penman 1948; Roderick et al. 2007; Stephens et al. 2018). Initially, Peterson et al. (1995) proposed that the decline in Epan was a result of reduced solar radiation due to an increase in aerosol concentration and cloud cover. A complementary theory was presented by Brutsaert and Parlange (1998), highlighting that the decline in Epan could be a sign of increased regional moisture availability, resulting in decreased sensible heat and energy available for Epan. However, further studies by Roderick et al. (2007) suggested that the primary cause of Epan decreases was due to reductions in wind speed, which may not always remain the dominant driver of Epan changes. While Epan had been decreasing for several decades, recent studies have shown that these trends have either plateaued or reversed in some regions, which has been linked to an increase in VPD (Li et al. 2013; Stephens et al. 2018). VPD describes the difference between the water vapor pressure at saturation and the actual water vapor pressure for a given temperature. The increasing trend of global temperatures in recent decades has caused a significant increase in land surface VPD, resulting in an increased atmospheric demand for evaporation water (Grossiord et al. 2020; Massmann et al. 2019; Monteith 1965; Penman 1948; Yuan et al. 2019).
The D20 pans were the primary choice for evaporation observations from the 1950s until 2001 in China. Subsequently, the E-601B pans have replaced the D20 pans in most stations. The Epan observed by E601 and D20 has a good linear relationship, which is the most direct method for reconstructing the evaporation data. However, there were few stations that observed Epan of E601 and D20 simultaneously. Moreover, the E-601B pans are unsuitable for observation when evaporation rates are low or when water freezes during winter, except for those located in northern regions. Utilization of different types of evaporation pans has resulted in a lack of continuous observations in recent years, making it necessary to explore effective reconstruction techniques to supplement Epan observations (Wang et al. 2019). To estimate atmospheric evaporation demand, several physical models have been developed, including the Penman–Monteith model, which is recommended by the Food and Agriculture Organization (Allen et al. 1998), as well as the Priestley–Taylor (Kingston et al. 2009; Priestley and Taylor 1972) and Hargreaves–Samani (Thompson et al. 2014) models. With the advancement of computing technology, machine learning methods, such as random forest (Lu et al. 2018), artificial neural networks (Goyal et al. 2014; Keskin et al. 2009; Kim et al. 2012), multivariate adaptive regression splines (Ghaemi et al. 2019), and support vector machines (Kisi 2015), have demonstrated the potential in reconstructing Epan (Guan et al. 2020; Kisi and Heddam 2019; Qasem et al. 2019; Seifi and Soroush 2020). However, it is important to consider their limitations, particularly in scenarios where multiple complex factors influence evaporation and their poor interpretability or “black box” phenomenon makes attribution challenging (Coyle and Weller 2020; Yu et al. 2021). Therefore, a physically based tool, called PenPan, for reconstructing the monthly Epan was developed (Roderick et al. 2007; Rotstayn et al. 2006). The model is grounded on mass and energy balances and is based on the Penman equation for potential evapotranspiration (Penman 1948). It incorporates the aerodynamic and radiative components based on Thom et al. (1981) and Linacre (1994), respectively. The PenPan model has undergone extensive validation and has been adeptly adapted for use with the D20 evaporation pan in China (Feng et al. 2018; Liu and Sun 2016; Wang et al. 2018c). Additionally, the model’s efficacy has also been demonstrated in its ability to perform attribution analysis.
Based on the PenPan model, this study is to construct an Epan dataset for China, to capture its temporal and spatial variations, and to examine the underlying factors that drive these variations. In section 2, we provide a description of the study’s data and methods. In section 3, we simulate Epan and analyze its spatiotemporal pattern and attribution of variations. Last, the conclusions are drawn.
2. Data and methods
Figure 1a displays the spatial locations of meteorological stations used in this study. The daily surface data for 1951–2020 were collected from 2410 meteorological stations available at the China Meteorological Data Service Center (CMDC) (http://data.cma.cn/). The meteorological parameters considered in this study included atmospheric pressure, sunshine hours, relative humidity, air temperature measured 2 m above ground level, wind speed measured at a height of 10 m above ground level, and Epan. The Epan was measured using a D20 pan with a diameter of 20 cm and a height of 10 cm (Fig. 1b). All collected data underwent strict quality control measures, including data integrity check, climate consistency check, internal consistency check, time consistency check, space consistency check, multisource data consistency check, and human–computer interaction check, among others (http://data.cma.cn/). After excluding stations that were impacted by site relocation, resulting in changes to wind speed and other meteorological parameters, a total of 1786 stations were identified for inclusion in the study. To minimize the statistical deviation due to nonuniform station distribution, we defined grid cells with a spatial resolution of 1° × 1° and identified effective grid cells that contained at least one station. A total of 641 effective grids were selected. For a grid containing more than one observation station, the average was used for calculation and analysis.
