1. Introduction
A large amount of water in the form of glaciers, snowpack, lakes, and rivers is stored in the Tibetan Plateau (TP), which is the world’s highest and largest plateau. The TP is recognized as “Asia’s water tower” and supplies several major rivers and a large amount of the population on the Asian continent (Xu et al. 2008). According to the great rivers and watersheds, the TP is divided into 12 drainage basins, namely the Amu Dayra, Indus, Ganges, Brahmaputra, Salween, Mekong, Yangtze, Yellow, Hexi Corridor, Qaidam, Tarim, and Inner Plateau basins (Tang et al. 2019; Zhang et al. 2013), as shown in Fig. 1. The Inner Plateau is an endorheic basin. Apart from it, many major Asian rivers originate from other exorheic basins over the TP. Therefore, it is vital to understand the large-scale aspects driving the water budget over the TP and on its basin scales. Thus, the following questions are posed: Where is the water sourced from? Has the supply to water resources changes? These questions lead to cascading impacts in downstream areas where billions of people live.
The study area is divided into 12 drainage basins (from 1 to 12 the Amu Dayra, Indus, Ganges, Brahmaputra, Salween, Mekong, Yangze, Yellow, Hexi, Qaidam, Tarim, and Inner Plateau basins, respectively) over the TP. The discharge stations (Zhimenda and Tangnaihai) and the corresponding upper Yangtze and Yellow River basins (UYZR and UYLR) are also marked with shaded regions.
Citation: Journal of Climate 34, 15; 10.1175/JCLI-D-20-0692.1
The study area is divided into 12 drainage basins (from 1 to 12 the Amu Dayra, Indus, Ganges, Brahmaputra, Salween, Mekong, Yangze, Yellow, Hexi, Qaidam, Tarim, and Inner Plateau basins, respectively) over the TP. The discharge stations (Zhimenda and Tangnaihai) and the corresponding upper Yangtze and Yellow River basins (UYZR and UYLR) are also marked with shaded regions.
Citation: Journal of Climate 34, 15; 10.1175/JCLI-D-20-0692.1
The study area is divided into 12 drainage basins (from 1 to 12 the Amu Dayra, Indus, Ganges, Brahmaputra, Salween, Mekong, Yangze, Yellow, Hexi, Qaidam, Tarim, and Inner Plateau basins, respectively) over the TP. The discharge stations (Zhimenda and Tangnaihai) and the corresponding upper Yangtze and Yellow River basins (UYZR and UYLR) are also marked with shaded regions.
Citation: Journal of Climate 34, 15; 10.1175/JCLI-D-20-0692.1
For the water cycle within a specific area, all components described by Eqs. (1) and (2) represent area-averaged or area-integrated values. Developing regional estimates of P and E is challenging due to the lack of observations supporting their spatiotemporal variability. Such a lack of observations leads to reliance on atmospheric analysis estimates. Reanalysis data can provide an observationally constrained dataset representing the hydrological cycle, but previous studies have suggested that surface fluxes derived from reanalysis should be used with great caution (Trenberth et al. 2011; Bosilovich et al. 2017). Imperfect parameterizations of the model processes and analysis increments arising from model biases and observational uncertainties impact the simulation of components of the water budget during reanalysis. With the improvement of numerical modeling capabilities and the advent of new satellite observation techniques and associated data assimilation methodologies, improved estimates of surface fluxes in the latest reanalysis ERA5 from ECWMF are expected and in progress (Hersbach et al. 2020; Eicker et al. 2020). The improved global balance of precipitation and evaporation is one of the strengths of ERA5 compared to ERA-Interim.
The present study addresses the net water flux (P − E) and its variability at daily to decadal time scales over the TP and its drainage basins. The starting point is the atmospheric water balance Eq. (1), which provides a method to estimate the net water flux based on reanalysis data (Oki et al. 1995). The ERA5 P − E [the model estimate on the left-hand side of Eq. (1)] is compared with the residual moisture budget (the analysis estimate on the right-hand side). Although comparisons of both estimates have revealed systematic differences and errors in the moisture budget, former generations of reanalyses have provided useful information to represent global and continental water cycles (Trenberth et al. 2011; Brown and Kummerow 2014; Bosilovich et al. 2017).
Previous studies have investigated components of atmospheric water budgets over the TP with various reanalysis datasets (Feng and Zhou 2012; Gao et al. 2014; Wang et al. 2017; Zhao and Zhou 2019; Yan et al. 2020). Water vapor convergence and P − E generally maintain equilibrium, but the climatology and long-term trends of the water balance exhibit distinct spatial variability across the TP. This work evaluates the net water flux estimates and variability parameters derived from ERA5 over the TP, while characteristics at the drainage basin scale are highlighted on multiple time scales ranging from daily, and submonthly to long-term trends.
In addition, the terrestrial water cycle over the TP and its drainage basins is discussed in this work. The nature of the water cycle involves the terrestrial branch described by Eq. (2). Long-term measurements of river discharge R are limited in availability (Rodell et al. 2015), while several hydrological stations over the TP provide streamflow measurements dating back to the 1950s (Tang et al. 2019). Accurate estimates of water storage S, including surface water, soil moisture, groundwater, snow, and ice, are indirectly measured by GRACE (Gravity Recovery and Climate Experiment), which was designed to measure the mass distribution and mass flux of water through its effects on Earth’s gravity (Tapley et al. 2004). GRACE has contributed highly accurate estimates of monthly and seasonal terrestrial water storage changes (Swenson and Wahr 2006; Kunstmann et al. 2012; Springer et al. 2014; Lorenz et al. 2014). The observationally derived water budget has been improved through estimates of vertically integrated terrestrial water storage changes provided by GRACE. The inherent spatial and temporal resolutions of GRACE data have been developed in recent years. A new state-of-the-art GRACE time series dataset is ITSG-Grace2018, which includes unconstrained monthly and constrained daily solutions of total water storage anomalies (TWSAs), as well as a high-resolution static gravity field (Kvas et al. 2019). Therefore, hydrometeorological fluxes derived from ERA5, available river discharge observations, and ITSG-Grace2018 provide a new opportunity to address the characteristics of the atmospheric–terrestrial water cycle and water resource changes over the TP and its basins in this work.
