Anthropogenic impacts on mass change in North China
Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, Beijing, China
Corresponding author: Q. Tang, Institute of Geographic Sciences and Natural Resources Research, Chinese Academy of Sciences, No. 11A, Datun Rd., Chaoyang District, Beijing 100101, China. (email@example.com)
 The gravity fields from the Gravity Recovery and Climate Experiment (GRACE) satellite have been used to estimate groundwater depletion in many parts of the world. Groundwater depletion assessment in North China, however, is affected by large anthropogenic disturbance on mass variation. Using the water survey data and coal flow statistics, we show that the mass variations in North China during 2003–2011 were largely affected by reservoir regulation, water diversion, and coal transport. According to our estimates, the mass loss rate of groundwater depletion (−14 to −8.4 mm yr−1) was largely offset by the mass gains caused by reservoir regulation (1.3 mm yr−1), water diversion (4.4 mm yr−1), and coal transport (1.9 mm yr−1) in the North China Plain. We note that anthropogenic impacts on mass change must be taken into account when using the GRACE data to estimate groundwater change in the area intensively altered by human activities.
 The land surface has been utterly transformed by anthropogenic activities in North China which is one of the most uniformly and extensively altered areas by humans. Human alterations to the hydrologic cycles, solid earth, and ecosystems are substantial in this area, and its effects are manifested by changes in the gravity field. The distribution and flow of mass are dominated directly by humanity in North China. No land mass variations at the Earth's surface, including those inside the surface fluid envelops [Tang et al., 2008; Ministry of Water Resources of China (MWR), 2013] and at the surface of the solid Earth [Mou and Li, 2012], are free of pervasive human influence in this area. Among the components of mass variation, groundwater is an important resource for socioeconomic development [Cao et al., 2013] and is usually unfeasible to assess without labor intensive and expensive monitoring well networks. The satellite Gravity Recovery and Climate Experiment (GRACE) can provide mass variation measurements closely related to changes in terrestrial water storage [Syed et al., 2008]. It is thus of importance to determine the contributing sources of the mass variation and isolate groundwater storage change [Scanlon et al., 2012; Voss et al., 2013].
 GRACE detects combined mass change that includes signals from hydrology, solid earth, cryosphere, oceans, atmosphere, and tides [Jacob et al., 2012; Wang et al., 2013]. Although the nonhydrologic signals are removed as far as possible in the GRACE terrestrial water storage (TWS) estimates [Landerer and Swenson, 2012], the signal of anthropogenic mass flow such as bulk commodity transport is not considered during the data processing. Therefore, the GRACE TWS variations include the combined contributions of groundwater, soil water, surface water, snow, ice, biomass [Rodell et al., 2009], and the anthropogenic mass changes. Interannual variations in biomass have been shown to be small [Rodell et al., 2005], and the natural variations in the TWS components (except for groundwater) may be obtained from the land surface hydrologic models [Rodell et al., 2009; Tang et al., 2010; Pokhrel et al., 2012]. To estimate groundwater storage variation, the information on the generally region-specific anthropogenic alterations of mass variations [Chao, 1995; Wang et al., 2011; National Bureau of Statistics of China (NBS), 2012; Pokhrel et al., 2012] is needed and the anthropogenic impacts on the mass variations (other than groundwater storage variation) must be removed from the GRACE TWS data.
 Agriculture in North China Plain (NCP) relies heavily on groundwater, the unsustainable use of which has resulted in depletion of fossil groundwater resources [Foster et al., 2004; Cao et al., 2013]. Some studies have estimated the rate of groundwater depletion in North China [Liu et al., 2001; Wada et al., 2010; Cao et al., 2013]. These assessments are based on monitoring well networks which are typically unsystematic or model computations which are suffered from uncertainties in estimating groundwater recharge and extraction. Although a few other studies have used the temporal variation of the gravity fields measured by the GRACE satellites to assess groundwater depletion in the NCP [Moiwo et al., 2013; Feng et al., 2013], they isolated groundwater storage variation from the GRACE data without sufficient considerations for the changes in other mass components caused by anthropogenic activities such as reservoir regulation [Ren et al., 2002], interbasin water diversion [MWR, 2013], and bulk commodity transport [Mou and Li, 2012]. This study aims to assess the anthropogenic impacts on mass change and to investigate the effect of anthropogenic mass distribution on the GRACE-based estimates of groundwater depletion in North China.
2 Data and Methods
 We collected the coal transport data from China Statistical Yearbooks provided by the National Bureau of Statistics of China [NBS, 2012] and water resources data from Water Resources Bulletins provided by the Ministry of Water Resources of China [MWR, 2013]. The net coal mass change was calculated as the imported amount minus the exported amount and was transformed to equivalent water in thickness (see supporting information). The Water Resources Bulletins provided the total reservoir storage at the end of each year. The total reservoir storage was distributed to the reservoirs in proportion to the gross storage capacity. The equivalent water thickness of reservoir storage was calculated as the reservoir storage divided by the area of the administrative zone where the reservoirs are located (see supporting information for details).
