Cryosphere evapotranspiration in the Tibetan Plateau:A review

KunXin Wang,YinSheng Zhang,Ning Ma,YanHong Guo,YaoHui Qiang

1. Key Laboratory of Tibetan Environment Changes and Land Surface Processes, Institute of Tibetan Plateau Research,Chinese Academy of Sciences,Beijing 100101,China

2.University of Chinese Academy of Sciences,Beijing 100101,China

3.CAS Center for Excellence in Tibetan Plateau Earth Sciences,Beijing 100101,China

4. Key Laboratory of Water Cycle and Related Land Surface Processes, Institute of Geographic Sciences and Natural Resources Research,Chinese Academy of Sciences,Beijing 100101,China

5.State Key Laboratory of Cryospheric Science,Chinese Academy of Sciences,Lanzhou,Gansu 730000,China

6.China-Pakistan Joint Research Center on Earth Sciences,CAS-HEC,Islamabad 45320,Pakistan

ABSTRACT Land surface actual evapotranspiration is an important process that influences the Earth's energy and water cycles and determines the water and heat transfer in the soil-vegetation-atmosphere system.Meanwhile,the cryosphere's hydrologi‐cal process is receiving extensive attention, and its water problem needs to be understood from multiple perspectives.As the main part of the Chinese cryosphere, the Tibetan Plateau faces significant climate and environmental change.There are active interaction and pronounced feedback between the environment and ETa in the cryosphere. This article mainly focuses on the research progress of ETa in the Tibetan Plateau. It first reviews the ETa process, characteristics,and impact factors of typical underlying surfaces in the Tibetan Plateau (alpine meadows, alpine steppes, alpine wet‐lands, alpine forests, lakes).Then it compares the temporal and spatial variations of ETa at different scales. In addition,considering the current greening of cryosphere vegetation due to climate change, it discusses the relationship between vegetation greening and transpiration to help clarify how vegetation activities are related to the regional water cycle and surface energy budget.

Keywords:cryosphere evapotranspiration;Tibetan Plateau;transpiration;evaporation

1 Introduction

Land surface actual evapotranspiration (ETa) is the most basic liquid-vapor phase change process on land surfaces and is also the only pathway for water to re‐turn to the atmosphere. Driven by energy and motion,the process that natural water moves continuously from water bodies, vegetation, and land surface to the atmosphere in the form of water vapor is called evapo‐transpiration or evaporation. In vegetation-covered ar‐eas, ETaincludes soil evaporation, vegetation transpi‐ration and evaporation from the canopy due to precipi‐tation interception.There is,however,only soil evapo‐ration in bare soil areas. Similarly, only water evapo‐ration occurs in water areas.ETarefers to those quanti‐fied from liquid to gaseous within a certain period.The latent heat flux refers to the energy consumed during the water phase change from the underlying surface to the atmosphere. In the Earth system, ETais a key element of the energy, moisture, and carbon cy‐cles (Junget al., 2010). In total, almost 65% of global land precipitation returns to the atmosphere as ETa(Oki and Kanae, 2006; Rodellet al., 2015). Simulta‐neously, the release of latent heat from the surface to the atmosphere constitutes approximately 50% of the solar radiation absorbed by the surface (Trenberthet al.,2009).

With the increasing impact of global climate change on the cryosphere, the hydrological process‐es of the cryosphere elements and their responses to climate change have received widespread attention.The various elements of the cryosphere usually work together at the watershed scale.Therefore,water prob‐lems related to the cryosphere cannot be understood from the traditional perspective of a single cryosphere element. Instead, they should be viewed from the viewpoint of Earth system science (Dinget al.,2020).Phase change and transmission of water in na‐ture are among the most critical processes in Earth system science, and it governs the circulation of cryosphere water. Traditionally, cryospheric hydrolo‐gy research is mainly concerned with water stored in the solid state and released in the liquid state. How‐ever, the vaporization of water molecules consumes seven times as much heat as liquefaction. Therefore,ETa(including sublimation) on different underlying surfaces of the cryosphere deserves more attention.As the central part of the Chinese cryosphere,the en‐vironment of the Tibetan Plateau is undergoing sig‐nificant changes under global warming, including ac‐celerated glacier melting, frozen soil thawing, alpine meadows degradation, and natural disaster aggrava‐tion (Yaoet al., 2017). The environment and ETain the Tibetan Plateau interact and react. ETacan regu‐late the balance of temperature and humidity in the Tibetan Plateau climate system in which the Asian monsoon and the westerly interaction together. On the one hand, the release of sensible and latent heat over the Tibetan Plateau has an essential impact on the East Asian climatic pattern, the Asian monsoon process, and the northern hemisphere atmospheric circulation (Yaoet al., 2013). On the other hand, the underlying surface change caused by climate warm‐ing significantly affects the ETarates. The enhanced ETaand precipitation change jointly affect the runoff in the local water cycle, thus leadig to changes in the regional water balance of the Tibetan Plateau (Yanget al.,2014).

