Spatiotemporal Variations of Reference Evapotranspiration and Its Climatic Driving Factors in Guangdong, a Humid Subtropical Province of South China

Abstract

It is of great importance to study the changes in reference evapotranspiration (ET₀) and the factors that influence it to ensure sustainable and efficient water resource utilization. Daily ET₀ data calculated using the Penman–Monteith method from 37 meteorological stations located within Guangdong Province in the humid zone of southern China from 1960 to 2020 were analyzed. The trend analysis and Mann–Kendall test were used to analyze the time series changes in ET₀ and major climatic factors (air temperature (T), relative humidity (RH), sunshine duration (SD), and wind speed (u₂)) for over 61 years. Sensitivity and contribution analyses were used to evaluate the driving factors of ET₀. The main findings of the study are as follows: (1) the trend in average annual ET₀ time series in Guangdong slightly increased at a trend rate of 1.61 mm/10a over the past 61 years, with most stations experiencing an increase in ET₀. During the same period, air temperature significantly increased, while RH and SD decreased; u₂ also decreased. (2) Sensitivity analysis showed that ET₀ was more sensitive to RH and T than SD and u₂, with ET₀ being most sensitive to RH in spring and winter and T in summer and autumn. (3) The contribution analysis showed that T was the dominant factor for ET₀ variation in Guangdong, followed by SD. SD was found to be the dominant factor in ET₀ changes in areas where the “evaporation paradox” occurred, as well as in spring and summer. The study concludes that the climate in Guangdong became warmer and drier over the past 61 years, and if the current global warming trend continues, it will lead to higher evapotranspiration and drought occurrence in the future.

1. Introduction

Climate change characterized by global warming profoundly impacted agriculture, ecosystems, and human survival and development over the past few decades [1,2,3]. As a result, changes in the hydrological cycle and its other component processes can be expected worldwide, leading to a series of water resource-related problems [1,4].

Evapotranspiration (ET) is one of the most important components of the hydrological cycle and a key parameter in hydrological models and agricultural irrigation management [5,6]. It is a complex process that is not only controlled by climate variables but also influenced by underlying surface conditions, human activities, and other environmental conditions [7]. Therefore, estimating actual ET can be challenging [8]. As an alternative method, reference evapotranspiration (ET₀), also known as potential evapotranspiration in hydrometeorology, is used to assess the effects of climate change on evapotranspiration and the hydrological cycle. ET₀ is only affected by climatic factors and can reflect the evaporation capacity of the surface atmosphere under specific meteorological conditions in an integrated manner, while excluding the interference of other environmental conditions [9,10].

ET₀ is a basic parameter for regional water balance, irrigation scheduling, and water resource management [9,11]. Many past studies investigated the temporal and spatial patterns of ET₀, most of which focused on its response to the combined effects of climate change and human activities in different regions [12,13,14,15,16,17,18,19]. Additional studies showed an increasing trend in ET₀ in most parts of the world in recent decades, including in the United States [12], China [13,14], India [15,16], and Iran [17]. Some studies reported increasing trends in ET₀ (e.g., Ghafouri-Azar et al. [18] in the Korean Peninsula and Liu et al. [19] in the Tibetan Plateau). According to the definition of ET₀, the only factors that affect ET₀ are climatic variables [9]. Therefore, some studies focused on the relationship between ET₀ and influencing climatic variables [20,21,22,23,24]. On a global scale, a decrease in sunshine duration (SD) (or solar radiation) was the main cause of evaporation changes at the end of the 20th century [20]. However, due to the non-uniformity of the spatial and temporal distributions of climatic variables, the importance of climatic variables affecting ET₀ varies significantly from region to region. Vicente-Serrano et al. [21] analyzed changes in the annual ET₀ from 46 stations in Spain and reported increased values of ET₀ in the period 1961–2011. They also demonstrated that relative humidity (RH), wind speed (u₂), and maximum temperature (T_max) had stronger effects on ET₀ than SD and minimum temperature (T_min). Patle et al. [22] reported that the most sensitive parameter affecting ET₀ estimation in the eastern Himalayan region of Sikkim, India, was T_max, followed by SD, whereas u₂, T_min, and RH had a fluctuating effect on mean ET₀. Chu et al. [23] studied the effects of climate change on ET₀ in Jiangsu, eastern China, finding that u₂ contributed the most to ET₀, followed by SD. Liu et al. [19] reported that changes in ET₀ on the Qinghai–Tibet Plateau in the period 1961–2017 mainly depended on air temperature (T), followed by u₂ and SD, whereas RH had a negative effect. There is currently no consensus on the underlying causes of ET₀ variation. This issue exists because ET₀ is influenced by a combination of changes in climate variables, such as T, RH, u₂, and SD, and there is a complex non-linear relationship between ET₀ and these parameters, with significant variability existing between these meteorological factors [13,24].

Guangdong, which is located in the south of mainland China, was one of the first regions in China to undergo reform and opening up and has the most outward-looking economy. Due to rapid socio-economic development and human activities, coupled with its proximity to the South China Sea and significant maritime climatic features, both the ocean and the continent have a significant influence on the climate of the region, and changes in ET₀ in the region are likely to be complex. In addition, seasonal droughts and urban water shortages were commonly reported in the region in recent years [25,26]. However, there is a lack of study on the spatial and temporal variability and drivers of ET₀ in Guangdong.