In the following, section 2 details the datasets and methods used herein. Section 3 presents the ERA5-derived net water flux over the TP and its basins. The atmospheric water balance is evaluated. Sections 4 and 5 address the joint atmospheric–terrestrial water cycle over the TP and its basins. ERA5- and GRACE-derived water fluxes or storage changes are compared, with a particular focus on long- and short-term variations. Section 6 discusses the implications and limitations of the results.
2. Data and methodology
a. Study area
The study area is shown in Fig. 1. The TP is divided into 12 drainage basins (the Amu Dayra, Indus, Ganges, Brahmaputra, Salween, Mekong, Yangtze, Yellow, Hexi Corridor, Qaidam, Tarim, and Inner Plateau basins, marked 1–12) according to great rivers and watersheds provided by the HydroSHEDS dataset (Immerzeel and Bierkens 2012; Zhang et al. 2013). Daily streamflow records at the outlet hydrological stations Tangnaihai and Zhimenda (marked in Fig. 1) in the northern TP are employed in this work; thus, the corresponding drainage basins [the upper Yangtze River basin (UYZR) and upper Yellow River basin (UYLR)] are separately investigated as shaded areas in Fig. 1. The areas of the TP and its drainage basins are listed in Table 1.
Annual mean of the net water flux estimates over the TP, 12 drainage basins, and UYZR and UYLR basins, together with the percentage of each basin’s contribution to the total water flux over the TP. Boldface values highlight the larger term in comparisons of P − E and the residual in Eq. (1).
b. Water balance equations based on ERA5 and ITSG-Grace2018
The atmospheric–terrestrial water cycle is investigated according to Eqs. (1) and (2). Previous studies often quantify the water cycle in annual and monthly surface terrestrial or atmospheric terms. This work combines both the atmospheric and terrestrial components of the water cycle over the TP and its drainage basins and at time scales ranging from daily to long term.
The primary concern in this work is the net water flux (P − E) and its variability. The net water flux is estimated and described by Eq. (1). Previous studies suggested that the atmospheric moisture convergence term in reanalyses is more accurate in approximating P − E than the reanalysis diagnostic of P − E (Bosilovich et al. 2017). Reanalysis P − E is computed from the integration of the background forecast model in the analysis cycle, while the convergence term involves analyzed state variables of wind and moisture from observations. The analysis increments lead to an imbalance in the atmospheric water budget. The TP is a data-poor region. Here, we focus on the ERA5 P − E dataset and draw comparisons with the residual moisture budget in Eq. (1). Characterizing the water balance in ERA5 is a useful analysis of the quality and usability of hydrological studies over the TP.
ERA5, as the latest reanalysis from ECMWF, supersedes previous ERA-Interim data; it has a resolution of 31 km and 137 levels to 0.01 hPa, as well as an updated model physics and an improved data assimilation process (Hersbach et al. 2020). ERA5 provides hourly estimates of a large number of atmospheric, land, and ocean climate variables. Previous studies have shown that improvements in ERA5 include a better representation of developing weather events, cloud cover extents, surface irradiance parameters, atmospheric water fluxes, and other variables with respect to the former ERA-Interim dataset (Lei et al. 2020; Eicker et al. 2020; Yan et al. 2020; Zhao and Zhou 2019). In this work, the parameters P, E, total column water vapor, and vertically integrated moisture convergence from ERA5 are employed to investigate the atmospheric water cycle. The degree to which ERA5 satisfies the physical constraints of the water cycle over the TP and its drainage basins is of particular interest.
GRACE TWSA retrievals permit the validation of net water flux estimates derived from the reanalysis based on Eq. (2). This method has been employed to evaluate the closure of the terrestrial water budget over a number of river basins in Europe and around the globe with monthly data (Springer et al. 2014; Lorenz et al. 2014). The nonstandard daily time series of ITSG-Grace2018 represents a new and independent way to constrain the net water flux of ERA5 (Eicker et al. 2020).
Daily GRACE TWSAs from ITSG-Grace2018 are employed in this work. Validation of ITSG-Grace2018 with ERA5 and ERA-Interim has shown that GRACE-derived water-flux data produce realistic, high-frequency (5–30 days) information (Eicker et al. 2020). Changes in the TWSAs of two subsequent days are meaningless because the full sample period collected over the globe for GRACE is approximately 4–5 days. In according to Eq. (2), the daily GRACE TWSA is linked to integrated anomalies of net water flux and river discharge R (deviations from the climatology of annual mean). The daily GRACE TWSA can also be converted to water flux estimates by applying a numerical differentiation filter. Therefore, the terrestrial water balance is considered to represent storage changes (m3) or water fluxes (m3 day−1).
Independent observations of river discharge are available at two hydrological stations, Zhimenda (33.02°N, 97.13°E) in the upper Yangtze River Basin and Tangnaihai (35.5°N, 100.15°E) in the upper Yellow River Basin, over the TP. Daily streamflow records from 1983 to 2012 were provided by the Ministry of Water Resources of China (Liu et al. 2017). The terrestrial water balance is analyzed in the UYZR and UYLR basins.