(a) The locations of meteorological stations and selected grids, containing at least one station each (spatial resolution of 1° × 1°), and (b) the D20 pan.
Citation: Journal of Hydrometeorology 24, 11; 10.1175/JHM-D-23-0066.1
3. Results and discussion
a. Validation of the monthly pan evaporation simulation based on the PenPan model
Based on the PenPan model, the monthly Epan in China was calculated and the scatter diagram between the calculated and observed Epan is shown in Fig. 2a. The diagram shows a close agreement between the observed and calculated values, with the majority of points lying near the 1:1 line. A linear relationship was observed between the calculated and observed Epan, with y = 0.98x + 7.70, R2 = 0.93 ± 0.045, and RMSE = 21.48 ± 6.06 mm month−1. The spatial distribution of R2 is shown in Fig. 2b, which indicates that 98.28% of grids have R2 values greater than 0.80, whereas 81.41% of grids have R2 values greater than 0.90. The grids with small R2 values were mainly found in the southwest region of China, with the smallest R2 value being no less than 0.66. The spatial distribution of RMSE is shown in Fig. 2c, which shows that the grids with RMSE greater than 30 mm month−1 were primarily located in the northwestern region, where the annual Epan is approximately 2000 mm yr−1 (Fig. 2d) and RMSE accounted for 1.78% on average.
The statistical comparison between the observed and calculated monthly pan evaporation (Epan) over China. (a) The scatter density plot between the observed values and the values calculated by the PenPan model during 1951–2021. The legends are the counts, with the red color representing more and the blue color representing less. The (b) R2 and (c) RMSE values between the simulation and observation of Epan during 1951–2021. (d) The annual average Epan during 1961–2021.
Citation: Journal of Hydrometeorology 24, 11; 10.1175/JHM-D-23-0066.1
The results exhibited performance on par with the Epan prediction reported by Wang et al. (2018c) and Li et al. (2013), with R2 and RMSE values of 0.92 and 26.03 mm month−1 over China and 0.94 and 31 mm month−1 in Xinjiang, respectively. Despite the slightly lower accuracy compared to some machine learning methods, such as the random forest model used in Al-Mukhtar’s (2021) study in Iraq (R2 = 0.99; RMSE = 19.88 mm month−1) and Lu et al.’s (2018) study in Poyang Lake (R2 = 0.96; RMSE = 0.35 mm day−1), these results still provide valuable information for attribution analysis. Therefore, the PenPan model can be considered accurate in predicting the monthly Epan values for the entire country based on meteorological factors such as wind speed, atmospheric pressure, relative humidity, air temperature, and sunshine hours.
b. The spatiotemporal patterns of pan evaporation in the recent 70 years
Figure 3 shows the significance of the annual Epan change. The figure displays red grids for a significant decrease, yellow grids for a significant increase, blue grids for no statistically significant changes, and blank grids for incomplete data. Considering the progressive initiation of Epan observations in 1951 and the gradual expansion of site coverage until 1961, we have chosen data spanning each year from 1951 to 1961 through 2021 for trend and attribution analysis to assess result robustness. Extending the study period back to 1951 enables the capture of a broader range of climate variability and cyclic changes, as well as their impacts on evaporation. Figures 3a–i show the spatial distribution of Epan change trends with complete years of data from 1951 to 2021, from 1952 to 2021, …, and from 1961 to 2021, respectively. It can be seen that due to the different start time of station monitoring, the grids with different years have different distributions. The grids with effective data from 1951 to 2021 are the least, whereas those with data from 1961 to 2021 are the most. While the number of spatial grids with available data varies across the different time periods, the results consistently show a significant decrease in Epan across most regions, with a smaller proportion of grids showing a significant increase and a relatively constant proportion showing no significant change over time. On average, 46.85% of the grids showed no significant changed trend, 42.59% showed a significant decrease, and 10.56% showed a significant increase. From 1951 to 2021, there were 31 effective grids, and among them, 70.97% showed no significant change trend, 25.81% showed a significant decrease, and 3.23% showed a significant increase (Fig. 3a). The number of effective grids increased to 190 during 1954–2021, among which 46.32% showed no significant change trend, 41.05% showed a significant decrease, and 12.63% showed a significant increase (Fig. 3d). During 1961–2021, the number of effective grids increased to 580, of which 51.03% showed no significant change trend, 37.76% showed a significant decrease trend, and 11.21% showed a significant increase (Fig. 3i). The significance of the average annual Epan change and its average linear change rate during different periods are shown in Table S1 in the online supplemental material.
The significance of the annual Epan change. Significance is for the 90% confidence level and controlling the FDR with αFDR = 0.1. The red grids represent a significant decrease, the yellow grids represent a significant increase, the blue grids represent no statistically significant changes, and the blank grids represent incomplete data. (a) 1951–2021; (b) 1952–2021; (c) 1953–2021; (d) 1954–2021; (e) 1955–2021; (f) 1956–2021; (g) 1958–2021; (h) 1960–2021; (i) 1961–2021.