In this study, the objectives were to analyze the change trends of annual and seasonal ET₀ and major climatic factors through collecting meteorological data and calculating ET₀ for Guangdong from 1960 to 2020, and to identify the major climatic driving factors of ET₀ through quantifying the effects of meteorological variables on ET₀. The results of this study can improve our understanding of the factors contributing to ET₀ changes and the impact of climate change on the hydrological cycle in Guangdong Province. It is anticipated that the outcomes of this study will improve guidance for agricultural production and economic development in this vitally important region.

2. Materials and Methods

2.1. Study Area and Meteorological Data

This study was conducted in Guangdong Province (20°09′ N~25°31′ N, 109°45′ E~117°20′ E), which is located in the humid south of China and has a tropical and subtropical monsoon climate; the province covers an area of approximately 179,700 km². The region has abundant sunshine, heat, and water resources, with an annual mean temperature of 22.3 °C; annual sunshine duration of 1745.8 h; and annual precipitation of 1789.3 mm, which varies between 1300 and 2500 mm. However, the region has unevenly distributed water resources, with frequent floods in summer and autumn, and droughts in winter and spring. In addition, the region experiences water shortages caused by population growth, climate change, and water pollution [27]. Understanding climate change and its impact on the hydrological cycle is crucial, as the area has strong evapotranspiration (more than half of the total precipitation) [28].

This study used daily meteorological data from 37 stations (Figure 1) located in Guangdong, including average temperature (T_mean, °C), maximum temperature (T_max, °C), minimum temperature (T_min, °C), relative humidity (RH, %), wind speed at 10m height (u₁₀, m/s), and sunshine duration (SD, h); these data were obtained from the China Meteorological Administration (CMA). The analysis period ranged from 1960 to 2020.

Table 1. Basic information of meteorological stations used in study area.

Station Name	Station Code	Latitude (°)	Longitude (°)	Elevation (m)	T (°C)	P (mm year−1)	ET0 (mm year−1)
Nanxiong	57996	25.08	114.25	149.7	20.60	1516.93	1086.92
Lianzhou	59072	24.82	112.37	131.7	20.59	1630.72	1017.50
Shaoguan	59082	24.67	113.60	121.3	21.24	1598.76	1102.62
Fogang	59087	23.88	113.52	97.2	21.87	2185.18	1110.86
Yingde	59088	24.18	113.42	74.4	21.86	1875.78	1125.04
Lianping	59096	24.38	114.50	283.9	21.06	1761.46	1045.66
Xinfeng	59097	24.03	114.22	269.3	21.29	1902.73	1021.50
Longchuan	59107	24.12	115.28	179.6	21.75	1676.19	1113.47
Dabu	59116	24.35	116.70	80.1	22.29	1504.85	1063.31
Meixian	59117	24.28	116.07	116.0	22.34	1497.09	1125.14
Guangning	59271	23.63	112.42	92.7	22.05	1721.33	1048.16
Gaoyao	59278	22.98	112.48	60.0	22.96	1641.14	1132.58
Qingyuan	59280	23.72	113.08	79.2	22.48	2125.33	1150.73
Guangzhou	59287	23.22	113.48	70.7	22.81	1809.36	1123.41
Dongguan	59289	22.97	113.73	56.0	23.14	1826.26	1195.20
Heyuan	59293	23.80	114.73	71.1	22.41	1922.63	1154.25
Zengcheng	59294	23.33	113.83	30.7	22.67	1965.49	1155.53
Huiyang	59298	23.07	114.37	108.5	22.87	1751.19	1191.59
Wuhua	59303	23.92	115.75	135.9	22.18	1499.91	1151.35
Zijin	59304	23.63	115.18	176.6	21.86	1706.80	1066.10
Jiexi	59306	23.45	115.85	80.9	22.42	2063.74	1123.96
Fengshun	59310	23.77	116.18	45.3	22.43	1830.54	1147.24
Shantou	59316	23.38	116.68	2.3	22.39	1556.30	1186.02
Huilai	59317	22.98	116.30	42.0	22.68	1792.25	1210.90
Nan’ao	59324	23.43	117.03	8.0	22.13	1357.39	1284.52
Xinyi	59456	22.35	110.93	141.4	23.47	1790.40	1201.36
Luoding	59462	22.72	111.60	60.0	23.12	1373.51	1114.67
Taishan	59478	22.25	112.78	33.1	22.89	1965.12	1183.24
Zhongshan	59485	22.50	113.40	33.7	22.97	1859.13	1134.09
Zhuhai	59488	22.28	113.57	51.4	23.17	2031.49	1271.63
Shenzhen	59493	22.55	114.00	63.0	23.36	1911.09	1252.23
Shanwei	59501	22.80	115.37	16.7	22.85	1899.26	1229.55
Zhanjiang	59658	21.15	110.30	53.4	23.84	1675.38	1214.12
Yangjiang	59663	21.85	111.98	90.3	23.11	2353.95	1189.67
Dianbai	59664	21.55	110.98	31.8	23.82	1550.61	1220.74
Shangchuan Island	59673	21.73	112.77	21.9	23.20	2244.92	1271.17
Xuwen	59754	20.25	110.17	11.4	24.54	1393.28	1271.18

provides basic characteristics of the meteorological stations, such as latitude, longitude, and altitude. The FAO Penman–Monteith equation was used to calculate the daily ET₀ for each station. Some years were excluded from the analysis due to the missing data from several stations, including Station Zhuhai (years 1960 to 1964), Station Fengshun (years 2016 to 2020), and Station Jiexi (years 1965 to 1969). Furthermore, routine quality checks and error correction were performed on the meteorological data according to the methodology of Peterson et al. [29]. The four seasons were divided into spring (March–May), summer (June–August), autumn (September–November), and winter (December–February of the following year).