The joint atmospheric–terrestrial water cycle over the TP and its drainage basins is described by ERA5-derived net water flux and GRACE-derived TWSA data. A preliminary assessment of the discrepancies between the model-derived and space-based observations in the atmospheric–terrestrial water cycle over the TP is provided. Validation helps to identify the development of the latest atmospheric reanalysis ERA5 and GRACE data that constrain the water cycle independence. Without observed discharge records collected over the TP and other drainage basins, comparisons of GRACE-derived TWSA with ERA5-derived estimates demonstrate the agreement of variability in hydrometeorological fluxes. This study provides valuable information for researchers to gain a better understanding and an improved interpretation of the variability in regional water resources. It is vital for ensuring societal benefits over the TP and surrounding areas.
c. Using the Lanczos filter to derive water fluxes at periods of 5–30 days
Daily GRACE TWSA data are used to investigate high-frequency hydrometeorological fluxes over the TP and its basins. The Lanczos filter (Duchon 1979) is applied to isolate fluxes at periods in the range of 5–30 days. Submonthly variations are of interest in this work, as the daily GRACE data include additional information compared to the standard monthly solutions (Eicker et al. 2020; Kvas et al. 2019). Bandpass filtered water fluxes derived from GRACE provide a meaningful link with submonthly variations in surface water dynamics from ERA5 on various scales covering the TP and its drainage basins.
3. ERA5-derived net water flux
There have been many estimates of water cycle terms via Eq. (1). Figure 2 shows the climatology of the net water flux (mm day−1) estimates made during January, April, July, and October in the period covering 1979–2019 from ERA5. The reanalysis P − E dataset is compared with the residual moisture budget. Similarities between the two estimates are evident. Both estimates display distinct spatial and temporal characteristics. The net water flux estimates over the TP are impacted by westerly winds in winter and the southwest monsoon in summer and transitions in spring and autumn (Ye and Gao 1979). Due to the “heat pump effect” the TP attracts water vapor from surrounding oceans and seas and becomes a wet pool during summer (Wu et al. 2015).
Climatology of the net water flux estimates (P − E vs the residual
Citation: Journal of Climate 34, 15; 10.1175/JCLI-D-20-0692.1
Climatology of the net water flux estimates (P − E vs the residual
Citation: Journal of Climate 34, 15; 10.1175/JCLI-D-20-0692.1
Climatology of the net water flux estimates (P − E vs the residual
Citation: Journal of Climate 34, 15; 10.1175/JCLI-D-20-0692.1
Ideally, the net water flux estimates are identical on average, but model biases and the analysis increments preclude reanalyses from water cycle closure constraints. Differences between the ERA5 P − E dataset and its residuals are shown in Fig. 2. Biases are much smaller over the TP than in its surrounding areas. Biases over the TP are distinct at the drainage basin scale. The P − E estimate is always larger than the residual over the western and southern basins and smaller over the center (the Qaidam and the Inner TP basins). Persistent biases suggest a static impact, such as that of topography. Over the eastern TP, the overestimate of P − E is dominant in spring, but the opposite is true in summer. Changing biases imply an effect from outside of the TP.
Figure 3 compares the probability distributions (PDFs) of the net water flux estimates (mm day−1) over the TP during January and July. The joint probability distributions throughout the whole period of 1979–2019 are also calculated. The PDFs of the net water flux estimates and the moisture convergence terms have similar features and represent dry and wet conditions, respectively. The tendency of total column water vapor is a small term with a Gaussian distribution structure. The joint probability highlights the likelihood of the net water flux estimates over the TP. Its RMSE equals 0.38. The high correlation (0.94) and linear regression between the P − E term and the residuals close to 1:1 suggest an ERA5-derived atmospheric water balance over the TP.
(top) Probability distributions of the net water flux estimates (P − E vs the residual) and components of the residual
Citation: Journal of Climate 34, 15; 10.1175/JCLI-D-20-0692.1
(top) Probability distributions of the net water flux estimates (P − E vs the residual) and components of the residual
Citation: Journal of Climate 34, 15; 10.1175/JCLI-D-20-0692.1
(top) Probability distributions of the net water flux estimates (P − E vs the residual) and components of the residual
Citation: Journal of Climate 34, 15; 10.1175/JCLI-D-20-0692.1
The joint probability distributions of the net water flux estimates are examined over 12 drainage basins of the TP in Fig. 4. The atmospheric water balance is maintained at the basin scale. The resulting linear regressions are close to the 1:1 line, especially over the western and southern basins (Amu Dayra, Indus, Ganges, Brahmaputra, Salween, Mekong, and Yangtze). The net water fluxes there are relatively large and contribute the most to the TP. Atmospheric water imbalance is evident in Qaidam and other northeastern basins (Yellow and Hexi). Overestimates of P − E are dominant under dry conditions, but the opposite case is true under wet conditions. Significant correlations between net water flux estimates of 0.7 or more are found in most basins, excluding the Qaidam basin.
Joint probability distribution of the net water flux estimates (mm day−1) over the 12 basins of the TP in the period of 1979–2019 of ERA5. Regression lines are shown in black.
Citation: Journal of Climate 34, 15; 10.1175/JCLI-D-20-0692.1
Joint probability distribution of the net water flux estimates (mm day−1) over the 12 basins of the TP in the period of 1979–2019 of ERA5. Regression lines are shown in black.
Citation: Journal of Climate 34, 15; 10.1175/JCLI-D-20-0692.1
Joint probability distribution of the net water flux estimates (mm day−1) over the 12 basins of the TP in the period of 1979–2019 of ERA5. Regression lines are shown in black.
Citation: Journal of Climate 34, 15; 10.1175/JCLI-D-20-0692.1
Figure 5 presents the monthly mean net water flux estimates and the four components in Eq. (1) over the TP. The given unit (× 108 m3) is integrated values taking the area into account. The tendency term (−∂W/∂t) is small, so the primary balance lies between the net water flux P − E and the moisture convergence. The annual cycle is clear between wet and dry seasons over the TP, while P is the largest term. Compared with individual values of P and E, the net water flux estimates and the moisture convergence values are consistent and relatively small. Thus, the uncertainty in the net water flux estimates is relatively small. The corresponding high temporal correlations (approximately 0.9) are fairly stable. This implies that the net water flux estimates from ERA5 are in good agreement throughout the year.