This review first introduces the history of ETare‐searches and its measurements. Taking the Tibetan Plateau as the primary research area, it then reviews the ETaprocesses, variations and influencing factors over different underlying surfaces (alpine meadows,alpine grasslands, alpine wetlands, alpine forests, and lakes). Besides, the spatial and temporal distribution of ETaover the Tibetan Plateau is obtained based on the existing ETaresearches at the basin scale and the Tibetan Plateau scale. Finally, the relationship be‐tween vegetation greening and transpiration is dis‐cussed to further understand the relationship between vegetation activities and regional water cycle and sur‐face energy budget.

2 Development and methods

The processes of measurement and estimation of ETahave been developed for centuries, dating back to the 17th century. In London, England, scholars at Gresham College began to use evaporation pans to conduct evaporation observations (Halley, 1687).As early as 1703, the first comprehensive scientific ly‐simeter experiment was carried out in France by Philippe de la Hire (Puetzet al., 2018). The lysime‐ter method applies the principle of water balance to achieve accurate in-situ observations at point scales.At present, the lysimeter method is considered to be the closest observation technique for estimating ETa,and its application in the world is very mature(How‐ellet al., 1991). In 1926, Bowen calculated the ratio of sensible heat flux(H)and latent heat flux(LE)us‐ing the surface energy balance equation and the ver‐tical transport equation and developed the Bowen Ratio Energy Balance (BREB) method for assessing ETa(Bowen 1926; Lewis, 1995), which has been widely used in ecosystems with relatively homoge‐neous underlying surfaces. In 1939, Holzman and Thornthwatie proposed the aerodynamic method for evaluating ETa(Holzman, 1941). Penman (1948)treated soil evaporation and crops evapotranpiration as a certain proportion of the potential evapotranspi‐ration and proposed the Penman hypothesis.Through synthesizing the energy balance and water vapor transmission equations, he developed the Pen‐man equation for determining the ETaof water sur‐faces, bare land, and pasture. Later, Monteith (1965)introduced surface resistance and aerodynamic im‐pedance and derived the Penman-Monteith (P-M)equation, which clearly describes ETaunder unsatu‐rated conditions. In the 1950s, Swinbank proposed an eddy covariance (EC) (Swinbank, 1951) method that can be used to calculate ETaon underlying un‐saturated surfaces. Based on the EC method, the global scale flux observation network FLUXNET was fully established to measure the exchange of wa‐ter vapor, energy and carbon between the atmo‐sphere and ecosystems (Baldocchiet al., 2001). The theory of "complementarity correlation" proposed by Bouchet (1963) was the basis of a variety of ETamodels. Among them, the widely used ones include the Advection-Aridity model (Venturimet al.,2008), the Complementary Relationship Areal Evapotranspiration (CRAE) model (Morton, 1983),and the Granger model (Granger, 1989). Philip(1966) proposed a relatively complete Soil-Plant-At‐mosphere Continuum (SPAC) system theory. Built on this theory, the physical process models were es‐tablished and used in ETaestimation, including the Simultaneous Heat and Water (SHAW) model(Flerchinger, 2012) and the Distributed Hydrology Soil Vegetation (DHSVM) model (Wigmostaet al.,1994).

By the late 20th century, remote sensing informa‐tion was widely utilized globally to compute regional ETa. Since this method can gauge large-scale ETa, the prospective applications are very broad. In recent years, watershed water balance estimation, reanalysis data, remote sensing evapotranspiration products, ter‐restrial ecosystem models, and land surface model simulations have gradually become the main tools in regional ETaresearch (Olesonet al., 2013; Yinet al.,2013a; Liet al., 2014; Maet al., 2014; Maoet al.,2016; Sunet al., 2017; Maet al., 2019). They allow for accurate estimation of regional ETaand further our understanding of how climate and underlying terrain influence ETa, revealing the response mechanism of basin water cycles to future climate change (Junget al.,2010).

The observation and estimation of ETacan be roughly divided into observation and estimation meth‐ods. These methods can be applied to the research of ETaat different scales, including the point, watershed and regional scales.The classification tree that consist of main methods for ETaobservations and estimations is shown in Figure 1.