2.2. Reference Evapotranspiration Computation

As measured ET₀ values were unavailable, we calculated daily ET₀ using the FAO56 Penman-Monteith (PM) method, which is the most widely used and accurate method for estimating ET₀ across various climatic regions. The equation is expressed as follows [9]:

$${\mathit{ET}}_0=\frac{0.408\Delta \left(R_n-G\right) + \gamma \frac{900}{T+273}{u}_2{(e}_s-e_a)}{\Delta +\gamma (1+0{.34u}_2)}$$

(1)

where R_n is the net radiation (MJ m⁻² d⁻¹), G is the soil heat flux (MJ m⁻² d⁻¹), T is the mean daily air temperature at 2 m height (°C), e_s is the saturation vapor pressure (kPa), e_a is the actual vapor pressure (kPa), Δ is the slope of the vapor pressure curve (kPa °C⁻¹), and γ is the psychrometric constant (kPa °C⁻¹).

To convert the wind speed data observed at the meteorological station, which were measured at 10 m above ground level, to the corresponding value at 2 m height, we used Equation (2) [9]:

$$u_2{=u}_z\frac{4.87}{\mathit{ln}(68.7z-5.42)}$$

(2)

where u_z is the wind speed at a height of z m above ground level (m s⁻¹).

2.3. Climatic Trend

Climate tendency refers to the changing trend of meteorological variables over time, which can be estimated using an ordinary linear regression equation, as given using the following equation [30,31]:

$$y\left(t\right)=\mathit{at}+b$$

(3)

where t represents the long-time series (year), y(t) is the ET₀ and other meteorological variables corresponding to t, a is the linear slope, and b is the intercept. In general, the climatic tendency rate (β) is equal to 10a with a unit of value per decade [30].

The significance of the trends in climatic series is evaluated using the Mann–Kendall trend test technique (MK test). The MK test is a rank-based non-parametric method that is widely applied for trend detection in hydro-climatic time series [32,33]. The MK test is described as follows:

$$S={\displaystyle \sum }_{i=1}^{n-1}{\displaystyle \sum }_{j=i+1}^n\mathrm{sgn}\left(x_j -{ x}_i\right)$$

(4)

where

$$\mathrm{sgn}\left(x_j{-x}_i\right)=\left\{\begin{array}{c}+1 \left(x_j{ > x}_i\right)\\ 0 \left(x_j=x_i\right)\\ -1 \left(x_j<x_i\right)\end{array}\right.$$

(5)

where x_i and x_j are the sequential data values, and n is the length of the data set. The mean and variance of the statistic S are given as:

$$E\left(S\right)=0$$

(6)

$$V\left(S\right)=\left[n\left(n-1\right)\left(2n+5\right)-{\displaystyle \sum }_{i=1}^n{t}_{i}i\left(i-1\right)\left(2i+5\right)\right]$$

(7)

where t is the extent of any given time.

The standardized statistic Z for a one-tailed test is formulated as follows:

$$Z=\left\{\begin{array}{l}\left(S-1\right)/\sqrt{\mathrm{var}\left(S\right)}\left(S>0\right)\\ 0 \left(S=0\right)\\ \left(S+1\right)/\sqrt{\mathrm{var}\left(S\right)}\left(S<0\right)\end{array}\right.$$

(8)

where a positive value of Z denotes an increasing trend, and a negative value indicates a decreasing trend. |Z|>1.96 and 2.32 indicate passing the significance level test of 0.05 and 0.01, respectively.

2.4. Assessing the Impact of Climate Variables on ET₀

In this study, the impact of meteorological factors on ET₀ was assessed through combining the sensitivity analysis with the contribution rate of a single climate factor to ET₀.

Sensitivity analysis is a widely used method to identify the changes in the dependent variable (ET₀) caused by the change in an independent meteorological variable [13,14,32], and the sensitivity coefficient is defined by [34]:

$$S_x=\underset{\Delta x\to 0}{\mathit{lim}}\frac{\Delta {\mathit{ET}}_0{/\mathit{ET}}_0}{\Delta x/x}=\frac{\partial {\mathit{ET}}_0}{\partial x}·\frac{x}{{\mathit{ET}}_0}$$

(9)

where Δx is the relative change in the model input value x, x is the meteorological factors, ΔET₀ is the relative change in ET₀ induced by Δx, and S_x is the dimensionless sensitivity coefficient. A positive (negative) S_x means that ET₀ increases (decreases) with the increase in meteorological factors. Larger |S_x| means higher sensitivity of ET₀ to meteorological factors. In order to quantitatively assess the sensitivity of ET₀ to different meteorological factors, the S_x was divided into four levels, as shown in

Table 2. Sensitivity coefficient level classification.

|Sx|	Sensitivity Level
|Sx| < 0.05	Negligible
0.05 ≤ |Sx| < 0.20	Moderate
0.20 ≤ |Sx| < 1.00	High
|Sx| ≥ 1.00	Very high

[35].