Monthly mean of the net water flux estimates (×108 m3; red line for P − E and black line for the residual) over the TP, together with four components (P, E, and
Citation: Journal of Climate 34, 15; 10.1175/JCLI-D-20-0692.1
Monthly mean of the net water flux estimates (×108 m3; red line for P − E and black line for the residual) over the TP, together with four components (P, E, and
Citation: Journal of Climate 34, 15; 10.1175/JCLI-D-20-0692.1
Monthly mean of the net water flux estimates (×108 m3; red line for P − E and black line for the residual) over the TP, together with four components (P, E, and
Citation: Journal of Climate 34, 15; 10.1175/JCLI-D-20-0692.1
Figure 6 shows the monthly mean values of the net water flux estimates and its components over 12 drainage basins of the TP. The monsoon system, with a rainy season, contributes most of the water resources over the TP and its drainage basins, excluding the Amu Dayra and Indus Basins. The westerlies impact the west basins and lead to the moisture convergence and rainy seasons occurring in winter and spring. Consistent with Fig. 2, differences between the P − E estimates and the residuals show spatial and temporal characteristics. Overestimations are evident in the western and southern basins (Amu Dayra, Indus, Ganges, Brahmaputra, and Salween), but underestimations occur over the eastern and northern basins (Yellow, Hexi, and Qaidam) during summer. Daily variations are highly correlated over basins in the south. Correlations decrease over the eastern and northern basins, especially during dry seasons. There is little correlation of the net water flux estimates over Qaidam. Consistent with the features shown in Fig. 4, the analysis increment and simulation biases in P and E preclude ERA5 from providing water cycle closure constraints in arid regions. The imbalance of the atmospheric water budget in arid basins has little impact on the total water balance over the TP, which is dominated by southern basins such as the Brahmaputra and Yangtze basins.
As in Fig. 5, but over the 12 drainage basins of the TP.
Citation: Journal of Climate 34, 15; 10.1175/JCLI-D-20-0692.1
As in Fig. 5, but over the 12 drainage basins of the TP.
Citation: Journal of Climate 34, 15; 10.1175/JCLI-D-20-0692.1
As in Fig. 5, but over the 12 drainage basins of the TP.
Citation: Journal of Climate 34, 15; 10.1175/JCLI-D-20-0692.1
Table 1 summarizes the annual mean of the net water flux estimates over the TP and the contributions of each drainage basin. The net water flux P − E is comparable to the residual over the TP and the scale of its basins. The net water fluxes over the Brahmaputra and Yangtze basins contribute over 50% to the total net water flux over the TP. Toward the central and northern basins (Hexi, Qaidam, Tarim, and Inner TP), the net water fluxes decrease and contribute less than 10% to the TP.
A brief evaluation implies that ERA5 maintains the atmospheric water balance over the TP. Variations in the net water flux estimates satisfy the physical constraints on a daily scale over the drainage basins. ERA5 is capable of providing useful information to evaluate changes in hydrometeorological fluxes over the TP and its drainage basins. It is vital for society to track changes in water availability over the TP as it affects billions of people.
4. Joint atmospheric–terrestrial water cycle over the UYZR and UYLR
ERA5-derived net water fluxes, GRACE-derived TWSAs, and independent observations of river streamflow collected at two hydrological stations (Zhimenda and Tangnaihai) are employed to investigate the atmospheric–terrestrial water cycle over the upper Yangtze and Yellow River basins on the TP via the terrestrial water balance equation, Eq. (2). To allow a meaningful comparison with the TWSA, the net water flux and river streamflow are converted to an integration of anomalies (deviations from climatology of the annual mean). Therefore, the terrestrial water balance is discussed in terms of storage changes and the imbalanced terrestrial water estimate (m3).
Figure 7 shows the climatology of daily streamflow anomalies (×108 m3 day−1) at Zhimenda and Tangnaihai, together with the ERA5-derived net water flux anomalies in the corresponding basins of the UYZR and UYLR. The annual mean climatology is removed. Similar to other basins on the TP, the climatology of the ERA5-derived net water flux P − E anomalies is consistent with the residual term over the UYZR and UYLR. The amplitudes of streamflow are approximately one-third smaller than the net water flux. Lag correlation analysis indicates that the smoothly varied streamflow lags behind the rapidly changing net water flux by 24 days in the UYZR and by 17 days in the UYLR to achieve their maximum correlations (0.85 and 0.77, respectively).
Climatological anomalies of the daily mean streamflow (×108 m3 day−1; river discharge R in brown lines) and ERA5-derived net water flux estimates (P − E in solid black lines; the residual in dashed blue lines) are shown over the (top) upper Yangtze River basin and (bottom) upper Yellow River basin. Integration of P − E − R anomalies and climatology of GRACE-derived TWSA (×108 m3, purple and red lines) are also shown.
Citation: Journal of Climate 34, 15; 10.1175/JCLI-D-20-0692.1
Climatological anomalies of the daily mean streamflow (×108 m3 day−1; river discharge R in brown lines) and ERA5-derived net water flux estimates (P − E in solid black lines; the residual in dashed blue lines) are shown over the (top) upper Yangtze River basin and (bottom) upper Yellow River basin. Integration of P − E − R anomalies and climatology of GRACE-derived TWSA (×108 m3, purple and red lines) are also shown.
Citation: Journal of Climate 34, 15; 10.1175/JCLI-D-20-0692.1
Climatological anomalies of the daily mean streamflow (×108 m3 day−1; river discharge R in brown lines) and ERA5-derived net water flux estimates (P − E in solid black lines; the residual in dashed blue lines) are shown over the (top) upper Yangtze River basin and (bottom) upper Yellow River basin. Integration of P − E − R anomalies and climatology of GRACE-derived TWSA (×108 m3, purple and red lines) are also shown.