 The Variable Infiltration Capacity (VIC) model [Liang et al., 1994] was used to estimate the natural surface water variations. The model was driven by the gridded forcings based on the meteorological observations from China Meteorological Administration and was optimized using the naturalized streamflow (see supporting information). The VIC model accounted for the water storage variation in various natural surface water storage components (e.g., soil water, natural river channel, and snow and ice) but not groundwater and anthropogenic effects (e.g., reservoir regulation and interbasin water diversion). The VIC model did not account for lakes and glaciers. There are no glaciers and few large lakes in the study area. The largest lake in the NCP, Lake Baiyangdian, has a storage capacity of 1.1 km3 which is about 2% of the total gross storage capacity of the reservoirs in the study area [Cui et al., 2010]. The modeled evaporation and runoff, together with the precipitation climatology, were used to calculate natural surface water storage change according to the surface water balance equation [Tang et al., 2010]. The water resources survey data [MWR, 2013] and coal flow statistics [NBS, 2012] together with VIC model estimates were used to produce the time series of the contribution to total land mass variation due to various natural and anthropogenic factors in North China during 2003–2011. Besides the VIC estimates, the Noah estimates from the Global Land Data Assimilation System [Rodell et al., 2004] were used to produce a comparable time series of natural surface water and the uncertainties of the hydrologic modeling are analyzed.
 We used the latest GRACE TWS land products (RL05) provided by the Center for Space Research (CSR), GeoForschungsZentrum Potsdam (GFZ), and Jet Propulsion Laboratory (JPL). The monthly TWS was scaled by using the scaling factors provided with the data in order to restore much of the energy removed by destriping, filtering, and truncation processes [Landerer and Swenson, 2012]. We did not use the method performed in Scanlon et al.  and Feng et al.  in which the filtered groundwater storage was corrected (by applying scaling factor) using information on the spatial distribution of groundwater storage changes. Thus, we avoided using the groundwater statistical data to tune the GRACE signal. The water storages at the beginning of 2003 are arbitrarily set to zero for reference purpose. The groundwater storage was derived as the GRACE TWS minus the sum of the natural surface water storage from the hydrologic model (VIC or Noah), reservoir storage, interbasin water diversion when applicable, and coal mass equivalent water.
 Anthropogenic impacts on surface water and bulk commodity mass flow specially merit considerations in North China. The river systems have been strongly affected by dams in our study area (Figure 1a). There are 16 large reservoirs in the study area, capable of holding 52.6 km3 of water [MWR, 2013], or 141 mm water thickness if the water amount of the gross storage capacity was evenly distributed over the study area (Tables S1 and S2 in the supporting information). The bulk commodity transport is not to be neglected. The rapid development of transport infrastructure has facilitated the bulk commodity transport and boosted China's economy [Yu et al., 2012]. Coal transport is a major bulk commodity mass flow, accounting for about half of China's railway commodity transportation and a great deal of truck transportation [Mou et al., 2012]. In this study, we consider coal but not other bulk commodities because only the coal mass flow data are systematically available in the statistical yearbooks [NBS, 2012] and coal transport is the main mass component of the bulk commodity transport in North China. This should not undermine the potential importance of certain bulk commodity in particular areas such as crude oil in the Middle East [Aleklett et al., 2010] and iron ore in Western Australia [Yellishetty et al., 2010]. The west region of the study area (Figure S2 and Table S2) is a coal-producing region where coal is transported to the coastal demand centers around Beijing, Shanghai, and Hong Kong [Mou et al., 2012]. The accumulated net coal exportation in the west region is 6.64 × 109 metric tons during 2003–2011, equivalent to 28.5 mm water in thickness if water amount of the same weight coal was evenly distributed over the west region (Figure 1b). The accumulated net coal importation in the east region (Figure S2 and Table S1) is 2.31 × 109 metric tons during 2003–2011, equivalent to 16.6 mm water thickness if the same weight water amount was evenly distributed over the east region (Figure 1b).
 Figure 2 plots the linear trend of the GRACE TWS from 2003 to 2011. Although the linear trends are different in the change magnitude and spatial coverage for the GRACE TWS products from different sources and between the estimates using the original products and the scaled data, all the GRACE-based TWS estimates show a general decreasing trend over North China in 2003–2011 (Figures S1 and S2). The GRACE TWS data show significant decreasing trend across the entire study area, including the east region where the monitoring well network demonstrates groundwater depletion and the west region where groundwater depletion is small [MWR, 2012; Feng et al., 2013]. The hotspot of groundwater depletion is the east region, i.e., the NCP, where the Ministry of Water Resources of China provides the estimates of shallow aquifer storage change according to the monitoring well network observations (Figure S3).