Sensitivity analysis cannot determine the actual contribution of each variable change to ET₀. In order to quantify the contributions of meteorological variables to the change trend in ET₀, we calculated the contribution rate (C_x) through multiplying the multi-year relative change rate of meteorological factors by its sensitivity coefficient, as shown in the following equation [36,37]:

$$C_x{=S}_x·R_c$$

(10)

$$R_c=\frac{n·a}{ \overline{x}}$$

(11)

where R_c is the relative change rate of certain meteorological factors and ET₀ (%); a is the linear slope, as mentioned in Equation (3); and $\overline{x}$ is the mean of the meteorological factor time series.

In this study, ArcGIS 10.8 was used to map the distribution of meteorological stations and spatial variation in ET₀ in the study area, Matlab 2018a was used for MK testing, IBM SPSS Statistics 26 was used for significance analysis, and Origin 2021 was used for other plots.

3. Results

3.1. Change Trends of Climatic Factors

The variations in climatic factors, which were averaged based on the 37 meteorological stations in Guangdong from 1960 to 2020, are shown in Figure 2 and

Table 3. Trend analysis results of climatic variable with linear regression, MK test.

Climatic Factors	Mean Value	Linear Regression		MK Test
		Slope	Std	Z	p-Value	Change Point (Year)
Spring
T (°C)	22.25	0.015	0.74	2.41	0.008	2000, 2011
RH (%)	82.20	−0.038	2.51	−2.38	0.046	1963, 1994
SD (h)	3.50	−0.008	0.74	−0.94	0.132	1968
u2 (m s−1)	1.53	−0.001	0.09	−1.60	0.083	1964
Summer
T (°C)	28.65	0.016	0.30	5.63	0.000	1993
RH (%)	82.03	−0.037	1.61	−3.30	0.003	1983
SD (h)	6.28	−0.010	0.60	−2.30	0.027	1979
u2 (m s−1)	1.45	0.002	0.08	3.18	0.004	2010
Autumn
T (°C)	23.99	0.024	0.50	5.30	0.000	1995
RH (%)	75.27	−0.057	3.04	−2.47	0.014	1966
SD (h)	6.01	-0.009	0.69	−1.86	0.086	2011
u2 (m s−1)	1.58	−0.0006	0.12	−0.29	0.458	-
Winter
T (°C)	15.01	0.026	0.94	3.58	0.000	1991, 2011
RH (%)	74.42	−0.036	3.31	−1.60	0.142	1997
SD (h)	4.17	−0.012	0.74	−1.30	0.032	-
u2 (m s−1)	1.69	−0.001	0.13	−0.22	0.276	-
Annual
T (°C)	22.47	0.019	0.35	5.41	0.000	1999
RH (%)	78.48	−0.042	1.73	−3.33	0.002	1987
SD (h)	4.99	−0.010	0.37	−2.73	0.001	1973
u2 (m s−1)	1.56	−0.0002	0.08	−0.38	0.702	-

Note: T is temperature, RH is relative humidity, SD is sunshine duration, u₂ is wind speed, slope is trend based on linear regression, Std is standard deviation of linear regression, and Z is Mann–Kendall test statistic.

. In line with the global warming trend, the T in the region exhibited a significant increase (p < 0.01) with a climate tendency rate of 0.19 °C/10a. The average annual RH was 78.51%, which declined significantly at a rate of −0.42%/10a, indicating a trend of drought in the atmosphere, with the most significant reduction occurring around 2010. Sunshine, which directly reflects solar radiation, is the energy source driving changes in other factors, such as T, RH, u₂, and ET₀. SD in Guangdong ranged from 4.26 to 6.34 h, with a multi-year average of 4.99 h, and showed a significant decreasing trend (p < 0.01) with a climate trend rate of −0.10 h/10a (Figure 2c). The decline in SD could be related to human activities and urbanization, which cause air pollution and an increase in aerosols in the air. In contrast, the multi-year average value of u₂ was 1.56 m/s, with a variation range of 1.39 m/s to 1.74 m/s, and showed a non-significant decreasing trend in spring, autumn, and winter (Figure 2d).

On the seasonal scale, T displayed a significant declining trend (p < 0.01) in all four seasons, and the increasing trend was stronger in cooler seasons than in warmer seasons. Autumn and winter warming rates were 0.24 °C/10a and 0.26 °C/10a, respectively, which were higher than the respective spring and summer warming rates of 0.15 °C/10a and 0.16 °C/10a. RH showed a decreasing trend in all seasons, with significant decreases in summer (p < 0.01), spring, and autumn (p < 0.05). Similarly, SD showed a decreasing trend in all seasons, with significant decreases in summer and winter (p < 0.05). u₂ showed a non-significant decreasing trend in spring, autumn, and winter and a significant increasing trend in summer (p < 0.01), with a tendency rate of 0.02 m/(s·10a).

