GRACE-based estimates of water discharge over the Yellow River basin

g e o d e s y a n d g e o d y n a m i c s 2 0 1 6 , v o l 7 n o 3 , 1 8 7 e1 9 3

Available online at www.sciencedirect.com

ScienceDirect journal homepage: www.keaipublishing.com/en/journals/geog; http://www.jgg09.com/jweb_ddcl_en/EN/volumn/home.shtml

GRACE-based estimates of water discharge over the Yellow River basin Qiong Lia, Bo Zhongb,c,*, Zhicai Luob,c, Chaolong Yaob a

MOE Key Laboratory of Fundamental Physical Quantities Measurement, School of Physics, Huazhong University of Science and Technology, Wuhan 430074, China b School of Geodesy and Geomatics, Wuhan University, Wuhan 430079, China c Key Laboratory of Geospace Environment and Geodesy, Ministry of Education, Wuhan University, Wuhan 430079, China

article info

abstract

Article history:

As critical component of hydrologic cycle, basin discharge is a key issue for understanding

Received 8 February 2016

the hydrological and climatologic related to water and energy cycles. Combining GRACE

Accepted 13 April 2016

gravity field models with ET from GLDAS models and precipitation from GPCP, discharge of

Available online 1 June 2016

the Yellow River basin are estimated from the water balance equation. While comparing the results with discharge from GLDAS model and in situ measurements, the results reveal

Keywords:

that discharge from Mosaic and CLM GLDAS model can partially represent the river

GRACE

discharge and the discharge estimation from water balance equation could reflect the

Gravity field model

discharge from precipitation over the Yellow River basin.

Terrestrial water storage

© 2016, Institute of Seismology, China Earthquake Administration, etc. Production and

Water discharge

hosting by Elsevier B.V. on behalf of KeAi Communications Co., Ltd. This is an open access

Yellow River basin

article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

How to cite this article: Li Q, et al., GRACE-based estimates of water discharge over the Yellow River basin, Geodesy and Geodynamics (2016), 7, 187e193, http://dx.doi.org/10.1016/j.geog.2016.04.007.

1.

Introduction

Global warming expedites the global water cycle process, and leads to the intensity variations of the regional precipitation and evapotranspiration as well as the water discharge. Water discharge is also a kind of important water resource, and the temporal and spatial variations of regional water

discharge result in difficult exploitation of surface water and watershed management. As the critical component of hydrologic cycle, basin discharge is a key issue for understanding the hydrological and climatologic related to the water and energy cycles [1]. Currently the seasonal and inter-annual variability of land-scale surface water changes, including water discharge, relying on sparse in situ gauge measurement and partially verified hydrological models. In

* Corresponding author. School of Geodesy and Geomatics, Wuhan University, Wuhan 430079, China. E-mail address: [email protected] (B. Zhong). Peer review under responsibility of Institute of Seismology, China Earthquake Administration.

Production and Hosting by Elsevier on behalf of KeAi

http://dx.doi.org/10.1016/j.geog.2016.04.007 1674-9847/© 2016, Institute of Seismology, China Earthquake Administration, etc. Production and hosting by Elsevier B.V. on behalf of KeAi Communications Co., Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

188

g e o d e s y a n d g e o d y n a m i c s 2 0 1 6 , v o l 7 n o 3 , 1 8 7 e1 9 3

situ gauge measurements could quantify the value of the river channel runoff, but little information about subsurface runoff [2]. Thought mostly hydrological model driving by real-time hydro-meteorological data such as precipitation, surface temperature, lacking of spatially measurement of regional basin characteristic make the hydrological model unable to estimate the water cycle component including water discharge properly. The Gravity Recovery and Climate Experiment (GRACE) mission, launched on March 17, 2002, provides a unique opportunity to detect continental water storage variations [3]. GRACE's temporal gravity field has been used to detect the water storage changes at a global or very large scale, and to quantify fluxes and storages for the validation and improvement of the terrestrial water balance in global land surface hydrological models [4e7]. Syed et al. [8] proposed a method for estimating monthly river basin outflows based on GRACE satellite measurements in a coupled water balance equation, which was tested on the Amazon and Mississippi River basin showing good correlation with observed streamflow. Syed et al. [9] used the same method to estimate monthly fresh water discharge from continents, drainage regions, and global land, and the comparisons to observations indicate that the method shows important potential for global-scale monitoring discharge at near-realtime. Sproles et al. [10] used GRACE measurements to estimate runoff in three regional-scale watersheds of Columbia River basin, USA and Canada, and discussing the hysteresis loops between runoff and terrestrial water storage, subsurface water and groundwater. The Yellow River (Huanghe) is the second largest river in China, and the annual runoff accounts for 2% of the whole country but offers 12% water consumption in china. As the most important water resource in the Yellow River basin, the variations of the water discharge in Yellow River affects the personal life directly and restricted the development of social economy. In this paper, we estimate the Yellow River discharge combined with the monthly terrestrial water storage based on GRACE observation, precipitation and evapotranspiration from space observation. The GRACE-based estimates of water discharge are compared with the in situ gauge station observation over the sub-streams of the Yellow River basin.