Citation: Journal of Hydrometeorology 24, 11; 10.1175/JHM-D-23-0066.1
Although Epan has shown a decreasing trend or no significant change over the past 6–7 decades, there have been shifts in the annual average Epan trends around 1961 and 1993. Figure 4 shows the average linear change rate of Epan and its aerodynamic component (Epa) and radiative component (Epr) during 1961–93, 1994–2021, and 1961–2021. The red dot represents a decreasing trend and the blue dot represents an increasing trend, with larger dots indicating greater change rates. It can be seen that most grids showed a decreasing trend during 1961–2021 (Fig. 4a). During 1961–93, most grids (83.79%) showed a decreasing trend (Fig. 4b), while during 1994–2021, most grids (62.59%) showed an increasing trend (Fig. 4c). Similarly, from 1954 to 2021, most grids showed a decreasing trend (Fig. S1a). Although most grids (78.95% showed a decreasing trend from 1954 to 1993 (Fig. S1b), from 1994 to 2021 most grids (58.42% showed an increasing trend (Fig. S1c). The Epan also showed similar trends during different periods, such as 1951–2021, 1952–2021, …, and 1960–2021. Hence, it can be seen that most grids showed a significant decreasing trend before 1993 but began to show an increasing trend after 1994, which has slowed down the Epan trend in the past 60–70 years. Moreover, Fig. 4d shows the linear change rate of Epa from 1961 to 2021, indicating that most grids showed a decreasing trend. Similar to the Epan trend, Epr decreased in most grids (76.03%) during 1961–93 (Fig. 4e), while an increasing trend was observed in most grids (62.59%) during 1994–2021 (Fig. 4f). Furthermore, Fig. 4g shows the linear change rate of Epr during 1961–2021. It can be seen that Epr also showed a decreasing trend in most grids during 1961–93 and an increasing trend during 1994–2021. The Epr exhibits stronger regionalism, which can be attributed to the higher solar radiation in northwest China than in southeast China, and the solar radiation began to reverse from decline to rise around the 1990s (Wang et al. 2018a).
The average linear change rate of Epan and its aerodynamic component (Epa) and radiative component (Epr) during 1961–2021, 1961–93, and 1994–2021. The red dot represents a decreasing trend and the blue dot represents an increasing trend, with larger dots indicating greater change rates. (a)–(c) Epan during 1961–2021, 1961–93, and 1994–2021; (d)–(f) Epa during 1961–2021, 1961–93, and 1994–2021; (g)–(i) Epr during 1961–2021, 1961–93, and 1994–2021.
Citation: Journal of Hydrometeorology 24, 11; 10.1175/JHM-D-23-0066.1
The annual average Epan showed a decreasing trend (0.1 level) over the past 61–71 years, with a decreasing rate ranging from 1.02 to 1.29 mm yr−2 and an average of 1.13 mm yr−2 (Fig. 5a). In the period before 1993 (1951–93, 1952–93, …, and 1961–93), Epan also showed a statistically significant decreasing trend (0.1 level), with the decrease rate first increasing and then stabilizing as the number of years increases. Specifically, the decrease rate was 1.95 mm yr−2 during 1951–93 and gradually stabilized to about 5.00 mm yr−2 during 1954–93. This was mainly because Epan showed an increasing trend before 1961, primarily due to the offsetting of increasing and decreasing trends before and after 1961. Thus, 1961 and 1993 can be used as subsection points to divide annual Epan into three periods, namely, 1951–2021, 1954–2021, and 1961–2021. Epan showed an increasing trend during 1951–61 and 1954–61, with an increasing rate of 15.28 and 3.83 mm yr−2, respectively, while during 1961–93, Epan showed a significant decreasing trend with slightly different rates due to different effective regions in different periods (Fig. 5b). From 1994 to 2021, Epan showed an increasing trend in 11 periods (1951–2021, 1952–2021, …, and 1961–2021), with an increasing rate ranging from 0.05 to 1.63 mm yr−2 and an average of 1.11 mm yr−2 (Fig. 5b). The Epa also showed a significant (0.1 level) decreasing trend during 1951–2021 (1951–2021, 1952–2021, …, 1961–2021), with the period before 1993 showing a significant decrease and the period 1994–2021 showing a significant increase or no statistically significant change (Fig. 5c). Epr showed no statistically significant changes in the recent 61–71 years, showing significant decreasing trends before 1993 with a small decreasing rate and increasing trends from 1994 to 2021 without any statistically significant changes (Fig. 5d). Overall, the change trend of Epa is more significant and has a decisive effect on the change trend of Epan. The average reduction rate of Epan was 1.08 mm yr−2, with the contribution of Epa reduction accounting for 82.75%, followed by Epr. Further analysis was conducted to determine the contribution of the main influencing factors of Epa.