3.2. Spatial and Temporal Variation Characteristics of ET₀

Due to the non-uniformity of the distribution of climatic factors, the spatial and temporal distribution of ET₀ in the study area is uneven. Figure 3a shows the spatial distribution and trend of annual average ET₀ in the study area, which shows that ET₀ gradually increased from north to south in spatial distribution. The differences in ET₀ among meteorological stations were evident, with higher ET₀ values found in coastal areas, such as western Guangdong and the Pearl River Delta (PRD). The three stations with the highest ET₀ values were Nan’ao (1284.52 mm), Zhuhai (1271.63 mm), and Xuwen (1271.18 mm), while the lowest three stations were Lianzhou (1017.50 mm), Xinfeng (1021.50 mm), and Lianping (1045.66 mm).

Of the 37 meteorological stations in the study area, 25 stations showed an increasing trend in ET₀, of which 9 stations showed a increasing trend with a selected significance level of 0.05. The stations with significant increases in ET₀ were found in Qingyuan, Gaoyao, Dongguan, Zhongshan, and Shangchuan Island in the PRD, as well as in Dabu, Jiexi, Shantou, and Longchuan in the east. On the other hand, 12 stations showed a decreasing trend in ET₀, with 7 stations showing a significant decrease, including Luoding and Dianbai in the west; Guangzhou, Zengcheng, Zhuhai, and Huiyang in the PRD; and Nan’ao in the east.

In terms of interannual variations, the overall annual ET₀ in the study area showed a slightly increasing trend (Figure 3b), with a climatic tendency rate of 1.61 mm/10a and an insignificant decreasing trend (

Table 4. Temporal trend analysis of seasonal ET₀ with linear regression and MK analysis.

Season	Mean Value	Linear Regression		MK Test
		Slope	Std	Z	p-Value	Change Point (Year)
Spring	3.00	−0.0239	22.29	0.24	0.885	2000
Summer	4.32	0.0013	18.83	0.03	0.993	2003
Autumn	3.29	0.1004	15.92	0.91	0.397	1990
Winter	2.01	0.0828	12.97	1.23	0.392	2002
Annual	3.15	0.1606	40.58	0.83	0.594	2003

Note: Slope is trend based on linear regression, Std is standard deviation, and Z is Mann–Kendall test statistic.

). The interannual ET₀ was unevenly distributed, with a variation range between 1069.27 and 1254.56 mm and a multi-year average of 1142.45 mm.

The spatial and temporal distribution of ET₀ in the study area varies across different seasons. Figure 4a–d show the spatial distribution characteristics and trends of multi-year average ET₀ at each station during the four seasons. During spring, the variation range of ET₀ is between 241.25 and 347.86 mm, and the highest ET₀ values are observed in Xuwen. Among the 37 stations in the study area, 23 stations show an increasing trend in ET₀, with 4 stations, such as Shantou and Dongguan, showing a significant increase; only 1 of the remaining 14 stations showing a significant decrease. During summer, ET₀ variation range is 355.14–431.06 mm, with high ET₀ values occurring in regions such as Zhanjiang and Zhuhai. Within the study area, 20 stations have an increasing trend in ET₀, while 17 stations have a decreasing trend. Among them, seven stations show a significant increase, while five stations show a significant decrease. The variation ranges of ET₀ in autumn and winter are 252.20 to 358.41 mm and 133.40 to 223.20 mm, respectively. The high ET₀ values are mainly concentrated in coastal areas, such as Zhuhai, Shenzhen, and Shanwei. In autumn, 24 stations show an increasing trend in ET₀, while 13 stations show a decreasing trend, with 8 stations showing a significant increase and 5 stations showing a significant decrease. During winter, ET₀ at 26 stations show an increasing trend, with 8 stations showing a significant increase; however, 11 stations show a decreasing trend, and 2 stations showing a significant decrease. The trend in ET₀ changes in different seasons also show that the study area has the most stations with increasing ET₀ in winter, while there are relatively more stations with decreasing ET₀ trends in summer.

The annual ET₀ tends to increase in all seasons except spring, albeit not significantly. The climatic tendency rate of ET₀ is highest in autumn and winter, being 1.00 mm/10a and 0.83 mm/10a, respectively, and lowest in spring and summer, being −0.24 mm/10a and 0.01 mm/10a, respectively. ET₀ is unevenly distributed throughout the year in the study area, with the highest ET₀ in summer (34.5% of the year), followed by autumn (299.27 mm), which accounts for 25.9% of the year, and spring (276.10 mm), which accounts for 23.9% of the year. However, in some years, spring ET₀ is higher than autumn. Winter ET₀ is the smallest (80.87 mm), accounting for only 15.7% of the year.

3.3. Sensitivity Analysis of ET₀ to Climatic Factors

The spatial distributions of the sensitivity coefficients (S_x) of annual ET₀ for each climatic factor were analyzed and visualized in Figure 5. The S_x of ET₀ to T ranged from 0.49 to 0.75, with an average of 0.65, as well as an increasing trend from north to south. Moreover, the S_x values were higher in the southern coastal areas, with most regions having values greater than 0.70. The S_x of ET₀ to RH ranged from −1.53 to −0.36, with an average value of −0.74. The spatial distribution of |S_x| showed an increasing trend from north to south, with |S_x| in the southern coastal areas having values greater than 1.00. However, the spatial distribution difference of the Sx of ET₀ to SD and u₂ was not significant, ranging from 0.23 to 0.30 and 0.06 to 0.13, with an average of 0.26 and 0.09, respectively. The sensitivity coefficients of ET₀ were positive for T, SD, and u₂ and negative for RH. Therefore, this result indicates that ET₀ in the study area increases with T, SD, and u₂ and decreases with RH. The analysis of the sensitivity of ET₀ to climatic factors showed that ET₀ was highly sensitive to T, RH, and SD and moderately sensitive to u₂.