Figure 1 Main methods for observations and estimations of ETa

3 ETa on typical underlying surfaces of the Tibetan Plateau

The Tibetan Plateau belongs to a unique alpine cli‐mate zone,which can be subdivided into nine climatic sub-regions. The temperature zones include the pla‐teau sub-cold zone, the temperate plateau zone, and the subtropical plateau zone, while the arid/humid re‐gions cover the humid, the sub-humid, the semi-arid and the arid areas (Zhenget al., 2013). The distribu‐tion of vegetation types on the Tibetan Plateau has sig‐nificant horizontal differentiation. From southeast to northwest, forests, shrubs, alpine meadows, alpine steppes, alpine desert steppes and deserts are succes‐sively distributed (Zhaoet al., 2015). The underlying surfaces of the Tibetan Plateau are highly undulating and diverse in terms of types of terrain. The regional climate conditions are affected by westerly winds and monsoons to varying degrees,and ETaindicates signif‐icant surface heterogeneity. Due to the Tibetan Pla‐teau's harsh environment, the distribution of meteoro‐logical stations is inhomogeneous, with most located in the central and eastern regions.In order to compen‐sate for the shortage of observations, since the 1970s,the academic community has carried out several ob‐servation experiments of the land−atmosphere interac‐tions on the Tibetan Plateau, such as the Tibetan Pla‐teau atmosphere scientific experiments (Wanget al.,2016; Zhaoet al., 2018), GEWEX Asian Monsoon Experiment Tibet (GAME-Tibet) and CAME-Tibet(Maet al., 2006). In recent years, under the Third Pole Environment program (Yaoet al., 2012) and the impetus of the Second Tibetan Plateau Scientific Expedition and Research, several large experiments, in‐cluding the Tibetan Observation and Research Plat‐form (TORP) (Maet al., 2008) and the High-cold Re‐gion Observation and Research Network for Land Surface Processes & Environment of China (HORN),were successfully established.Coupled with the obser‐vation experiment of the Heihe River Basin (Liuet al., 2018) and the ChinaFLUX National Flux Obser‐vation Network (Yuet al., 2006b), scholars have gradually carried out observations of the ETaprocess on various underlying surfaces of the Tibetan Pla‐teau, including alpine meadows and grasslands, wet‐lands, forests, and lakes. Figure 2 illustrates more than forty ETaobservation sites compiled based on incomplete statistics, includes alpine meadow and steppe meadow stations, lake stations, alpine wet‐land stations, alpine forest and shrub stations, and al‐pine desert stations.

Figure 2 Distribution map of ETa observation sites on the Tibetan Plateau.The land use data was based on Xu(2019)

3.1 Alpine meadow and steppe

The Tibetan Plateau is characterized by higher so‐lar radiation than other regions at the same latitude.The precipitation mainly occurs during the growing season from May to September. The temperatures are always low throughout the year, with mean annual values of less than 6 °C (Songet al., 2014). In this unique ecological environment, ETacharacteristics in the alpine meadow and steppe ecosystem are distinc‐tive. Compared with other ecosystems, researchers have made more progress in terms of understanding its ETacharacteristics and variations by using the EC method, the BREB method, lysimeters, empirical models, and other methods, covering the northeastern,eastern, southeastern, and central regions of the Tibet‐an Plateau(Table 1).

ETastudies have been carried out in the upper Heihe River Basin in the northeastern Tibetan Pla‐teau over many years. The results revealed that the ratio of annual evapotranspiration to precipitation(referred to as ETa/P) for alpine meadows was only 0.6.During the growing season from May to Septem‐ber,the ETaaccounted for more than 80% of the total annual value, and the energy consumed by ETadur‐ing the period was about 44% of the net radiation (Liet al., 2013). Indeed, a recent study reported that the total ETain this region was controlled by energy, be‐cause of the high water availability (Sunet al., 2019).Interannual ETawas highly correlated with precipita‐tion, and the daily and seasonal changes were mainly dependent on energy changes.The low ratio of net ra‐diation to solar radiation and insufficient water supply resulted in low-level ETa(Guet al., 2008). Huet al.(2009) used the Shuttleworth−Wallace model to sim‐ulate soil evaporation and vegetation transpiration on alpine swamp meadows, alpine shrub meadows,and the alpine meadow steppe in the northeastern Ti‐betan Plateau. The results showed that soil evapora‐tion accounted for a large proportion of the ETa. The ratios of annual average soil evaporation to ETa(re‐ferred to as E/ETahereafter) for the three ecosystems were 0.6, 0.56 and 0.64, respectively. During the growing season, their ratios were 0.41, 0.42 and 0.58, respectively. Canopy stomatal conductance and leaf area index (LAI) were found to be the main fac‐tors governing the changes of E/ETaat different spa‐tiotemporal scales. Zhanget al. (2016) used the BREB method to study ETaseasonal change in three different ecosystems in the Qinghai Lake Basin.They found that the annual ETaof alpine meadows was 16% less than the annual precipitation,the annu‐al ETaof alpine shrub meadows was 26% higher than the precipitation, whereas the annual ETaof al‐pine steppes was close to the annual precipitation. In general, the daily ETaof the alpine ecosystem in the Qinghai Lake Basin was dominated by solar radia‐tion. Less precipitation and limited soil moisture al‐so controlled ETain the alpine steppe. Fanet al.(2010)compared the ETacharacteristics and the driv‐ing factors of alpine meadows under different cover‐age conditions in the Fenghuoshan Basin of the Yangtze River source area using a small lysimeter.With a decrease in coverage, the ETashowed a de‐clining trend, and the effect of the vegetation change on the ETaduring the growing season was particular‐ly significant.The average daily ETafrom January to April exhibited an increasing trend with diminishing vegetation coverage but was the opposite from May to December. The ETaof alpine meadows with vari‐ous levels of vegetation coverage reached its maxi‐mum value in July.In Naqu,located in the central Ti‐betan Plateau, the ETaof alpine meadows was 4 −6 mm/d during humid summers and close to 2 mm/d under dry conditions. Periodic water shortages likely affected the grass ETa(Conerset al., 2016). The sea‐sonal distribution of LAI, soil freezing and thawing,and precipitation significantly affected the seasonal variation of the surface energy exchange in the al‐pine grassland of the eastern Tibetan Plateau (Shanget al., 2015). In the Tanggula Mountain valley of the central Tibetan Plateau, the ETaof alpine meadows increased during the growing season from 2007 to 2013. The ETaon the precipitation days was mainly controlled by the net radiation and wind speed,while ETaon the non-precipitation days was closely related to the net radiation and air temperature(Wanget al., 2019). Seasonal variations of ETawere regulated by the freeze-thaw cycle, and the evapora‐tion during the thawing period was 4 to 10 times greater than that during the freezing period (Zhanget al.,2003).