Citation: Journal of Climate 34, 15; 10.1175/JCLI-D-20-0692.1
The climatology of integrated anomalies of the net water flux and associated streamflow (the imbalanced terrestrial water estimate P − E minus river discharge, P − E − R) over the UYZR and UYLR basins are then compared considering the TWSA based on the ITSG-Grace2018 daily data shown in Fig. 7. GRACE-derived TWSAs are linked to the imbalanced terrestrial water estimate of P − E − R via Eq. (2). Good correspondence between TWSAs and the imbalanced terrestrial water estimate is distinguished over the UYLR basin, with a similar annual cycle time series with comparable amplitudes ranging from −60 to 60 × 108 m3. The corresponding relationship is complex over the UYZR basin. The imbalanced terrestrial water estimate is similar to that over the UYLR basin, but the amplitude of the TWSAs decreases by half. In addition, the time series over the UYZR basin demonstrates a small accumulation of TWSAs during the early portion of the year. The missing water may imply a significant amount of groundwater outflow (Yong et al. 2021). The discrepancy suggests that the joint atmospheric–terrestrial water cycle is imbalanced with current datasets.
Figure 8 extends the time series of observed streamflow anomalies and ERA5-derived net water fluxes, the integrated anomalies of P − E − R, and the GRACE-derived TWSAs over the UYZR and UYLR basins during 2003–12. The streamflow anomalies are small compared with the net water flux. The net water flux contributes the most to the imbalanced terrestrial water estimate. Correlations between the integrations of P − E − R anomalies and corresponding TWSAs are similar between the two basins (0.45 and 0.47, respectively). In addition to their interannual variations, there are significant long-term trends in the GRACE-derived TWSAs.
Anomalies of the streamflow and ERA5-derived net water flux (×108 m3 day−1; brown and black lines) over the (top) UYZR and (bottom) UYLR basins during 2003–12. Integrations of P − E − R anomalies and GRACE-derived TWSA (×108 m3; purple and red lines) are also shown.
Citation: Journal of Climate 34, 15; 10.1175/JCLI-D-20-0692.1
Anomalies of the streamflow and ERA5-derived net water flux (×108 m3 day−1; brown and black lines) over the (top) UYZR and (bottom) UYLR basins during 2003–12. Integrations of P − E − R anomalies and GRACE-derived TWSA (×108 m3; purple and red lines) are also shown.
Citation: Journal of Climate 34, 15; 10.1175/JCLI-D-20-0692.1
Anomalies of the streamflow and ERA5-derived net water flux (×108 m3 day−1; brown and black lines) over the (top) UYZR and (bottom) UYLR basins during 2003–12. Integrations of P − E − R anomalies and GRACE-derived TWSA (×108 m3; purple and red lines) are also shown.
Citation: Journal of Climate 34, 15; 10.1175/JCLI-D-20-0692.1
Figure 9 shows ERA5-derived integrations of the net water flux (P − E) anomalies during 1979–2019, integrated anomalies of P − E − R during 1982–2012, and GRACE-derived TWSAs during 2002–16 over the UYZR and UYLR. Extensive records show features consistent with those in Fig. 8. The integrated anomalies of P − E only and P − E − R are consistent. They both are comparable with the TWSAs. The achieved significant temporal correlations are, respectively, 0.44 and 0.46 over the UYZR basin and 0.56 and 0.61 over the UYLR basin. The slight increases in the obtained correlations suggest a small contribution by streamflow. Nevertheless, the integration of the net water flux anomalies is an undoubtedly ERA5-derived major component corresponding to the GRACE-derived TWSAs.
ERA5-derived integrations of the net water flux P − E anomalies (×108 m3; black) over the (top) UYZR and (bottom) UYLR basins in the whole period of 1979–2019. Integrations of P − E − R anomalies (purple) during 1983–2012, and counterparts of GRACE-derived TWSAs during 2002–16 are also shown.
Citation: Journal of Climate 34, 15; 10.1175/JCLI-D-20-0692.1
ERA5-derived integrations of the net water flux P − E anomalies (×108 m3; black) over the (top) UYZR and (bottom) UYLR basins in the whole period of 1979–2019. Integrations of P − E − R anomalies (purple) during 1983–2012, and counterparts of GRACE-derived TWSAs during 2002–16 are also shown.
Citation: Journal of Climate 34, 15; 10.1175/JCLI-D-20-0692.1
ERA5-derived integrations of the net water flux P − E anomalies (×108 m3; black) over the (top) UYZR and (bottom) UYLR basins in the whole period of 1979–2019. Integrations of P − E − R anomalies (purple) during 1983–2012, and counterparts of GRACE-derived TWSAs during 2002–16 are also shown.
Citation: Journal of Climate 34, 15; 10.1175/JCLI-D-20-0692.1
5. Long- and short-term atmospheric–terrestrial water cycle over the TP
The ERA5-derived integration of the net water flux anomalies is compared with the GRACE-derived TWSAs over the TP in Fig. 10. In the 2002–16 period, the amplitudes of ERA5- and GRACE-derived water storage changes were comparable. The temporal correlation is significant at approximately 0.59. The long-term trends show consistent and significant increases over the TP during 2002–16. In addition, the ERA5-derived estimate shows decadal time-scale variations during the long period of 1979–2019.
ERA5-derived integration of the net water flux anomalies (× 108 m3) over the TP in the 1979–2019 period and GRACE-derived TWSA during 2002–16.
Citation: Journal of Climate 34, 15; 10.1175/JCLI-D-20-0692.1
ERA5-derived integration of the net water flux anomalies (× 108 m3) over the TP in the 1979–2019 period and GRACE-derived TWSA during 2002–16.
Citation: Journal of Climate 34, 15; 10.1175/JCLI-D-20-0692.1
ERA5-derived integration of the net water flux anomalies (× 108 m3) over the TP in the 1979–2019 period and GRACE-derived TWSA during 2002–16.