On the seasonal scale (Figure 6), ET₀ showed positive S_x for T, SD, and u₂ and negative S_x for RH in all seasons. The S_x of ET₀ to T was highest in summer and autumn, with an average of 0.73. The same measure was smaller in spring and winter, with values of 0.60 and 0.54, respectively. The S_x of ET₀ to RH varied significantly among different regions in all seasons, and the spatial distribution of |S_x| increased from north to south. The S_x to RH was relatively high in winter and spring, with |S_x| averages of 0.94 and 0.83, respectively, and smaller in autumn and summer, with |S_x| averages of 0.63 and 0.55, respectively. The S_x of ET₀ to RH in different regions differed little between seasons, and the ranking of the S_x to SD was as follows: summer (0.35) > autumn (0.31) > spring (0.20) > winter (0.19). The S_x for u₂ were ranked as follows: winter (0.15) > autumn (0.13) > spring (0.05) > summer (0.04). Overall, ET₀ was highly sensitive to T and RH in different seasons, while being sensitive to SD in spring, summer, and autumn and moderately sensitive in winter. ET₀ was moderately sensitive to u₂ in winter, autumn, and spring, but negligibly sensitive in summer.

3.4. Contributions of Climatic Factors to the Trends in ET₀

We calculated the contribution rates of T, SD, RH, and u₂ to ET₀ using Equation (10), before adding them to obtain the total contribution rate of climatic factor changes to ET₀, which were noted as ET_0-estimated. Next, the actual relative rate of change in ET₀ was calculated using Equation (11), which was noted as a ET_0-actual. A correlation analysis between ET_0-estimated and ET_0-actual for all stations showed that ET_0-estimated was relatively close to ET_0-actual (Figure 7). The fitting points were concentrated around the 1:1 line, and the R² values were greater than 0.90 in different seasons and annually. Therefore, it could be considered reliable to quantify ET₀ changes based on the contributions of T, SD, RH, and u₂.

The spatial distribution of the contribution rate (C_x) of T, RH, SD, and u₂ to ET₀ variation in the study area is shown in Figure 8. The results indicate that the C_x of T to ET₀ variation is positive at all stations, with an average of 3.78% and a range in variation from 0.96% to 8.16% (

Table 5. Relative changes in climate variables for different seasons and their contributions to ET₀ change in study area.

Season	Relative Change Rc (%)					Sensitivity Coefficient Sx				Contribution Rate Cx (%)
	T	RH	SD	u2	ET0	T	RH	SD	u2	T	RH	SD	u2
Spring	4.17	−2.93	−14.68	1.77	−0.27	0.60	−0.83	0.20	0.05	2.51	2.28	−2.88	0.12
Summer	3.49	−2.78	−9.73	13.07	0.07	0.73	−0.55	0.35	0.04	2.55	1.40	−3.37	0.73
Autumn	6.11	−4.61	−9.26	3.95	2.15	0.73	−0.63	0.31	0.13	4.47	2.61	−2.86	0.22
Winter	10.40	−3.15	−17.58	0.92	3.22	0.54	−0.94	0.19	0.15	5.59	2.67	−3.39	−0.22
Annual	6.04	−3.37	−12.81	4.93	1.29	0.65	−0.73	0.26	0.09	3.78	2.24	−3.12	0.21

), with high values occurring in the PRD and eastern coastal areas, while C_x is relatively small in the north and west. The C_x of RH to ET₀ variation ranged from −2.55% to 10.22%, with 31 stations having positive C_x and 6 stations having negative C_x. Negative values were found in stations such as Zhanjiang, Shaoguan, and Nan’ao, where RH showed an increasing trend during the study period. High values of C_x were found in the PRD, and low values were mainly found in the north and west. The C_x of SD to ET₀ change ranged from −7.67% to 1.84%. |C_x| high values were mainly found in the PRD, Luoding, and Heyuan. C_x was negative in most regions and positive only in Huilai and Yingde, where C_x was 0.90% and 1.84%, respectively. SD showed an increasing trend in these two regions. The C_x of u₂ to ET₀ variation at different stations ranged from −5.70% to 10.91%, with 21 stations having positive C_x and 16 stations having negative C_x. Negative values were mainly found in coastal areas where u₂ decreased; however, in the remaining 21 stations, u₂ showed an increasing trend, resulting in an overall positive C_x on average.

Overall, the ranking of the contribution of each meteorological factor to ET₀ in the study area was T (3.78%) > SD (3.27%) > u₂ (2.73%) > RH (2.58%). The C_x of T, RH, and u₂ to ET₀ was positive on average, indicating that the temperature changes, RH, and u₂ in Guangdong over the last 61 years caused an increase in ET₀. In contrast, the C_x of SD to ET₀ was negative on average, indicating that the changes in SD in Guangdong decreased ET₀ during the study period.