Maet al. (2015a) analyzed the dynamic process of alpine steppe ETain the hinterland of the Tibetan Plateau. During the observation period, the ETaval‐ues in the ground frozen period and the transition pe‐riod were much lower than the reference evapotrans‐piration value, due to reduced soil water content.The ETain the rainy season accounted for more than 70% of the total for the whole year, with solar radia‐tion acting as the central control factor. ETaand pre‐cipitation were basically equal, suggesting that most atmospheric precipitation returned to the atmo‐sphere through ETaor sublimation. In addition, the potential evapotranspiration and ETain this alpine steppe had a symmetrical complementary relation‐ship, which was different from a previous finding(Maet al., 2015b). The latent heat flux between the semi-arid alpine steppe in the central Tibetan Pla‐teau and the humid alpine meadow in the southeast‐ern Tibetan Plateau was evidently different. The al‐pine grassland had sparse vegetation, and its evapo‐ration rate in the rainy season was appreciably relat‐ed to the normalized difference vegetation index(NDVI) and soil moisture, with an ETa/P of 1.36.The atmosphere was nearly saturated during the rainy season in the alpine meadows. The evapora‐tion rate was close to constant, had a low correlation with NDVI and soil moisture, and was mainly con‐trolled by net radiation, with an ETa/P of 0.43 (Wanget al.,2018).

3.2 Alpine forest and shrub

The alpine forests are mainly distributed on the slopes of the river valleys in the southeastern Tibet‐an Plateau. The trees distributed in the northwestern Sichuan, the southern Qinghai and the eastern Tibet‐an Plateau were gradually replaced by alpine mead‐ows in the western and northern Tibetan Plateau.Temperate grasslands and temperate arid deciduous shrubs are also distributed below 4,400 m in the val‐leys of the southern Tibetan Plateau (Zhaoet al.,2015). However, only a few studies on alpine forest ETain the Tibetan Plateau have been carried out to date, as summarized in Table 1. Notably, the latent heat flux of Jiuzhaigou coniferous and broadleaved mixed forest in the eastern Tibetan Plateau was consis‐tent with the daily and monthly variations in net radia‐tion.Particularly,the ratio of latent heat flux to net ra‐diation was as high as 0.69−0.75 in the 2014−2015 growing season, which may have been caused by high ETaduring forest restoration (Liet al., 2018).The ETa/P was close to 1, indicating that precipita‐tion in this area was mainly consumed by forest ETa(Yanet al., 2017). Several studies on the sub-alpine Qinghai spruce forests in the northeastern Tibetan Plateau revealed that rainfall events significantly af‐fected the energy flux distribution and ETa. The amount of sublimation in 2011 accounted for 7.1%of the total ETa; thus, it cannot be ignored in the wa‐ter balance of sub-alpine forests. During the grow‐ing season, the soil water content was low but not easily lost (Zhuet al., 2014). A long-term growing season flux observation of humid alpine shrubs in the northeastern Tibetan Plateau revealed that alpine shrubs served as a water source during the growth period, and seasonal changes in ETawere deter‐mined by net radiation and water vapor pressure loss. At different growth stages, there was a signifi‐cant difference in response to water vapor exchange(Liet al.,2019).