 Figure 3 shows the time series of the contributions (natural surface water storage, reservoir storage, coal transport, interbasin water diversion when applicable, and groundwater) of the GRACE TWS from 2003 to 2011. The linear trends of the time series are shown in Table 1. The statistical significance of the trends was calculated according to the two-tailed Student's t test. We note the limitations of trend analysis on short samples. The GRACE TWS peaked at the end of the wet year 2003 [MWR, 2012] and decreased thereafter. The storages of natural surface water varied with high storage in wet years and low storage in dry years [MWR, 2012]. Both the VIC and Noah models showed there was a peak of natural surface water storage in 2003 while the Noah-estimated peak seemed much larger than that of VIC (Figures 3c and 3d). The uncertainty of natural surface water estimate may affect the groundwater estimates (Figures 3a and 3b). Nonetheless, both models show that the trends in natural surface water storage are statistically insignificant (Figure 3 and Table 1). Reservoir storages increased during 2003–2011, indicating that more water was impounded behind the dams. Coal transport has caused a steady mass loss in the west region (Figure 3a) and mass gain in the east region (Figure 3b). The trends of the coal transport are statistically significant during the study period (Table 1). After removing the contribution of coal mass loss, reservoir storage, and natural surface water storage, the estimated groundwater depletion at the end of 2011 (comparing to the beginning of 2003) is 16–21 mm in the west region, a value smaller than the coal mass loss (28.5 mm). The trend of the estimated groundwater storage in the west region is insignificant, with a value between −4.33 (VIC estimated) to 2.65 (Noah estimated) mm yr−1. The coal transport has caused mass loss in the west region at a rate of −3.13 mm yr−1. In the east region, the signal of groundwater depletion was largely offset by the mass gains from reservoir regulation, interbasin water diversion (Table S3), coal transport, and nature surface water storage change. Comparing to the beginning of 2003, the GRACE TWS decline at the end of 2011 is 9 mm, which is much smaller than the estimated groundwater depletion (112–136 mm). The trend of GRACE TWS is insignificant with a value of −5.37 mm yr−1 while the estimated rate of groundwater depletion is significant with a value between about −14 (VIC estimated) and −8.4 (Noah estimated) mm yr−1. The estimated groundwater depletion in the east region is close to the estimated change in shallow aquifer (92 mm during the study period or at a rate of −12 mm yr−1) from the monitoring well network. The monitoring well network did not account for deep aquifers, and thus may underestimate the total groundwater depletion in the NCP [Cao et al., 2013]. Meanwhile, our approach is conservative because other anthropogenic mass such as that of building materials and living materials, which are likely accumulating in the NCP, is not considered. Our estimates show that comparing to the beginning of 2003, the accumulated mass gain at the end of 2011 caused by natural surface water, reservoir, interbasin water diversion, and coal transport is 14–39, 26, 46, and 16.6 mm in equivalent water thickness, respectively. The insignificant trend of natural surface water ranges from 1.14 (VIC) to −4.5 (Noah) mm yr−1. The trend of reservoir, interbasin water diversion, and coal transport is 1.3, 4.4, and 1.85 mm yr−1, respectively. These results suggest that the GRACE TWS in North China was largely affected by both natural and anthropogenic mass changes (Figure S5). The GRACE TWS decrease includes the coal mass loss signal in the west region, whereas the groundwater depletion signal in the GRACE TWS was largely offset by reservoir storage increase, water diversion, and coal mass gain in the east region. The results suggest that the anthropogenic impacts on mass variation must be taken into account when using the GRACE TWS to estimate groundwater depletion in North China.
Table 1. Linear Trends of the Contributions of the Scaled GRACE TWS During 2003–2011
 Boosted by its rapid economic growth, China has largely altered the Earth's surface, including its gravity field. We use the water resources survey data and coal flow statistics together with land surface hydrologic models to produce the time series of the contribution to total land mass variation due to various natural and anthropogenic factors in North China between 2003 and 2011. Our results show that the GRACE-detected TWS decline during the above period is small because the mass loss signal of groundwater depletion (112–136 mm) was largely offset by the mass gains of natural surface water (14–39 mm), reservoir (26 mm), water diversion (46 mm), and coal transport (16.6 mm) in the North China Plain. The estimated rate of groundwater depletion ranged from −14 (VIC estimated) to −8.4 (Noah estimated) mm yr−1 which was offset by reservoir (1.3 mm yr−1), water diversion (4.4 mm yr−1), and coal transport (1.85 mm yr−1) caused mass changes. The anthropogenic mass redistribution has to be accounted for in studies on groundwater depletion as the induced equivalent water thickness change would have a comparable magnitude to the groundwater depletion in North China. The results show the importance of anthropogenic impacts on mass variation in the area intensively altered by human activities. The study further demonstrates the capabilities of GRACE to detect the effects of the exploitation of natural resources such as groundwater and fossil resources.
 We thank Editor M. Bayani Cardenas and two anonymous reviewers for their comments and suggestions that improved the manuscript. This work has been supported by the National Natural Science Foundation of China (grant 41171031), National Basic Research Program of China (grant 2012CB955403), and Hundred Talents Program of the Chinese Academy of Sciences.
 The Editor thanks Chunmiao Zheng and an anonymous reviewer for their assistance in evaluating this paper.