On the seasonal scale (Figure 9), the C_x of T to ET₀ change was positive in all seasons, with high C_x mainly found in the PRD and the eastern coastal region. The mean C_x gradually increased from spring (2.51%) to winter (5.59%) (Table 5), which is consistent with the ranking of T tendency rate in different seasons. The C_x of RH to ET₀ variation was the smallest in summer (1.40%) and the highest in winter (2.67%), mainly due to the negative tendency rate and negative S_x of RH in different seasons. High C_x was observed in the PRD in all seasons. The average C_x of RH to ET₀ changes in all seasons was negative, with the highest |S_x| observed in winter (3.39%), followed by summer (3.37%), while the lowest was observed in autumn (2.86%). High |S_x| values were mainly found in the PRD, Luoding, and Heyuan. The high value area of |S_x| in winter was relatively lower than in other seasons. In conclusion, the contribution rate of T and SD to ET₀ change was higher in different seasons, followed by RH, while u₂ was very small. The C_x of SD was higher than T in spring and summer and was the dominant factor influencing ET₀ variation. However, in autumn and winter, the C_x of T was higher than SD, and T became the dominant factor of ET₀ change in the study area.

4. Discussion

4.1. Changing Trends of Meteorological Factors and ET₀

Over the past few decades, the air temperatures in most regions showed an unprecedented increasing trend, making global warming is indisputable. The overall T in Guangdong increased at a rate of 0.17 °C/10a during the study period, and the main reason explaining the regional warming is the increase in greenhouse gas emissions due to global population growth and economic development. The decreasing trend of RH in the study area and the increase in T indicate that the climate became drier to some extent over the 61 years studied. SD in Guangdong is on a downward trend; in fact, most of the world, such as Asia and Europe, is experiencing a decrease in sunshine hours to varying degrees, i.e., global dimming [38,39,40]. It was previously reported that human activities can cause a reduction in SD because atmospheric pollution from human activities leads to an increase in aerosols in the air. The aerosols increase the reflection and absorption of sunlight by the atmosphere, which, in turn, reduces the solar radiation reaching the ground, causing the reduction in SD [41]. Moreover, the decrease in wind speed makes the pollutants in the atmosphere less diffusible, which increases the near-surface aerosol concentration and contributes to the decrease in SD to some extent. u₂ in the study area decreases along with the trend in global wind speeds [42]. In this study, the distribution of u₂ changes was irregular, with stronger declines in Nan’ao, Shanwei, Shantou, and Shenzhen; these declines were concentrated in the central–eastern coastal region. Changes in meteorological factors also showed different seasonal characteristics variation. For example, T increased in all seasons, and RH and SD decreased in different seasons. Meanwhile, u₂ increased only in summer and decreased in other seasons.

Annual ET₀ in Guangdong is slightly increasing at a rate of 2.76 mm/10a; it appears that many regions around the globe significantly influenced by oceanic climate recorded an increasing trend in ET₀, such as the Korean Peninsula, which is in a subtropical monsoon climate zone [43]; Paraíba, Brazil, and Madagascar, which are in tropical high-temperature climate zones [44]; and Austria, which is in a maritime temperate broad-leaved forest climate zone [45]. This trend is different from those recorded in other regions in China, such as the Yellow River Basin [46], North China Plain [47], Northwest China [36], and Beijing–Tianjin–Hebei regions [48], where ET₀ decreased since the 1960s, creating an “evaporation paradox”. This paradox exists at about 62% of the stations across China, where ET₀ decreases at a rate of 5.2 mm/10a despite an increase in temperature [14]. However, this study showed some subtle differences in the spatial distribution in Guangdong: due to the influence of topography and geomorphology, ET₀ is higher in the southern low-elevation coastal areas than in the northern high-elevation areas. The evaporation paradox phenomenon was found at 12 of 37 stations in the study area, where ET₀ decreased with increasing temperature. Among these stations, Luoding and Dianbai in western Guangdong, Guangzhou, Zhuhai, Zengcheng, and Huiyang in the PRD, as well as 6 stations in Nan’ao in eastern Guangdong, showed significant decreasing trends in ET₀. The study also revealed seasonal variations in ET₀, with a decreasing trend in spring and an increasing trend in the other seasons, particularly autumn and winter. These results show that various factors affect ET₀, with each factor having a different weight.