Table 1 Summary of multi-year average precipitation(P)and ETa data of typical underlays in the Tibetan Plateau

Table 1 Summary of multi-year average precipitation(P)and ETa data of typical underlays in the Tibetan Plateau

3.3 Alpine wetland

The alpine wetlands on the Tibetan Plateau are mainly distributed in the source areas of the Yangtze River and the Yellow River, the Ruoergai Plateau, the Hengduan Mountain valleys, and the southern and eastern margins of the Qiangtang Plateau. Wetlands are highly sensitive to climate change,and their degra‐dation is closely related to precipitation and ETa(Baiet al.,2004).Some scholars believe that the current al‐pine wetlands and alpine meadows vegetation share the same response to climate change, showing a greening trend (Wanget al., 2020). The wetland ETaresearch on the Tibetan Plateau is concentrated in the northeastern and eastern regions, with little occurring in other areas (Table 1). During the growing season,the wetland ecosystem energy in the Qinghai Lake source was dominated by latent heat flux and was mainly controlled by the canopy conductance, water vapor pressure, and available energy. Meanwhile, the energy in the non-growing period was governed by sensible heat. Wetland moisture was in surplus from July to December (Caoet al., 2019). In the degraded Yushu Longbaotan wetland located within the Yang‐tze River source region, the correlation coefficient be‐tween ETaand the annual temperature reaches 0.91. It is worth noting that, between 2012 and 2013, ETawas much greater than precipitation, which exacerbated the degradation of this alpine wetland. On a daily scale, the net radiation trend was consistent with the actual ETa, whereas the relative humidity had a strong negative correlation with ETa(Quanet al., 2016).While the daily average latent heat flux of the Zoige alpine wetland in the eastern Tibetan Plateau was mainly controlled by solar radiation and water vapor pressure,the latent heat flux during daytime was more sensitive to solar radiation (Chenet al., 2020). Based on multi-source remote sensing data, Wanget al.(2020) analyzed the ETacharacteristics of five wet‐lands, including the Mapam Yumco, Qiangtang Pla‐teau, Maidika, Gyaring-Ngoring Lake and Zoige.Their results revealed that precipitation was one of the most important factors in alpine wetlands. From 2000 to 2015,the Zoige wetlands ETashowed a signif‐icant increase, whereas the Qiangtang and Maidika wetlands showed a declining trend due to soil mois‐ture changes.For the complex response of alpine wet‐lands to climate change,a more comprehensive under‐standing of the ETacharacteristics and energy distribu‐tion at different time scales of the Tibetan Plateau will be conducive to protecting and restoration of alpine wetland ecosystems.

3.4 Lake evaporation

The Tibetan Plateau has the largest lake number and area in China. The lakes are mainly located in‐land, accounting for about 90% of the total lake area,and are mostly distributed at 4,000 −5,000 m above sea level (Maet al., 2011). In the past 30 years, the Qiangtang Plateau endorheic lakes have shown a sig‐nificant expansion trend (Leiet al., 2013). Lake evap‐oration is the only expenditure item of surface water in endorheic lakes, and its accurate quantification is essential to elucidate the mechanism of lake expan‐sion on the Tibetan Plateau. Analysis of the Qinghai Lake evaporation revealed a downward trend from 1961 to 2007,while precipitation increased(Shiet al.,2010). During the period 1958 −1998, the surface evaporation of the Zigtangcuo Lake fluctuated signifi‐cantly but exhibited an upward trend (Liet al., 2001).Haginoyaet al. (2009) simulated seasonal variations of the Namco Lake evaporation using the energy bal‐ance model and found that the maximum lake evapo‐ration occurred from October to November. However,the study lacked a verification of the observation data.Maet al.(2016) simulated the Namco Lake evapora‐tion variation from 1979 to 2012 using the CRLE model that fully considers heat storage changes. It was found that lake evaporation decreased only slight‐ly during this period. However, lake evaporation fell significantly at −12 mm/a from 1998 to 2008,contrib‐uting about 4%to the rapid lake area expansion.Xuet al. (2009) and Yuet al. (2011) used the energy bal‐ance model to simulate lake surface evaporation changes in the Yongzhuo Lake.They found that it de‐creased by about 7% from 1961 to 2005. Guoet al.(2019) simulated lake evaporation from 1961 to 2015 in the Selinco, using a single-layer lake evaporation model.They found that the lake evaporation showed a significantly rising trend from 1961 to 1984 and a slight increase from 2007 to 2015,whereas it fell con‐siderably from 1985 to 2006.The main governing fac‐tor in lake evaporation from 1961 to 2006 was identi‐fied to be wind speed. Table 1 lists some of the lake evaporation data.