The annual average Epan and its aerodynamic component (Epa) and radiative component (Epr) over the past 61–71 years. The blue line represents the period 1951–2021, the black line represents the period 1954–2021, the red line represents the period 1961–2021, and the gray lines with different shades represent other periods.
Citation: Journal of Hydrometeorology 24, 11; 10.1175/JHM-D-23-0066.1
c. Attribution recognition of the pan evaporation variations
The results indicated that the average rate of Epan decreased from 1961 to 2021 at a rate of −1.06 mm yr−2. It is noteworthy that between 1961 and 1993, the average rate of Epan decrease was −4.59 mm yr−2, with 65.40% of the contribution coming from Epa and 34.60% from Epr. However, from 1994 to 2021, Epan showed a reversal and increased at an average rate of 1.63 mm yr−2, with Epa contributing 75.58% and Epr contributing 24.42%. The study further decomposed the contributions of wind speed, VPD, and temperature within the Epa during the entire period (1961–2021) as well as the subperiods of 1961–93 and 1994–2021, as depicted in Fig. 6. The red dots represent negative contributions, the blue dots represent positive contributions, and the size of the dots denotes the magnitude of the contribution.
The contributions of wind speed, vapor pressure deficit (VPD), and air temperature during 1961–2021 as well as the subperiods of 1961–93 and 1994–2021. The red dot represents a decreasing trend and the blue dot represents an increasing trend, with larger dots indicating greater contributions. (a)–(c) The contributions of wind speed (U*) during 1961–2021, 1961–93, and 1994–2021. (d)–(f) The contributions of VPD (D*) during 1961–2021, 1961–93, and 1994–2021. (g)–(i) The contributions of air temperature (T*) during 1961–2021, 1961–93, and 1994–2021.
Citation: Journal of Hydrometeorology 24, 11; 10.1175/JHM-D-23-0066.1
Figure 6a shows the contribution of wind speed in different regions during 1961–2021. The results indicated that wind speed predominantly has a negative impact on Epan, with an average contribution of −2.14 mm yr−2. Furthermore, the reduction in wind speed contribution in the north was more substantial than that in the south. Most grids made negative contributions during 1961–93, with an average value of −2.68 mm yr−2 (Fig. 6b). However, from 1994 to 2021, the average contribution of wind speed was −0.20 mm yr−2, and many grids changed from negative to positive contributions (Fig. 6c). These findings suggest that the influence of wind speed on evaporation is weakening over time, mainly due to the continuous decrease in wind speed during 1961–93, but its stagnation and reversal upward during 1994–2021 (Fig. 7d). Figure 6d shows the contribution of VPD in different regions during 1961–2021. The results indicated that VPD makes a positive contribution to Epan in most grids, with an average value of 1.32 mm yr−2, and the contribution is greater in northern and coastal areas. During 1961–93, VPD had a negative contribution in some areas and a positive contribution in others (Fig. 6e), with an average value of −0.35 mm yr−2. However, from 1994 to 2021, most regions had positive contributions (Fig. 6f), with an average value of 1.61 mm yr−2. The contribution of VPD increased over time, primarily due to the slight decrease in VPD during 1961–93, but a significant increase during 1994–2021 (Fig. 7e). Figures 6g–i show the spatial distribution of temperature contribution to Epan change during 1961–2021, 1961–93, and 1994–2021, respectively. The temperature had a negative contribution to Epan during all three periods, with an average contribution from −0.08 to −0.04 mm yr−2.
The annual Epan and its driving factors wind speed and VPD. The blue line represents the Epan, the black line represents the wind speed, and the red line represents the VPD. (a)–(c) 1951–2021; (d)–(f) 1961–2021. (top) The Epan vs wind speed; (middle) Epan vs VPD; (bottom) wind speed vs VPD.