Citation: Journal of Climate 34, 15; 10.1175/JCLI-D-20-0692.1
Figure 11 shows the ERA5-derived integrations of the net water flux anomalies and GRACE-derived TWSAs over the 12 drainage basins of the TP. The amplitudes of the two estimates are comparable at the scale of some basins (Amu Dayra, Indus, Ganges, Brahmaputra, Yangtze, and Yellow River basins). Significant temporal correlations (approximately 0.5 or more) are also evident there. The highest correlation over Amu Dayra (0.78) implies good correspondence between the ERA5- and GRACE-derived estimates, while decadal variations are also distinct there. Over relatively small or arid basins (Salween, Mekong, Hexi, Qaidam, Tarim, and Inner TP), the obtained correlations decrease to 0.3 or less. It may be impacted by the ratio of signal and noise in both the ERA5 and GRACE datasets.
As in Fig. 10, but over the 12 drainage basins of the TP.
Citation: Journal of Climate 34, 15; 10.1175/JCLI-D-20-0692.1
As in Fig. 10, but over the 12 drainage basins of the TP.
Citation: Journal of Climate 34, 15; 10.1175/JCLI-D-20-0692.1
As in Fig. 10, but over the 12 drainage basins of the TP.
Citation: Journal of Climate 34, 15; 10.1175/JCLI-D-20-0692.1
Opposite trends in GRACE-derived TWSAs are evident in the southern and northern basins. Descending trends occur in the Indus, Ganges, Brahmaputra, Salween, and Mekong basins. Ascending trends are dominant in the northern basins (Yangtze, Yellow, Hexi, Qaidam, Tarim, and Inner TP). The ERA5-derived estimates have similar features of opposing long trends in the southern and northern basins during 1979–2019. However, more consistent increasing trends in the ERA5-derived estimates are evident within the short period of 2002–16. Long-term variations in the ERA5- and GRACE-derived atmospheric–terrestrial water cycles, especially at the basin scale, require further investigation in the future.
Submonthly variations in the ERA5- and GRACE-derived hydrometeorological fluxes are investigated in Fig. 12. In contrast to conventional monthly GRACE products, the daily time series of TWSAs include high-frequency hydrometeorological fluxes at periods between 5 and 30 days (Eicker et al. 2020). Bandpass filtering analysis is performed to isolate submonthly variations in the hydrometeorological fluxes over the 12 drainage basins, the UYZR and UYLR basins, and the TP. Considering the availability of continuous records in ITSG-Grace2018, Fig. 12 shows the results during 2004. ERA5- and GRACE-derived hydrometeorological fluxes show generally good agreement with comparable amplitudes over the TP and at the basin scale. Significant temporal correlations (approximately 0.6 or more) are evident over the TP and the southern basins (Brahmaputra, Salween, Mekong, and Yangtze).
Comparison of the bandpass filtered daily time series during 2004 from the GRACE-derived TWSAs (×108 m3 day−1; red) and the ERA5-derived integration of the net water flux anomalies (black) over the 12 drainage basins, UYZR and UYLR basins, and the TP.
Citation: Journal of Climate 34, 15; 10.1175/JCLI-D-20-0692.1
Comparison of the bandpass filtered daily time series during 2004 from the GRACE-derived TWSAs (×108 m3 day−1; red) and the ERA5-derived integration of the net water flux anomalies (black) over the 12 drainage basins, UYZR and UYLR basins, and the TP.
Citation: Journal of Climate 34, 15; 10.1175/JCLI-D-20-0692.1
Comparison of the bandpass filtered daily time series during 2004 from the GRACE-derived TWSAs (×108 m3 day−1; red) and the ERA5-derived integration of the net water flux anomalies (black) over the 12 drainage basins, UYZR and UYLR basins, and the TP.
Citation: Journal of Climate 34, 15; 10.1175/JCLI-D-20-0692.1
GRACE-derived TWSAs and ERA5-derived water fluxes provide independent ways to describe the atmospheric–terrestrial water cycle. By comparison of ERA5- and GRACE-derived estimates, consistent long- and short-term variations are found over the TP. Discrepancies are evident in long trends at the drainage basin scale. Over relatively small or arid basins, the correspondence of short-term variations is weak. The ratio of signal and noise in both ERA5 and GRACE may cause discrepancies.
6. Discussion and conclusions
The better representation of the atmospheric–terrestrial water cycle is a goal of recent reanalysis efforts because water availability is vital for human and society. Validation of the ERA5-derived atmospheric–terrestrial water fluxes is performed over the TP and drainage basin scales in this work. The regional scale for which the water budget is locally close within a specific threshold (in this case 10%); that is, LWB = (P − E)/P < 0.1 (Dagan et al. 2019). The degree of closure for the local water budget depends on how similar P and E are (a perfect P = E is achieved only on the global scale). Over the TP region, the mean scale of LWB is 0.54 based on ERA5. Therefore, the net water flux (P − E) is heavily affected by moisture transport from the surrounding areas (Wu et al. 2015). ERA5 maintains the atmospheric water balance over the TP. Although model biases and the analysis increments preclude closure of the water cycle in arid basins (Qaidam, Yellow, and Hexi), variations in the net water flux estimates satisfy the physical constraints on a daily scale over the drainage basins of the TP. Given the importance of water cycle variability, the accuracy of ERA5 is evident over the TP and its drainage basins. It is concluded that ERA5 is capable of providing realistic variations in the net water flux and has great potential for hydrometeorological applications over the TP.
ITSG-Grace2018 provides an opportunity to evaluate the terrestrial water budget in a new and completely independent way (Eicker et al. 2020; Kvas et al. 2019). Daily GRACE-derived TWSA data offer great potential for assessing model-derived atmospheric–terrestrial fluxes via water budget-based methods (Lorenz et al. 2014). The TWSAs from ITSG-Grace2018 deviate from a long-term mean gravity field model. The limited satellite ground track coverage of 1 day does not allow for stable Kalman smoother-based global gravity field inversion based on the track patterns of GRACE, and the information content on daily time scales is fully sampled approximately every 4–5 days in a nonrepeating orbit cycling the globe. Because GRACE observations with fewer than 10 000 observations per day were excluded, the time series of TWSAs includes many data gaps over the TP and its drainage basins. Data gaps have little impact on monthly and longer time-scale variations, but they affect the high-pass filtering process. Therefore, bandpass filtering focusing on the submonthly variations was performed on a continuous daily time series collected during 2004.