4 ETa at a regional scale

4.1 Basin scale

The Tibetan Plateau, known as the "Asian Water Tower" creates many rivers. The water balance be‐tween watershed ETa, precipitation and runoff greatly affects the water resources available downstream. Es‐timates of basin-scale ETaare mainly based on the wa‐ter balance method.The actual ETain the upper Heihe River Basin is primarily controlled by precipitation and temperature. High temperature promotes vegeta‐tion transpiration in the upper Heihe River Basin and increases soil evaporation. Precipitation and relative humidity further enhance vegetation transpiration. Es‐pecially in the growing season, the contribution of vegetation transpiration to local precipitation accounts for nearly half, which plays an essential role in redis‐tributing regional water resources (Zhaoet al., 2019).In addition, diverse vegetation types increase their spatial heterogeneity. The highest ETaareas are domi‐nated by shrubs and alpine meadows, with an eleva‐tion of approximately 3,000 −3,600 m (Gaoet al.,2016).Yaoet al.(2016)used the revised semi−empiri‐cal Penman LE algorithm to clarify that the annual variation in latent heat flux in the Three Rivers source region during 1982−2010 was not significant. The la‐tent heat flux in the eastern watersheds declined due to the weakening of solar radiation, which offset the latent heat flux increase in the western watershed due to risen precipitation and soil moisture. More than 80% of precipitation could be consumed by ETain the source area of the Yangtze River(Yaoet al.,2014).Liet al. (2019) combined multi-source satellite remote sensing data and water balance calculation to provide a new approach for the estimation of watershed ETa.From 2003 to 2012, the multi-year ETaof the the source area of the Yellow River, the Yangtze River, the Salween River, and the Mekong River was 447 mm,366 mm, 430 mm, and 342 mm, respectively. The ETaof these source regions accounted for 57%−78% of the precipitation. From 1983 to 2006, the multi-year aver‐age ETain the upper Yellow River basin and the upper Yangtze River basin was consistent with the estimates of Liet al. (2019), which were 408 mm and 353 mm,respectively (Liet al., 2014). The multi-year average ETain the Qiangtang and Qaidam watersheds was lower than that in other watersheds, at 268 mm and 171.6 mm, respectively (Liet al., 2014). In the past thirty five years,the ETaof various basins revealed in‐creasing trends in each season and annual, although the change patterns of precipitation, runoff and tem‐perature in various watersheds were different (Liet al., 2014). Zhouet al. (2016) simulated the ETaof the land and water components of the Selinco Basin in the Qiangtang Plateau from 2003 to 2012, using the WEB-DHM distributed hydrological model and the P-M equation, respectively. The multi-year average land ETawas 379 mm, which showed an insignifi‐cant increase at 7.03 mm/a and was mainly con‐trolled by precipitation. The multi-annual average lake evaporation was 1,074 mm, decreasing signifi‐cantly at 4.17 mm/a. The lake evaporation variations were most sensitive to water vapor pressure changes and wind speed.

4.2 Plateau scale

Many studies have investigated ETaat the Tibetan Plateau scale. However, due to variations in the study period, research methods and geographical scope, the ETacharacteristics and spatiotemporal pattern on the Tibetan Plateau are not entirely consistent. Most re‐search results revealed that the Tibetan Plateau ETain‐creased in recent decades, including the simulations from 1981 to 2010 by Yinet al. (2013a) based on the LPJ model and ETacalculated using the CR model from 1982 to 2017 by Maet al. (2019). From 1982 to 2017,ETaon the Tibetan Plateau increased significant‐ly at the rate of 0.8 mm/a (Figure 3).Wanget al.(2018)revealed ETaand its components of the Tibetan Plateau at the 8−day scale during 1982−2012 based on PML model. In addition, Songet al. (2017)showed that ETaon the Tibetan Plateau followed a downward trend during 2000 −2010, with a decline rate of 6.6 mm/a.The results of Zhanet al. (2017) in‐dicated that ETaon the Tibetan Plateau remained rela‐tively stable from 2000 to 2014.

Figure 3 Variation of annual ETa during 1982−2017 in the Tibetan Plateau(data is from Ma et al.(2019))

Between 1981 and 2010, the P −ETavalue in‐creased in the northwestern Tibetan Plateau, especial‐ly in the alpine steppe of the Qiangtang Plateau. In contrast, it decreased in the southeastern Tibetan Pla‐teau, particularly in the eastern Himalayas' seasonal rainforest and the subtropical evergreen broad-leaved forest areas (Yinet al., 2013a). The P −ETavalue in the central Tibetan Plateau also increased significant‐ly from 2000 to 2010 (Songet al., 2017). The in‐crease in the P/ETaratio from 1982 to 2012 indicates that the Tibetan Plateau has become wetter over the past few decades (Wanget al., 2018). According to Table 2, the regional multi-year actual ETain the Ti‐betan Plateau is roughly 300−400 mm.

The seasonal variation of the actual ETaon the Ti‐betan Plateau is generally characterized by low value in spring and winter, and high value in summer and autumn. The cessation of photosynthesis in the eco‐system caused by low temperatures in winter leads to very small amounts of ETa. For multi-year monthly ETa, the value of February is the smallest throughout the year (Songet al., 2017). The ETaon the Tibetan Plateau has obvious spatial differentiation characteris‐tics, with high values in the east and south and low values in the west and north (Wanget al., 2018). For‐ested areas have the largest ETa, mainly distributed in the valleys of the southern and eastern Tibetan Pla‐teau, including downstream of the Yarlung Zangbo River and in the northern Sichuan Province.The max‐imum annual ETain these areas can reach 1,017.2 mm(Songet al., 2017). ETais also relatively high at river edges and in irrigation areas(Guoet al.,2011).Lower ETais found in areas with sparse vegetation and bare surfaces and is the smallest on the northern Tibetan Plateau and north of Bayankra Mountain (Songet al.,2017).

