Potential evapotranspiration change and its attribution in the Qinling Mountains and surrounding area, China, during 1960–2012

Potential evapotranspiration (ET₀), which was also called reference crop evapotranspiration, is a key factor in the process of the surface water cycle and energy balance. Its temporal and spatial variation is not only an important aspect in the study of climate change, but also an important manifestation of water resources and the eco-environment in response to climate change (Hupet & Vanclooster 2001). In arid and semi-arid regions, water is a major constraint to agricultural production, so the estimation of evapotranspiration is of great importance for water resource planning and efficient utilization (Liu et al. 2008). Actual observed evaporation capacity is more difficult to determine, therefore potential evapotranspiration has been widely used in water resource research and production practices of agricultural irrigation.

The fifth assessment report of the Intergovernmental Panel on Climate Change noted that the global average land surface temperature increased by 0.85 (0.65–1.06) °C in the period from 1880 to 2012. During 1951–2012, the rising rate of the global average land surface temperature (0.12 (0.08–0.14) °C decade^–1) was almost double that of 1880 (IPCC 2013). In this century, China's trend has been consistent with the global warming trend. The average land surface temperature in China increased 1.38 °C, which is a warming rate of 0.23 °C decade^–1, during 1951–2009 (IPCC 2013). In the context of climate warming, atmospheric evaporation should exhibit an upward trend, but the actual observations found that pan evaporation and ET₀ in different parts of the world have shown a downward trend over the last few decades (Peterson et al. 1995; Moonen et al. 2002; Roderick et al. 2009a, 2009b; Shen et al. 2010; Yin et al. 2010). The phenomenon of the ‘evaporation paradox’ attracted the widespread attention of domestic and foreign scholars, who conducted research into the trend and possible causes of pan evaporation and ET₀. In China, the existing research had studied and analyzed spatial and temporal changes in the ET₀ and pan evaporation of these regions, such as the whole territory of China (Liu & Zhang 2011), the alpine region (Wang et al. 2009), arid and semi-arid areas (Li et al. 2012), the Northwest Territories (Cao et al. 2012), the Tibetan Plateau (Zhou et al. 2013), the North China Plain (Mo et al. 2005), the Loess Plateau (Li 2012), the Yellow River Basin (Ma et al. 2012a, 2012b), the Yangtze River Basin (Pei et al. 2010), Hanjiang River (Zhang et al. 2005), the Weihe River (Zuo et al. 2011), the Heihe River Basin (Ma et al. 2012a, 2012b) and other regions. Further studies of the main factors that caused evaporation changes and their contributions were conducted by using correlation analysis, partial correlation analysis, sensitivity analysis and other methods. However, we found that in most studies, the time series of pan evaporation in China ended in approximately 2001. This was because the evaporation observation program at weather stations stopped in approximately 2001 in some regions in China, especially in the northern region. Since then, reports on pan evaporation have been relatively rare (Shen et al. 2010).

The Qinling Mountains and the surrounding area are important eco-geographical boundaries, and mark the boundary of the warm temperate zone and northern subtropical zone. To the north side of the Qinling Mountains is the Weihe River (a tributary of the Yellow River) and to the south is the Han River (a tributary of the Yangtze River). The Han River is a source of water for the ‘South–North Water Diversion Project’. This area is extremely rich in flora and fauna, with many rare species, and has been important to ecology, geography, botany, soil and water conservation, and other related disciplines. However, previous studies have focused on aspects of biodiversity, protection of water resources and soil conservation; the research on climate change is relatively insufficient. The study of Jiang et al. (2012) and Zhou et al. (2011) showed that during 1960–2011 (2009), the heating rate of the north and south sides of the Qinling Mountains was overall 0.18 °C decade^–1. The heating rate in the North Qinling Mountains was faster than that in other sub-districts (up to 0.25 °C decade^–1) and was significantly higher than the national and global averages (Qin et al. 2005). Therefore, this sensitive area experienced amplified warming. Some studies were conducted on climate changes of Qinling and its surrounding areas, mostly in the Qinling Mountains in the Shaanxi region (the Guanzhong and southern Shaanxi area); these studies were focused on the Qinling Mountains or watershed units (the Guanzhong region of the Weihe River and the southern Shaanxi region of the Han River). Research mostly focused on temperature, precipitation, runoff, and other variables, and paid less attention to other meteorological and hydrological factors. Fewer studies of ET₀ have been conducted. The current representation of study regions is limited by a lack of coverage, single factor selections and broad scales. The previous studies were insufficient to determine the spatial distribution patterns of meteorological elements and their spatio-temporal variations in NSQ. Scientific management decisions of authorities were limited because they could not fully grasp the climate changes in this area. Many studies (Jiang et al. 2013a, 2013b, 2013c, 2013d) showed that the climate system in NSQ has undergone significant changes since the 1990s. In this area, a clear warming and humidifying phenomenon occurred; additionally, radiation and wind speed (WS) decreased. All of these factors modified the regional energy and water cycle, of which the evapotranspiration process is an important component. At the same time, the land surface environment also dramatically changed. For example, some parts of the river near the Qinling Mountains nearly dried up, lakes shrunk, and the eco-environment in the basin changed. Based on these foundations, this paper applied the observational data of 47 meteorological stations in the northern and southern regions of the Qinling Mountains (NSQ) during 1960–2012 to estimate the potential evapotranspiration (ET₀) by using the FAO Penman–Monteith equation. ET₀ trends were analyzed at different times using the non-parametric Mann–Kendall trend test. Further, a quantitative study of the time-varying characteristics of ET₀ was conducted under the context of climate change; its response to meteorological factors was investigated by using correlation analysis, multiple regression analysis and other methods. This research focused on the recent dynamic changes of ET₀ to analyze the internal link of these various changes from the perspective of climate changes and the causes of the reversal of the evaporation trend.