Citation: Journal of Hydrometeorology 24, 11; 10.1175/JHM-D-23-0066.1
Wind speed, VPD, and temperature in the other 10 periods (1951–2021, 1952–2021, …, and 1960–2021) showed a similar positive and negative contribution trend to Epan change as the period 1961–2021, with slightly different contribution values. For instance, Fig. S2 illustrates the contribution values for the period 1954–2021. The average annual reduction rate of Epan was 1.08 mm yr−2, with an average reduction rate of Epa of 0.89 mm yr−2 in recent 61–71 years of 11 periods. The average contributions of wind speed, VPD, and temperature to Epan reduction were −2.11, 1.32, and −0.08 mm yr−2, respectively. Before 1993, the average annual reduction rate of Epan was 3.48 mm yr−2, with an average reduction rate of Epa of 2.16 mm yr−2. Wind speed and VPD had contributions to the reduction of Epan of −2.21 and −0.01 mm yr−2, respectively. However, the period before 1993 was further divided into two periods: before 1961 and after 1961. Before 1961, both wind speed and VPD made positive contributions to Epan (Figs. 7a,b). During 1961–93, the average reduction rate of Epan was 5.11 mm yr−2, with an average reduction rate of Epa of 3.24 mm yr−2. Wind speed and VPD had an average contribution to the reduction of Epan of −2.88 and −0.37 mm yr−2, respectively. From 1994 to 2021, Epan began to rise with an average increase rate of 1.07 mm yr−2, and the average increase rate of Epa was 0.69 mm yr−2. The average contributions of wind speed and VPD were −0.67 and 1.50 mm yr−2, respectively. It can be seen that before 1961 the increase in wind speed and VPD promoted evaporation. During 1961–93, a slight decrease in VPD, coupled with a significant decrease in wind speed, led to a reduction in evaporation. During 1994–2021, the stagnation of wind speed, coupled with an increase in VPD, promoted an increase in evaporation (Fig. 7).
4. Conclusions
There has been a decreasing trend observed in Epan over the past 71 years. The first period (1951–60) exhibited an increasing trend, whereas the second period (1961–93) displayed a decreasing trend. Notably, during the third period (1994–2021), there was an increase in the trend. The trend of decreasing Epan is more prominent in the spatial distribution of the data than the trend of increasing Epan. The regions with a significant decreasing trend are mainly distributed in the northern and western regions of China, whereas the regions with a significant increasing trend are mainly distributed in the eastern and southern regions. These suggest that the changes in Epan are likely related to regional climate and environmental conditions. It is important to continue monitoring and analyzing the changes in Epan in order to better understand the impacts of climate change and human activities on water resources and ecosystems.
The radiation, wind speed, VPD, and temperature are important factors affecting Epan, and their contributions vary over different time periods. Notably, the dynamic term was found to contribute more than 60% to Epan. Further decomposition of the contributions of wind speed, VPD, and temperature was conducted within the aerodynamic component. The period before 1961 showed a positive contribution of wind speed and VPD to Epan, whereas the period from 1961 to 1993 showed a significant decrease due to a decrease in wind speed. The contribution of wind speed was consistently negative, indicating that slower winds reduce evaporation. From 1994 to 2021, the increase in VPD promoted the increase of Epan, suggesting that drier air promotes evaporation. These findings suggest that changes in climate variables can have significant impacts on evaporation, which can in turn affect water availability and ecosystem functioning. Nonetheless, there remains a considerable degree of uncertainty regarding trends in climate variables, and the available data are limited in terms of spatial and temporal coverage. To effectively manage water resources and mitigate the impacts of climate change, it is crucial to comprehend the relationships between evaporation and its drivers.
Acknowledgments.
This study was supported by the National Natural Science Foundation of China (42025104 and 42022005), the Program for the “Kezhen-Bingwei” Youth Talents (2021RC002) from the Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, and the Top-Notch Young Talents Program of China (Fubao Sun).
Data availability statement.
The meteorological data used in this study were obtained from meteorological stations of the China Meteorological Data Service Center (CMDC) (http://data.cma.cn/).
REFERENCES
Abtew, W., J. Obeysekera, and N. Iricanin, 2011: Pan evaporation and potential evapotranspiration trends in South Florida. Hydrol. Processes, 25, 958–969, https://doi.org/10.1002/hyp.7887.
Al-Mukhtar, M., 2021: Modeling the monthly pan evaporation rates using artificial intelligence methods: A case study in Iraq. Environ. Earth Sci., 80, 39, https://doi.org/10.1007/s12665-020-09337-0.
Allen, R. G., L. S. Pereira, D. Raes, and M. Smith, 1998: Crop evapotranspiration: Guidelines for computing crop water requirements. FAO Irrigation and Drainage Paper 56, 300 pp., www.fao.org/docrep/X0490E/X0490E00.htm.
Brutsaert, W., and M. B. Parlange, 1998: Hydrologic cycle explains the evaporation paradox. Nature, 396, 30, https://doi.org/10.1038/23845.
Budyko, M. I., 1974: Climate and Life. Elsevier, 507 pp.
Chiew, F. H. S., N. N. Kamaladasa, H. M. Malano, and T. A. McMahon, 1995: Penman-Monteith, FAO-24 reference crop evapotranspiration and class-A pan data in Australia. Agric. Water Manage., 28, 9–21, https://doi.org/10.1016/0378-3774(95)01172-F.
Coyle, D., and A. Weller, 2020: “Explaining” machine learning reveals policy challenges. Science, 368, 1433–1434, https://doi.org/10.1126/science.aba9647.