4.2. Climatic Factors Affecting the Variation in ET₀

Climate change is the key factor driving ET₀ variation; however, there are differences and uncertainties in the factors influencing ET₀ variation between global regions, mainly due to the interactions between meteorological elements. It is generally accepted that ET₀ is positively correlated with u₂, T, and SD and negatively correlated with RH [49]. The sensitivity analysis conducted in this study shows that ET₀ in Guangdong is highly sensitive to RH, T, and SD and moderately sensitive to u₂, which is consistent with the results recorded for the Poyang Lake catchment, China [32], and for the Korean Peninsula [43]. However, the sensitivity of ET₀ to climatic factors varies across different regions, and ET₀ was most sensitive to RH in Guizhou Province [50], Jiangsu Province [23], the Beijing–Tianjin–Hebei region [48], and the Huai River Basin [51]. In contrast, ET₀ was most sensitive to u₂ in the northwest inland region, followed by RH, T, and SD [36]. Although ET₀ was more sensitive to RH than to T, the contribution analysis showed that RH contributed less to the increase in ET₀ than T and SD. In the tropical high-temperature climate zone in Brazil, the main climatic factor driving ET₀ changes is temperature, while the most critical impact factor alternates between temperature and sunshine hours in the rainy and dry seasons. In Austria, which is located in a temperate broad-leaved forest climate zone, the main reason for ET₀ rise is the increase in solar radiation [45]. In this study, T was the main cause of ET₀ changes in Guangdong, and SD was the second main cause of ET₀ changes, while in spring and summer, SD was the dominant factor of ET₀ changes because SD decreased more significantly. It was reported that the drier the climate in China, the greater the contribution of wind speed to ET₀, especially in the arid northwest, where u₂ is the main cause of ET₀ decrease [26,36]. The results of this study showed a relatively large contribution of u₂ to ET₀ variation in summer, and the same low RH and strong u₂ rise were found at these stations.

Overall, the increasing effect of rising T and falling RH on ET₀ in Guangdong during the study period exceeded the decreasing effect of falling SD on ET₀, ultimately leading to an overall increasing trend of ET₀. However, in regions where the evaporation paradox exists, i.e., T rise is accompanied by ET₀ decline, the contribution of SD to ET₀ is usually more significant than that of T. Moreover, in inland areas of Guangdong, such as Guangzhou, Zengcheng, Huiyang, and Luoding, the strong decreasing effect of SD on ET₀ masks the increasing effect of T and RH, while in coastal regions, such as Dianbai and Nan’ao, the decrease in SD and rise in RH offset the increase in ET₀, suggesting that there are spatial differences and uncertainties in the weights of each factor affecting the variation in ET₀.

4.3. Impact of Climate and ET₀ Changes on Agricultural Production

Guangdong belongs to the tropical and subtropical monsoon climate zone, with a humid climate, abundant heat, and abundant but unevenly distributed precipitation. This region is a significant producer of grain crops (e.g., rice, corn, and tubers) and tropical crops (e.g., sugarcane, rubber trees) in China [52,53]. Climate change may complexly impact agricultural production and water resource management in this region. In fact, an increasing drought trend was consistently observed in recent years, which was both global and regional in scale, including in the Guangdong [54,55]. Climate warming will result in a richer agro-climatic heat resource, a longer crop growing season, and more heat in the growing season in Guangdong. This change, in turn, will push the existing agro-climatic zone and crop maturity boundaries northward and to higher elevations, which favors the cultivation of tropical crops in the region, while the northern boundary of the second and third maturity zones of crops also moves northward and the area is expanded. However, the increase in temperature may also lead to drought and summer heat disasters, which could reduce agricultural yield or affect the quality of crops. The decrease in sunshine duration will negatively impact the high and stable yield factors of tropical fruits in the region, such as fruit enlargement and sugar accumulation in sugarcane, as well as increase the late rice seed setting rate [53]. In addition, ET₀ increased in the province, particularly in eastern Guangdong and the Leizhou Peninsula. Crops’ evapotranspiration water consumption in these agricultural areas increased, leading to an increase in demand for irrigation water. The contradiction between the supply of and demand for water resources in the future is prominent, which may aggravate the drought and water shortage situation in water-restricted areas, especially in spring and winter, when precipitation is at the lowest annual level. Corrective and preventive measures should be considered in water resources planning and agricultural production.

For a complete understanding of the mechanism driving changes in regional evapotranspiration responses to climate change, further attention should be paid to the feedback and quantitative relationship between actual evapotranspiration and ET₀, as well as how this relationship affects regional hydrological cycles.

5. Conclusions

In this study, we conducted a comprehensive analysis of the trends of ET₀ and major climatic factors (T, RH, SD, and u₂) in Guangdong from 1960 to 2020. Our findings provide important insights into the factors driving variations in ET₀ and implications for future regional water management. Based on the results, the following conclusions can be made:

(1)

The annual ET₀ in Guangdong increased at a rate of 1.61 mm/10a. ET₀ increased at most stations; only 6 stations had a decreasing trend for all 37 analyzed samples. ET₀ decreased in spring and increased in the other seasons, though these trends were not statistically significant. Meanwhile, T significantly increased during the study period, while RH and SD significantly decreased.

(2)

Sensitivity analysis showed that ET₀ was more sensitive to RH and T than SD and u₂ in Guangdong. ET₀ was most sensitive to RH in spring and winter and T in summer and autumn. Through considering the variation in variables and their sensitivity to ET₀, the results showed that T was the dominant factor for ET₀ variation in Guangdong, followed by SD. In the areas where the “evaporation paradox” occurs, as well as in spring and summer, SD was the dominant factor in ET₀ variation. Therefore, the trend of climatic factors plays a critical role in analyzing the variation in ET₀.

(3)

The increasing effect of rising T and decreasing RH on ET₀ masked the decreasing effect of SD on ET₀, resulting in an overall increase in ET₀ in Guangdong. This result suggests that a potential future increase in SD combined with a decrease in RH may lead to higher evapotranspiration rates and drought events in Guangdong. Therefore, we recommend adopting long-term water management strategies for sustainable development to cope with regional climate change.