laser ranging (SLR) will be used to substitute for the C20 series from GRACE in the study [11]. Owing to the hybrid effects of the satellite orbit errors, ocean and atmospheric mode errors and the correlated errors of Stokes coefficients, the surface mass variations exhibit obvious NeS strip and high-frequency errors while calculate without filtering. In this paper, a hybrid filtering scheme with the combination of de-correlation filter P3M6 [12] and 300 km Fan filter [13] is taken in the subsequent calculations. Their calculation models are as follows: Dhðq; lÞ ¼

∞ l  X 2aprave X 2l þ 1 ~ lm cosðmlÞ ,Wl Plm ðcos qÞ,Wm DC 3rwater l¼0 1 þ kl m¼0  þ DS~lm sinðmlÞ

(1) where Wl and Wm are the smoothing kernel related to order 2b  1b, and degree respectively, namely W0 ¼ 1, W1 ¼ 1þe 1e2b ln 2 W þ W , b ¼ , r is the filtering radius, Wnþ1 ¼ 2nþ1 n n1 b 1cosðr=aÞ ~ lm and DS~lm are the variations of Stokes coefficients and DC after de-correlated filtering respectively.

2.2.

Estimation of terrestrial freshwater discharge

The basic water balance equation can be described as: dS ¼ PðtÞ  RðtÞ  ETðtÞ dt

where t is the time, S is the terrestrial water storage inferred by GRACE, P and ET are the basin-wide totals of precipitation and evapotranspiration, R represented as total basin discharge, or the net surface and groundwater outflow [8]. The TWS variations recovered from GRACE Spherical Harmonic solutions were converted into the anomaly with respect to a mean gravity field of a selected period, so called terrestrial water storage anomaly (TWSA). While estimating the GRACE-based ET by water balance method, dS dt is referred to a monthly scale value, or so called terrestrial water storage change (TWSC). We used a briefly and efficient way to calculate the TWSC following the method provided in Ramillien et al. [14]. The TWSC of month ith could be denoted as: TWSCðiÞ ¼

2.

Data and methods

2.1.

GRACE data and TWS

137 approximately monthly average gravity field models from April 2002 to December 2014 are used in the paper, provided by RL05 solutions from the Center for Space Research (CSR), University of Texas at Austin. Each monthly gravity field consists of fully normalized spherical harmonic coefficients Clm and Slm to degree and order 96. Though comparing with CSR RL04, the degree-2 zonal harmonic C20 of RL05 have been improved, it is still of relatively low quality, due in part to the orbital geometry and the short separation between the satellites. The time series of C20 from satellite

(2)

TWSAði þ 1Þ  TWSAði  1Þ 2

(3)

We used ten years monthly gravity field model covering the period from 2003 to 2013, with six month data missing due to technical failure. The TWSA missing data were retrieved by interpolation for calculating TWSC as integrated as possible. Here the Global Precipitation Climatology Project (GPCP) monthly precipitation analysis (version 2) is used to obtain monthly estimates of precipitation averaged over the Yellow River basin and several sub-streams respectively. The GPCP produces global analyses of monthly precipitation available from 1979 to the present on a 2.5  2.5 grid. It is a merged analysis that incorporates precipitation estimates from microwave-based data from the Special Sensor Microwave Imager (SSM/I), geosynchronous-orbit satellite infrared data and surface rain gauging data from over more than 6000