Cunliffe, A. M., and Coauthors, 2022: Strong correspondence in evapotranspiration and carbon dioxide fluxes between different eddy covariance systems enables quantification of landscape heterogeneity in dryland fluxes. J. Geophys. Res. Biogeosci., 127, e2021JG006240, https://doi.org/10.1029/2021JG006240.
Delgado-Torres, C., and Coauthors, 2023: Multi-annual predictions of the frequency and intensity of daily temperature and precipitation extremes. Environ. Res. Lett., 18, 034031, https://doi.org/10.1088/1748-9326/acbbe1.
Elbaum, E., C. I. Garfinkel, O. Adam, E. Morin, D. Rostkier-Edelstein, and U. Dayan, 2022: Uncertainty in projected changes in precipitation minus evaporation: Dominant role of dynamic circulation changes and weak role for thermodynamic changes. Geophys. Res. Lett., 49, e2022GL097725, https://doi.org/10.1029/2022GL097725.
Feng, Y., Y. Jia, Q. Zhang, D. Gong, and N. Cui, 2018: National-scale assessment of pan evaporation models across different climatic zones of China. J. Hydrol., 564, 314–328, https://doi.org/10.1016/j.jhydrol.2018.07.013.
Fu, B.-P., 1981: On the calculation of the evaporation from land surface. Chin. J. Atmos. Sci., 5, 23–31, https://doi.org/10.3878/j.issn.1006-9895.1981.01.03.
Ghaemi, A., M. Rezaie-Balf, J. Adamowski, O. Kisi, and J. Quilty, 2019: On the applicability of maximum overlap discrete wavelet transform integrated with MARS and M5 model tree for monthly pan evaporation prediction. Agric. For. Meteor., 278, 107647, https://doi.org/10.1016/j.agrformet.2019.107647.
Goyal, M. K., B. Bharti, J. Quilty, J. Adamowski, and A. Pandey, 2014: Modeling of daily pan evaporation in subtropical climates using ANN, LS-SVR, Fuzzy Logic, and ANFIS. Expert Syst. Appl., 41, 5267–5276, https://doi.org/10.1016/j.eswa.2014.02.047.
Grigorev, V. Y., M. A. Kharlamov, N. K. Semenova, A. A. Sazonov, and S. R. Chalov, 2023: Impact of precipitation and evaporation change on flood runoff over Lake Baikal catchment. Environ. Earth Sci., 82, 16, https://doi.org/10.1007/s12665-022-10679-0.
Grossiord, C., T. N. Buckley, L. A. Cernusak, K. A. Novick, B. Poulter, R. T. W. Siegwolf, J. S. Sperry, and N. G. McDowell, 2020: Plant responses to rising vapor pressure deficit. New Phytol., 226, 1550–1566, https://doi.org/10.1111/nph.16485.
Guan, Y. Q., B. Mohammadi, Q. B. Pham, S. Adarsh, K. S. Balkhair, K. U. Rahman, N. T. T. Linh, and D. Q. Tri, 2020: A novel approach for predicting daily pan evaporation in the coastal regions of Iran using support vector regression coupled with krill herd algorithm model. Theor. Appl. Climatol., 142, 349–367, https://doi.org/10.1007/s00704-020-03283-4.
Huntington, T. G., 2006: Evidence for intensification of the global water cycle: Review and synthesis. J. Hydrol., 319, 83–95, https://doi.org/10.1016/j.jhydrol.2005.07.003.
Keskin, M. E., Ö. Terzi, and D. Taylan, 2009: Estimating daily pan evaporation using adaptive neural-based fuzzy inference system. Theor. Appl. Climatol., 98, 79–87, https://doi.org/10.1007/s00704-008-0092-7.
Kim, S., J. Shiri, and O. Kisi, 2012: Pan evaporation modeling using neural computing approach for different climatic zones. Water Resour. Manage., 26, 3231–3249, https://doi.org/10.1007/s11269-012-0069-2.
Kingston, D. G., M. C. Todd, R. G. Taylor, J. R. Thompson, and N. W. Arnell, 2009: Uncertainty in the estimation of potential evapotranspiration under climate change. Geophys. Res. Lett., 36, L20403, https://doi.org/10.1029/2009GL040267.
Kisi, O., 2015: Pan evaporation modeling using least square support vector machine, multivariate adaptive regression splines and M5 model tree. J. Hydrol., 528, 312–320, https://doi.org/10.1016/j.jhydrol.2015.06.052.
Kisi, O., and S. Heddam, 2019: Evaporation modelling by heuristic regression approaches using only temperature data. Hydrol. Sci. J., 64, 653–672, https://doi.org/10.1080/02626667.2019.1599487.
Li, Z., Y. Chen, Y. Shen, Y. Liu, and S. Zhang, 2013: Analysis of changing pan evaporation in the arid region of Northwest China. Water Resour. Res., 49, 2205–2212, https://doi.org/10.1002/wrcr.20202.