Good correspondence between the GRACE-derived TWSAs and ERA5-derived net water flux anomalies was found over the TP and at the scale of some drainage basins via the terrestrial water budget. The ERA5-derived integration of the net water flux anomalies as a major component is mostly balanced by GRACE-derived TWSAs over the TP and some of its basins. Consistent long-term trends are evident over the TP in GRACE and ERA5. The disagreement of long-term trends between GRACE and ERA5 on the basin scales is difficult to explain. Uncertainty in both datasets may induce different long-term trends on the basin scale, which requires further study in the future.
With limited time spans, short-term variations in GRACE- and ERA5-derived hydrometeorological fluxes show good agreement over the TP. The correlation coefficients of the two time series reach 0.6 or more, especially in basins with large flux occurring in summer. The correlation drops to approximately 0.4 over the two western basins where large fluxes occur in spring. The reasons why the correlation is the lowest in the Inner Basin are unclear, but the GRACE-derived flux dataset shows large variability during winter and spring. This implies that other processes in or under the surface influence GRACE-derived flux estimates (Yong et al. 2021).
In summary, this study investigates ERA5-derived net water flux estimates over the TP and its drainage basins. The atmospheric–terrestrial water cycle is represented by ERA5-derived net water fluxes, GRACE-derived TWSAs and daily streamflow measures collected at Zhimenda and Tangnaihai. Agreements and discrepancies between the model-derived and space-based observations in the atmospheric–terrestrial water cycle are discussed. The main results are summarized as follows:
The net water fluxes in the TP and the scale of drainage basins are closely tied to local dynamical and physical processes and to large-scale circulation and atmospheric water vapor. The net water flux P − E is nearly balanced by the estimate of the atmospheric branch (atmospheric moisture convergence and tendency of precipitable water) via the atmospheric water budget. ERA5 maintains the atmospheric water balance over the TP. Variations in the net water flux estimates satisfy the physical constraints on a daily scale over drainage basins.
The ERA5-derived integration of net water flux anomalies is a major component and corresponds to GRACE-derived TWSAs together with streamflow anomalies over the UYZR and UYLR basins via the terrestrial water balance equation.
The water budget-based approach with ERA5 and ITSG-Grace2018 provides realistic variations in hydrometeorological fluxes and water storage changes. The ERA5- and GRACE-derived estimates contain consistent long- and short-term atmospheric–terrestrial water cycles over the TP. Discrepancies are evident at the scale of drainage basins.
The documented good correspondence between the GRACE- and ERA5-derived atmospheric–terrestrial water cycles highlights the potential value of both datasets in the rational application of water resource information over the TP.
Acknowledgments
This paper was jointly funded by the Strategic Priority Research Program of the Chinese Academy of Sciences (Grant XDA20100300) and the Second Tibetan Plateau Scientific Expedition and Research Program (STEP; Grant 2019QZKK0206). ERA5 data were developed by ECWMF and supplied by the Climate Data Store (data used here can be found online at https://cds.climate.copernicus.eu/). ITSG-Grace2018 data are publicly available from TU Graz (https://www.tugraz.at/institute/ifg/downloads/gravity-field-models/itsg-grace2018). Discharge data are provided by Dr. X. Liu, who also contributed valuable discussions.
REFERENCES
Bosilovich, M. G., F. R. Robertson, L. Takacs, A. Molod, and D. Mocko, 2017: Atmospheric water balance and variability in the MERRA-2 reanalysis. J. Climate, 30, 1177–1196, https://doi.org/10.1175/JCLI-D-16-0338.1.
Brown, P. J., and C. D. Kummerow, 2014: An assessment of atmospheric water budget components over tropical oceans. J. Climate, 27, 2054–2071, https://doi.org/10.1175/JCLI-D-13-00385.1.
Dagan, G., P. Stier, and D. Watson-Parris, 2019: Analysis of the atmospheric water budget for elucidating the spatial scale of precipitation changes under climate change. Geophys. Res. Lett., 46, 10 504–10 511, https://doi.org/10.1029/2019GL084173.
Duchon, C. E., 1979: Lanczos filtering in one and two dimensions. J. Appl. Meteor. Climatol., 18, 1016–1022, https://doi.org/10.1175/1520-0450(1979)018<1016:LFIOAT>2.0.CO;2.
Eicker, A., L. Jensen, V. Wöhnke, H. Dobslaw, A. Kvas, T. Mayer-Gürr, and R. Dill, 2020: Daily GRACE satellite data evaluate short-term hydro-meteorological fluxes from global atmospheric reanalyses. Sci. Rep., 10, 4504, https://doi.org/10.1038/s41598-020-61166-0.
Feng, L., and T. Zhou, 2012: Water vapor transport for summer precipitation over the Tibetan Plateau: Multidata set analysis. J. Geophys. Res., 117, D20114, https://doi.org/10.1029/2011JD017012.
Gao, Y. H., C. Lan, and Y. Q. Zhang, 2014: Changes in moisture flux over the Tibetan Plateau during 1979–2011 and possible mechanisms. J. Climate, 27, 1876–1893, https://doi.org/10.1175/JCLI-D-13-00321.1.
Hersbach, H., and Coauthors, 2020: The ERA5 global reanalysis. Quart. J. Roy. Meteor. Soc., 146, 1999–2049, https://doi.org/10.1002/qj.3803.
Immerzeel, W. W., and M. F. P. Bierkens, 2012: Asia’s water balance. Nat. Geosci., 5, 841–842, https://doi.org/10.1038/ngeo1643.