Table 2 Multi-year average ETa results at the Tibetan Plateau scale

ETawas found to be affected by both climate and human activities (Zhanet al., 2017). The in‐crease in soil moisture is the contributing factor in the rise of ETa. From 1981 to 2010, the decrease in wind speed weakened the evaporation capacity of water vapor in the atmosphere, which in theory would lead to a reduction in ETa. However, ETain most areas of the Tibetan Plateau demonstrated an increasing trend, and the variation in P −ETawas consistent with precipitation (Yinet al., 2013b).During the period 2000−2010,the correlation coeffi‐cient between ETaand relative humidity is the larg‐est, followed by temperature and LAI. The longterm ETavariation in the northwestern arid areas was mainly controlled by the relative humidity,while in the wetter areas, the LAI or temperature dominated the ETa(Songet al., 2017). Wanget al.(2018) analyzed the proportion of ETacomponents on the Tibetan Plateau and its influencing factors.Their results showed that Tibetan Plateau-scale multi-year average soil evaporation accounted for about 66% of the total ETa.Also, the contribution of precipitation, LAI, vapor pressure deficit and wind speed to ETaover the Tibetan Plateau accounted for 61%,46%,13% and 7%,respectively,while net radi‐ation contributed −20% to the ETatrend. It was found that LAI and precipitation affected the total ETa,veg‐etation transpiration and interception evaporation.Meanwhile, soil evaporation was primarily con‐trolled by water vapor pressure, air temperature, net radiation and precipitation. However, the effect of soil moisture on ETawas not considered.

5 Vegetation greening and transpiration

Terrestrial vegetation change is responding rapid‐ly to global environmental change. From 1982 to 2009, the temporal trend of LAI in global vegetation areas indicated that 25% to 50% of the vegetation ar‐eas increased during the growing season, while only 4% showed a decrease. Climate change is the main reason for vegetation greening on the Tibetan Plateau(Zhuet al.,2016).Zhonget al.(2019)analyzed the re‐lationship between vegetation greening and climate change on the Tibetan Plateau by reconstructing the vegetation index data set.The results revealed that the overall vegetation density increased from 1999 to 2014, and vegetation density in high-altitude cold ar‐eas was more sensitive to temperature.The vegetation density of grassland in semi-arid areas was found to be highly correlated with temperature and precipita‐tion, with correlation coefficients higher than those in semi-humid regions. During the growing season, the response of vegetation activity to temperature was positively correlated with precipitation. The positive response of alpine steppe vegetation activity to inter‐annual precipitation was enhanced, while the negative interannual relationship between alpine meadow and precipitation was weakened. The combined effect of air temperature and precipitation on alpine steppes'vegetation activity indicates that the warming and hu‐midification of the Tibetan Plateau may enhance the vegetation seasonal activity (Conget al., 2017). The vegetation activities in the Tibetan Plateau indicate negative feedback on temperature due to the increase in ETathroughout the growing season (Shenet al.,2015). ETais a crucial process to dissipate the energy absorbed by vegetation and determine the diurnal cycle of surface temperature. Under the intense solar radia‐tion on the Tibetan Plateau(Zhenget al.,2000),vegeta‐tion transpiration during the growing season is en‐hanced, thereby maintaining evaporation cooling,which is different from the situation in Arctic regions(Pearsonet al., 2013). At present, under continuous warming, the vegetation productivity on the Tibetan Plateau continues to increase. The region will continue to be cooled by ETa, which comes from the soil mois‐ture provided by ice and snow meltwater.However,the uncertainty of the current data and model makes it diffi‐cult to quantitatively describe the ETafeedback resulted from the increase in vegetation activity (Shenet al.,2015).The extent to which temperature rise and precip‐itation change regulate water and heat fluxes on the Ti‐betan Plateau is still unknown(Liuet al.,2018).

Transpiration links vegetation activity to the re‐gional water cycle and the surface energy budget. Its response to climate change may affect energy flux and further influence regional climate. Jasechkoet al.(2013) reported that "80% of ETaon the Tibetan Pla‐teau is due to vegetation transpiration".Coenders-Ger‐ritset al.(2014) proposed that the global transpiration to T/ETaratio ranges from 35% to 80%. To substanti‐ate these conclusions, scholars explored the propor‐tion of vegetation transpiration in ETaon the Tibetan Plateau based on models and other methods.Yinet al.(2010) used the SHAW model to simulate and sepa‐rate the ETaof shrub fir forests in the eastern Tibetan Plateau. Their results confirmed that vegetation tran‐spiration accounts for about 30% of the total ETa. Huet al. (2009) estimated the ETacomponents of three sites on the Tibetan Plateau based on the two-source model and reported that the T/ETaratio during the growth period was 42%−59%. Based on the SHAW model, this study examined the seasonal changes of vegetation transpiration and soil evaporation in the semi-arid alpine steppe of the central Tibetan Plateau(Zhou 2015). The results showed that vegetation tran‐spiration in the Qiangtang alpine steppe accounted for 23% of the total ETa. Vegetation transpiration in‐creased from June to mid-October, peaking at 47% in August, while vegetation first grew, then shrunk. Al‐though the proportion of transpiration increased again in early October, the total ETafell to a very low level under the influence of meteorological conditions (Fig‐ure 4). Therefore, the transpiration of the vegetation on the Tibetan Plateau makes a considerable contribu‐tion to ETa.The simulation results of Liuet al. (2018)showed that the proportion of transpiration in ETaris‐es under warming and wet conditions. This finding highlights the important role of vegetation activity in regulating surface water and energy fluxes, which may accelerate the negative feedback of the Tibetan Plateau to climate warming. In general, the ETasepa‐ration results currently measured on the Tibetan Pla‐teau are still scarce and have become the main"bottle‐neck"in exploring how vegetation changes affect ETa.