Lim, W. H., M. L. Roderick, M. T. Hobbins, S. C. Wong, and G. D. Farquhar, 2013: The energy balance of a US Class A evaporation pan. Agric. For. Meteor., 182–183, 314–331, https://doi.org/10.1016/j.agrformet.2013.07.001.
Lim, W. H., M. L. Roderick, and G. D. Farquhar, 2016: A mathematical model of pan evaporation under steady state conditions. J. Hydrol., 540, 641–658, https://doi.org/10.1016/j.jhydrol.2016.06.048.
Limjirakan, S., and A. Limsakul, 2012: Trends in Thailand pan evaporation from 1970 to 2007. Atmos. Res., 108, 122–127, https://doi.org/10.1016/j.atmosres.2012.01.010.
Linacre, E. T., 1994: Estimating U.S. class a pan evaporation from few climate data. Water Int., 19, 5–14, https://doi.org/10.1080/02508069408686189.
Liu, W., and F. Sun, 2016: Assessing estimates of evaporative demand in climate models using observed pan evaporation over China. J. Geophys. Res. Atmos., 121, 8329–8349, https://doi.org/10.1002/2016JD025166.
Lu, X., Y. Ju, L. Wu, J. Fan, F. Zhang, and Z. Li, 2018: Daily pan evaporation modeling from local and cross-station data using three tree-based machine learning models. J. Hydrol., 566, 668–684, https://doi.org/10.1016/j.jhydrol.2018.09.055.
Lugato, E., G. Alberti, B. Gioli, J. O. Kaplan, A. Peressotti, and F. Miglietta, 2013: Long-term pan evaporation observations as a resource to understand the water cycle trend: Case studies from Australia. Hydrol. Sci. J., 58, 1287–1296, https://doi.org/10.1080/02626667.2013.813947.
Massmann, A., P. Gentine, and C. Lin, 2019: When does vapor pressure deficit drive or reduce evapotranspiration? J. Adv. Model. Earth Syst., 11, 3305–3320, https://doi.org/10.1029/2019MS001790.
Matsoukas, C., N. Benas, N. Hatzianastassiou, K. G. Pavlakis, M. Kanakidou, and I. Vardavas, 2011: Potential evaporation trends over land between 1983–2008: Driven by radiative fluxes or vapour-pressure deficit? Atmos. Chem. Phys., 11, 7601–7616, https://doi.org/10.5194/acp-11-7601-2011.
McMahon, T. A., M. C. Peel, L. Lowe, R. Srikanthan, and T. R. McVicar, 2013: Estimating actual, potential, reference crop and pan evaporation using standard meteorological data: A pragmatic synthesis. Hydrol. Earth Syst. Sci., 17, 1331–1363, https://doi.org/10.5194/hess-17-1331-2013.
Monteith, J. L., 1965: Evaporation and environment. Symp. Soc. Exp. Biol., 19, 205–234.
Penman, H. L., 1948: Natural evaporation from open water, bare soil and grass. Proc. Roy. Soc. London, 193A, 120–145, https://doi.org/10.1098/rspa.1948.0037.
Peterson, T. C., V. S. Golubev, and P. Y. Groisman, 1995: Evaporation losing its strength. Nature, 377, 687–688, https://doi.org/10.1038/377687b0.
Priestley, C. H. B., and R. J. Taylor, 1972: On the assessment of surface heat flux and evaporation using large scale parameters. Mon. Wea. Rev., 100, 81–92, https://doi.org/10.1175/1520-0493(1972)100<0081:OTAOSH>2.3.CO;2.
Qasem, S. N., S. Samadianfard, S. Kheshtgar, S. Jarhan, O. Kisi, S. Shamshirband, and K.-W. Chau, 2019: Modeling monthly pan evaporation using wavelet support vector regression and wavelet artificial neural networks in arid and humid climates. Eng. Appl. Comput. Fluid Mech., 13, 177–187, https://doi.org/10.1080/19942060.2018.1564702.
Roderick, M. L., and G. D. Farquhar, 2002: The cause of decreased pan evaporation over the past 50 years. Science, 298, 1410–1411, https://doi.org/10.1126/science.1075390-a.
Roderick, M. L., L. D. Rotstayn, G. D. Farquhar, and M. T. Hobbins, 2007: On the attribution of changing pan evaporation. Geophys. Res. Lett., 34, L17403, https://doi.org/10.1029/2007GL031166.
Roderick, M. L., M. T. Hobbins, and G. D. Farquhar, 2009: Pan evaporation trends and the terrestrial water balance. I. Principles and observations. Geogr. Compass, 3, 746–760, https://doi.org/10.1111/j.1749-8198.2008.00213.x.
Rotstayn, L. D., M. L. Roderick, and G. D. Farquhar, 2006: A simple pan-evaporation model for analysis of climate simulations: Evaluation over Australia. Geophys. Res. Lett., 33, L17715, https://doi.org/10.1029/2006GL027114.