Daily meteorological observations at 47 weather stations in NSQ were obtained from the China Meteorological Administration Shared Services Network for 1960–2012 (http://cdc.cma.gov.cn/home.do). According to the criterion of surface meteorological observation (China Meteorological Administration 2008), the meteorological observation field should be covered by grassland with a height lower than 20 cm and remaining stable, which basically satisfies a reference surface condition of the FAO Penman–Monteith equation. Meteorological data mainly include maximum temperature (MAT), minimum temperature (MIT), average temperature (AT) at a 2 m height, sunshine hours (SH), WS at a 10 m height, water vapor pressure (WVP) and 20 cm diameter pan evaporation. Diurnal temperature range (DTR) was calculated based upon MAT and MIT. To guarantee the accuracy of the results, the data were preprocessed before the analysis. The observational data of missing data years for more than 10 years (including 10 years) were excluded. The time series data of partially relocated stations were unified, and the remaining missing observation data were completed with a linear regression method and adjacent station interpolation to ensure the integrity of the time series. The division of the year is based on conventional meteorological seasons: December–February (DJF), March–May (MAM), June–August (JJA), and September–November (SON). The regional averages refer to the arithmetic mean value of the stations within a region. ArcGIS 10.0 software was used to process the mapping data, and the spline function was used to interpolate the spatial distribution map of ET₀.

Following the recommendation of the FAO, the Penman–Monteith equation is developed to better account for the specific local conditions. To obviate the need to define unique evaporation parameters for each crop and stage of growth, the concept of a reference surface was introduced. In the past, an open water surface has been proposed as a reference surface. However, the differences in aerodynamic, vegetation control and radiation characteristics present a strong challenge in relating ET to measurements of free water evaporation. Relating ET₀ to a specific crop has the advantage of incorporating the biological and physical processes involved in ET from cropped surfaces (Allen et al. 1998). Grass, together with alfalfa, is a well-studied crop regarding its aerodynamic and surface characteristics and is accepted worldwide as a reference surface (Allen et al. 1998). Therefore, the FAO expert consultation on the revision of FAO methodologies for crop water requirements accepted the following unambiguous definition for the reference surface: a hypothetical reference crop with an assumed crop height of 0.12 m, a fixed surface resistance of 70 s·m^–1 and an albedo of 0.23 (Allen et al. 1998). The reference surface closely resembles an extensive surface of green grass of uniform height, actively growing, completely shading the ground and with adequate water. The requirements that the grass surface should be extensive and uniform result from the assumption that all fluxes are one-dimensional upwards.

The FAO Penman–Monteith method is selected as the method by which the ET₀ can be unambiguously determined, and as the method which provides consistent ET₀ values in all regions and climates. It has been widely used in the calculation of ET₀, and many scholars verified its accuracy and practicality at national (Liu & Zhang 2011) and regional scales (Mo et al. 2005; Zhang et al. 2005; Wang et al. 2009; Pei et al. 2010; Zuo et al. 2011; Cao et al. 2012; Li 2012; Li et al. 2012; Ma et al. 2012a, 2012b; Zhou et al. 2013). However, in the Qinling Mountains area, the terrain of mountains and valleys is extremely complicated. The change in meteorological elements, such as temperature and precipitation, is relatively obvious and depends on altitude; thus, it is necessary to further verify the accuracy and applicability of the Penman–Monteith formula under the complex terrain and climate conditions. Comparing the observational data of the evaporation capacity of the 20 cm diameter evaporation pan with the calculation results of the Penman–Monteith formula month by month, the analysis results indicate that the correlation coefficients of the average of multiple stations year by year, the average for many years station by station, and the average station-by-station from year to year are as high as 0.89, 0.89 and 0.75, respectively, which are apparently higher than the analysis results of Liu & Zhang (2011) at the national scale using the same method. Therefore, the Penman–Monteith equation is suitable for estimating the ET₀ in NSQ.

where n is the length of the dataset, t is the extent of any given time, and ∑ denotes the summation over all ties.