Kerandi, N., J. Arnault, P. Laux, S. Wagner, J. Kitheka, and H. Kunstmann, 2017: Joint atmospheric–terrestrial water balances for East Africa: A WRF-Hydro case study for the upper Tana River basin. Theor. Appl. Climatol., 131, 1337–1355, https://doi.org/10.1007/s00704-017-2050-8.
Kunstmann, H., B. Fersch, A. Bárdossy, B. Devaraju, and N. Sneeuw, 2012: Continental-scale basin water storage variation from global and dynamically downscaled atmospheric water budgets in comparison with GRACE-derived observations. J. Hydrometeor., 13, 1589–1603, https://doi.org/10.1175/JHM-D-11-0143.1.
Kvas, A., S. Behzadpour, M. Ellmer, B. Klinger, S. Strasser, N. Zehentner, and T. Mayer-Gürr, 2019: ITSG-Grace2018: Overview and evaluation of a new GRACE-only gravity field time series. J. Geophys. Res. Solid Earth, 124, 9332–9344, https://doi.org/10.1029/2019JB017415.
Lei, Y., H. Letu, H. Shang, and J. Shi, 2020: Cloud cover over the Tibetan Plateau and eastern China: A comparison of ERA5 and ERA-Interim with satellite observations. Climate Dyn., 54, 2941–2957, https://doi.org/10.1007/s00382-020-05149-x.
Liu, X., T. Yang, K. Hsu, C. Liu, and S. Sorooshian, 2017: Evaluating the streamflow simulation capability of PERSIANN-CDR daily rainfall products in two river basins on the Tibetan Plateau. Hydrol. Earth Syst. Sci., 21, 169–181, https://doi.org/10.5194/hess-21-169-2017.
Lorenz, C., H. Kunstmann, B. Devaraju, M. J. Tourian, N. Sneeuw, and J. Riegger, 2014: Large-scale runoff from landmasses: A global assessment of the closure of the hydrological and atmospheric water balances. J. Hydrometeor., 15, 2111–2139, https://doi.org/10.1175/JHM-D-13-0157.1.
Oki, T., K. Musiake, H. Matsuyama, and K. Masuda, 1995: Global atmospheric water balance and runoff from large river basins. Hydrol. Processes, 9, 655–678, https://doi.org/10.1002/hyp.3360090513.
Peixoto, J. P., and A. H. Oort, 1992: Physics of Climate. American Institute of Physics, 520 pp.
Rodell, M., and Coauthors, 2015: The observed state of the water cycle in the early twenty-first century. J. Climate, 28, 8289–8318, https://doi.org/10.1175/JCLI-D-14-00555.1.
Springer, A., J. Kusche, K. Hartung, C. Ohlwein, and L. Longuevergne, 2014: New estimates of variations in water flux and storage over Europe based on regional (re)analyses and multisensor observations. J. Hydrometeor., 15, 2397–2417, https://doi.org/10.1175/JHM-D-14-0050.1.
Swenson, S., and J. Wahr, 2006: Estimating large-scale precipitation minus evapotranspiration from GRACE satellite gravity measurements. J. Hydrometeor., 7, 252–270, https://doi.org/10.1175/JHM478.1.
Tang, Q. H., and Coauthors, 2019: Streamflow change on the Qinghai-Tibet Plateau and its impacts (in Chinese). Chin. Sci. Bull., 64, 2807–2821, https://doi.org/10.1360/TB-2019-0141.
Tapley, B. D., S. Bettadpur, J. C. Ries, P. F. Thompson, and M. M. Watkins, 2004: GRACE measurements of mass variability in the Earth system. Science, 305, 503–505, https://doi.org/10.1126/science.1099192.
Trenberth, K. E., and C. J. Guillemot, 1998: Evaluation of the atmospheric moisture and hydrological cycle in the NCEP/NCAR reanalyses. Climate Dyn., 14, 213–231, https://doi.org/10.1007/s003820050219.
Trenberth, K. E., J. T. Fasullo, and J. Mackaro, 2011: Atmospheric moisture transports from ocean to land and global energy flows in reanalyses. J. Climate, 24, 4907–4924, https://doi.org/10.1175/2011JCLI4171.1.
Wang, Z., A. Duan, S. Yang, and K. Ullah, 2017: Atmospheric moisture budget and its regulation on the variability of summer precipitation over the Tibetan Plateau. J. Geophys. Res., 122, 614–630, https://doi.org/10.1002/2016JD025515.
Wu, G., and Coauthors, 2015: Tibetan Plateau climate dynamics: Recent research progress and outlook. Natl. Sci. Rev., 2, 100–116, https://doi.org/10.1093/nsr/nwu045.
Xu, X., C. Lu, X. Shi, and S. Gao, 2008: World water tower: An atmospheric perspective. Geophys. Res. Lett., 35, L20815, https://doi.org/10.1029/2008GL035867.
Yan, H., J. Huang, Y. He, Y. Liu, T. Wang, and J. Li, 2020: Atmospheric water vapor budget and its long-term trend over the Tibetan Plateau. J. Geophys. Res., 125, e2020JD033297, https://doi.org/10.1029/2020JD033297.
Ye, D. Z., and Y. X. Gao, 1979: Meteorology of the Qinghai-Xizang Plateau. Chinese Science Press, 278 pp.
Yong, B., C.-Y. Wang, J. Chen, J. Chen, D. A. Barry, T. Wang, and L. Li, 2021: Missing water from the Qiangtang Basin on the Tibetan Plateau. Geology, https://doi.org/10.1130/G48561.1.
Zhang, G., T. Yao, H. Xie, S. Kang, and Y. Lei, 2013: Increased mass over the Tibetan Plateau: From lakes or glaciers? Geophys. Res. Lett., 40, 2125–2130, https://doi.org/10.1002/grl.50462.
Zhao, Y., and T. Zhou, 2019: Asian water tower evinced in total column water vapor: A comparison among multiple satellite and reanalysis data sets. Climate Dyn., 54, 231–245, https://doi.org/10.1007/s00382-019-04999-4.