Sabziparvar, A.-A., H. Tabari, A. Aeini, and M. Ghafouri, 2010: Evaluation of class a pan coefficient models for estimation of reference crop evapotranspiration in cold semi-arid and warm arid climates. Water Resour. Manage., 24, 909–920, https://doi.org/10.1007/s11269-009-9478-2.
Seifi, A., and F. Soroush, 2020: Pan evaporation estimation and derivation of explicit optimized equations by novel hybrid meta-heuristic ANN based methods in different climates of Iran. Comput. Electron. Agric., 173, 105418, https://doi.org/10.1016/j.compag.2020.105418.
Stephens, C. M., T. R. Mcvicar, F. M. Johnson, and L. A. Marshall, 2018: Revisiting pan evaporation trends in Australia a decade on. Geophys. Res. Lett., 45, 11 164–11 172, https://doi.org/10.1029/2018GL079332.
Sun, Z., Z. Ouyang, J. Zhao, S. Li, X. Zhang, and W. Ren, 2018: Recent rebound in observational large-pan evaporation driven by heat wave and droughts by the Lower Yellow River. J. Hydrol., 565, 237–247, https://doi.org/10.1016/j.jhydrol.2018.08.014.
Thom, A. S., J.-L. Thony, and M. Vauclin, 1981: On the proper employment of evaporation pans and atmometers in estimating potential transpiration. Quart. J. Roy. Meteor. Soc., 107, 711–736, https://doi.org/10.1002/qj.49710745316.
Thompson, J. R., A. J. Green, and D. G. Kingston, 2014: Potential evapotranspiration-related uncertainty in climate change impacts on river flow: An assessment for the Mekong River basin. J. Hydrol., 510, 259–279, https://doi.org/10.1016/j.jhydrol.2013.12.010.
Tochimoto, E., and W. Yanase, 2023: Structural and environmental characteristics of western Baiu frontal depressions. J. Climate, 36, 2349–2365, https://doi.org/10.1175/JCLI-D-22-0515.1.
Wang, H., F. Sun, and W. Liu, 2018a: Spatial and temporal patterns as well as major influencing factors of global and diffuse horizontal irradiance over China: 1960–2014. Sol. Energy, 159, 601–615, https://doi.org/10.1016/j.solener.2017.11.038.
Wang, H., F. Sun, T. Wang, and W. Liu, 2018b: Estimation of daily and monthly diffuse radiation from measurements of global solar radiation a case study across China. Renewable Energy, 126, 226–241, https://doi.org/10.1016/j.renene.2018.03.029.
Wang, K., and R. E. Dickinson, 2012: A review of global terrestrial evapotranspiration: Observation, modeling, climatology, and climatic variability. Rev. Geophys., 50, RG2005, https://doi.org/10.1029/2011RG000373.
Wang, K., X. Liu, Y. Li, X. Yang, P. Bai, C. Liu, and F. Chen, 2019: Deriving a long-term pan evaporation reanalysis dataset for two Chinese pan types. J. Hydrol., 579, 124162, https://doi.org/10.1016/j.jhydrol.2019.124162.
Wang, T., F. Sun, J. Xia, W. Liu, Y. Sang, and H. Wang, 2018c: An experimental detrending approach to attributing change of pan evaporation in comparison with the traditional partial differential method. J. Hydrol., 564, 501–508, https://doi.org/10.1016/j.jhydrol.2018.07.021.
Wilks, D. S., 2016: “The stippling shows statistically significant grid points”: How research results are routinely overstated and overinterpreted, and what to do about it. Bull. Amer. Meteor. Soc., 97, 2263–2273, https://doi.org/10.1175/BAMS-D-15-00267.1.
Xue, K., and Coauthors, 2023: Estimating ecosystem evaporation and transpiration using a soil moisture coupled two-source energy balance model across FLUXNET sites. Agric. For. Meteor., 337, 109513, https://doi.org/10.1016/j.agrformet.2023.109513.
Yang, D., F. Sun, Z. Liu, Z. Cong, and Z. Lei, 2006: Interpreting the complementary relationship in non-humid environments based on the Budyko and Penman hypotheses. Geophys. Res. Lett., 33, L18402, https://doi.org/10.1029/2006GL027657.
Yu, F., C. Wei, P. Deng, T. Peng, and X. Hu, 2021: Deep exploration of random forest model boosts the interpretability of machine learning studies of complicated immune responses and lung burden of nanoparticles. Sci. Adv., 7, eabf4130, https://doi.org/10.1126/sciadv.abf4130.
Yuan, W. P., and Coauthors, 2019: Increased atmospheric vapor pressure deficit reduces global vegetation growth. Sci. Adv., 5, eaax1396, https://doi.org/10.1126/sciadv.aax1396.