The Mann–Kendall test may then be simply stated as follows: the null hypothesis, H₀, means not significant; H₀ is accepted if the test statistic Z is not statistically significant, i.e. |Z| > Z₁₋α_/2 in which ±Z_1–α/2 are the standard normal deviations. A negative value of Z indicated a downward trend, and vice versa.

To further study the relationship between ET0 and meteorological factors, their correlation coefficients were calculated, as shown in Table 2. In the three periods from 1960 to 2012, the correlation coefficients of ET₀ and SH in each sub-region increased with southward latitude. Thus, the changes in the energy conditions as shown via sunlight gradually affected the ET₀. Energy became the main factor in the process of evapotranspiration in the southern area of the Qinling Mountains. The correlation coefficients of P and WVP versus ET₀ gradually changed from negative to positive with southward latitude, and reflect the limit of the water supply decline. However, during 1980–1993 and 1994–2012, the contributions of SH and WS to ET₀ were above 20%, respectively (Table 3). Thus, sunshine and wind played an important role in the process of evapotranspiration at this stage. It is noted that the contribution rates of the various meteorological factors were not the same in the three periods; this finding was mainly caused by the change of the dominant meteorological factors.

The analysis in NSQ showed that the ET₀ trend over the study period had distinct turning points in 1979 and 1993, i.e. the pre-1979 trend was not consistent with the post-1979 trend. Overall, during 1960–2012, the temperature rose sharply while the ET₀ slightly decreased in the study area. Thus, an ‘evaporation paradox’ was apparent and could be explained by the evaporation complementary theory. The evaporation complementary theory assumes that under the given conditions of radiation, the actual evapotranspiration (ET_a) equals the potential evapotranspiration (ET₀) with a sufficient water supply. When the water supply of the underlying surface decreases, the ET_a would also decrease; thus, more energy is released to become sensible heat, and the ET₀ increases (Han et al. 2009; Yang et al. 2009). The ET_a was controlled by precipitation and sunshine, namely, water and energy. If the energy conditions were fixed, then the complementary relationship was pertinent. If the energy conditions changed slightly, then the complementary relationship was still in place. If the energy conditions changed greatly, then the complementary relationship might change. The energy supply of NRQ was adequate, and the ET_a was mainly controlled by the water conditions. Therefore, regardless of the increase or decrease in the ET₀, the ET_a decreased with declining precipitation.

Some deficiencies and uncertainties still exist in current research. The approach (the estimate based FAO Penman–Monteith method) does not exactly express the regional evapotranspiration, especially at high altitudes, so that more parameter calibration and formula improvement need to be carried out in the future. According to the long length of data series, inevitably the dataset has a spot of missing data. We discarded some stations that had large amounts of missing data, which will influence the precious description of ET₀ distribution. In addition, WS plays an important role in changing ET₀. Changes in atmospheric circulation, air humidity, irrigation, and atmospheric conditions have all resulted in the change of WS; it is difficult to quantitatively pinpoint the cause of the converse changing WS.

1. The spatial distribution of ET₀ was high in the northeast but low in the southwest, with the highest value in NRQ and the lowest value in BWV, which decreased from north to south. The seasonal distribution was as follows: summer > spring > autumn > winter.

2. The turning points of the ET₀ trend over the whole region and sub-regions occurred in 1979 and 1993, at which the value began to decrease or increase over the whole region and sub-regions. At the seasonal scale, in the period of 1960–1979, ET₀ in spring, summer, and winter presented a decreasing trend, however, it increased slightly in autumn; during 1980–1993, ET₀ in most seasons showed a downward trend except for autumn; in the period of 1994–2012, ET₀ declined in summer and autumn, however it increased slightly in spring and winter.

3. During 1960–1979, the positive correlation coefficient between DTR and ET₀ was highest, and DTR had the largest contribution to ET₀. During 1980–2012, the correlations of WS and SH versus ET₀ were the highest, followed by AT. The decrease in WS and solar radiation were the main causes of the ET₀ decrease, which offset the effect of the increase in temperature.

This research was funded by the National Natural Science Foundation of China (No. 41171420), the Key Research Program of the Chinese Academy of Sciences (No. KZZD–EW–04), and the West Light Foundation of the Chinese Academy of Sciences (2013–165–04).