Figure 4 Ten-days soil evaporation,vegetation transpiration,and contribution to the total evapotranspiration(Sep.2013 to Oct.2014,Shuanghu Station)(data is from Zhou(2015))

In addition, extreme climate, deforestation and grazing impact the hydrological process to a variable degree on the Tibetan Plateau. The frequency of ex‐treme events on the Tibetan Plateau has increased over the past few decades (Youet al., 2008). Extreme drought and cold events hinder grassland growth, and the former is considered the most severe extreme event on the Tibetan Plateau.However,the increase in extreme humidity events in recent years may offset the adverse effects of drought. Extreme high-tempera‐ture events happening in May would also accelerate the growth of alpine grasslands (Liuet al., 2019). In future research on the hydrological cycle of the Tibet‐an Plateau,the study of extreme events'impact on hy‐drological processes should be strengthened for quan‐titative evaluation, especially regarding model param‐eterization improvement for extreme climate events.Deforestation in the southeastern Tibetan Plateau af‐fects the local climate and other adjacent regions,leading to a reduction in vegetation transpiration,cou‐pled with an increase in summer precipitation, which may elevate river runoff and exacerbate flooding di‐sasters in downstream areas(Cuiet al.,2007).

Pasture is one of the main land uses in the Tibetan Plateau ecosystem. It alters the vegetation community structure, especially in alpine meadow ecosystems,where the limiting factor of ETais shifted from soil moisture to water vapor pressure (Anet al., 2019).Specifically, grazing reduces the biomass on the land surface and thus decreases vegetation transpiration.Exposure of a large area of soil to strong solar radia‐tion enhances soil evaporation,resulting in a rise in to‐tal grassland ETa. Nevertheless, deep soil moisture is preserved. In arid years, reasonable grazing is benefi‐cial for storing soil moisture, thereby maintaining the high productivity of alpine meadows (Zhanget al.,2019). However, overgrazing will destroy the carbon sinks in the grasslands and lead to the degradation of the grasslands.The roots in bare soil have a rough tex‐ture and are extremely permeable. In addition, expo‐sure to strong solar radiation increases the water de‐mand for vegetation transpiration, accelerates the loss of precious water, and further erodes the soil (Heet al., 2017). Generally, sustainable and moderate graz‐ing improves plant photosynthetic potential by im‐proving soil nitrogen (Shenet al., 2019). Considering degraded pasture areas' impact on the local climate, it is imperative to restore the native grassland and pro‐mote sustainable development of the alpine meadow ecosystem.

6 Outlook and challenge

The Tibetan Plateau, where China's primary cryo‐sphere resides, is undergoing significant climate and environmental changes. The rise in air temperature and the decline in wind speed have caused significant changes in surface ETa. Thus, ETaloss is a factor that cannot be ignored in the study of the cryosphere. ETaresearch of alpine meadows, steppes, lakes, and other underlying surfaces has seen much progress. Still,most of the observations are presently concentrated in the northeastern, eastern and central regions of the Ti‐betan Plateau. Due to the uneven distribution of the study sites, the response mechanism of ETato climate change is still unclear. Establishing new ETaobserva‐tion stations in the western and southern regions of the Tibetan Plateau will provide additional observa‐tional verifications for regional studies.Nowadays,Ti‐betan Plateau-scale ETastudy is mainly based on re‐mote sensing inversion and model simulation. The simulation results are greatly influenced by atmo‐spheric forcing, surface data, and model parameters and structure, which comprise significant uncertainty.The authors believe that regional ETasimulation should focus on multi-source data fusion and parame‐terization scheme optimization to obtain high-confi‐dence analysis results.

Acknowledgments:

This research was jointly funded by the "Strategic Pri‐ority Research Program" of the Chinese Academy of Sciences (XDA2006020102), the Second Tibetan Pla‐teau Scientific Expedition and Research Program(2019QZKK0201), National Natural Science Founda‐tion of China (41801047, 41701082), the China Post‐doctoral Science Foundation funded project (2018M63 1589), and the Open Research Fund Program of State Key Laboratory of Cryospheric Science, Northwest In‐stitute of Eco-Environment and Resources, CAS(SKLCS-OP-2020-11).