g e o d e s y a n d g e o d y n a m i c s 2 0 1 6 , v o l 7 n o 3 , 1 8 7 e1 9 3

stations [15]. The GPCP data sets also provide error estimates for each field with both sampling and measurementalgorithm effects taken into account. Meanwhile bias effects in the final product are assumed independent to random errors due to systematic effects removed from the individual datasets prior to the merging process [16]. So the GPCP data provide a priori monthly precipitation products, with characterize of sophisticated quality control, low bias errors from the input information. The Global Land Data Assimilation System (GLDAS) provide global hydrological models with monthly time interval and 1  1 gridded resolution, which integrate satellite data and Land Surface Model data assimilation technique to generate global distribution of land surface states such as soil moisture, snow water equivalent, canopy plant water, evapotranspiration, surface and subsurface runoff and so on [17]. The four land surface models are Noah, Mosaic, the community land model (CLM) and variable infiltration capacity model (VIC). In this paper, the ET, Discharge and TWS are totally estimated from GLDAS models. The total river discharge and the TWS are the sum of surface and sub-surface runoff and the sum of soil moisture, snow water equivalent and canopy plant water respectively. In order to comparing the estimation discharge with in situ observations, we also used the 3 gauging stations locating on the main stream of the Yellow River basin with data from 2003 to 2012, which provide by Huanghe water resources commission of the ministry of water resources. The name of the three gauging stations are Tangnaihai (TNH), Toudaoguai (TDG) and Huayuankou(HYK), and catchment area for this three gauging stations are the source of the Yellow River basin, the upper stream, and the middle stream respectively. The locations of the three gauging stations are marked on Fig. 1.

Fig. 1 e Long trend of terrestrial water storage with EWH (Unit: mm/a) over the Yellow River basin from GRACE gravity models.

189

Fig. 2 e Time series of the (a)-TWSA and (b)-TWSC over the whole Yellow River basin.

3.

Results and discussion

3.1.

TWS over the Yellow River basin

Here 1  1 grid terrestrial water storages over the Yellow River basin were estimated from GRACE monthly gravity with the 300 km Fan filter and the P3M6 de-correlation filter. And the 1  1 grid monthly TWSA were obtained from each monthly TWS minus mean TWS of the whole time span. While removing the annual and semi-annual signals, the trends of each grid were estimated by least square method. Fig. 1 shows the long trend with EWH over the Yellow River basin. The Region of the Yellow River basin was superimposed as grey line, the white line is the boundaries of upper and middle reaches and the main stream of the Yellow River basin were marked with blue line. The watershed of the Yellow River basin are provided by the National Science & Technology Infrastructure of China, Data Sharing Infrastructure of Earth System Science e LakeWatershed Science Data Center (http://lake.data.ac.cn). As can be seen from Fig. 1, the TWS appears slowly increased over the source of the Yellow River basin and decrease over the lower stream of the Yellow River basin. Feng et al. [18] estimated the TWS changes over North China span from 2003 to 2010, and the TWS decrease were explained as the overexploitation of groundwater in most provinces especially in Beijing, Tianjin and Hebei province. This result agrees well with the long trend showed in Fig. 1. Here the time series from area weighted summing for each grid in the estimated region are regarded as the time series of regional average changes. The time series results from GRACE, GLDAS models as well as the GPCP data in each region discussed below are obtained from this method. The time series of TWSA over the Yellow River basin estimated from GRACE monthly gravity filed models and GLDAS global hydrological models were showed in Fig. 2(a). In order to comparing results

190

g e o d e s y a n d g e o d y n a m i c s 2 0 1 6 , v o l 7 n o 3 , 1 8 7 e1 9 3

based GRACE and GLDAS models reasonably, the GLDAS gridded models were converted to Spherical harmonic coefficients with maximum degree of 96, which is the same with the CSR RL05 models. And then the 300 km Fan filter were also used without the de-correlation filter. However the amplitudes were reduced as well as the high frequency errors for using the spatial smoothing filter reduced the errors. Here introducing an average scaling factor to retrieve the signals as mentioned in Long et al. [19]. The averaged scaling factors over the Yellow River basin is 1.25 for CSRRL05withdegree of 96 which minimizing the root-meansquare (RMS) differences between the filtered and original GLDAS-NOAH-based TWS. Fig. 2(a) shows that the GRACE and GLDAS based TWSA both exhibit seasonal signal, and the amplitude value of both results in 2002and 2003 appears abnormally negative and positive value, respectively. Wang et al. [20] analyzed the precipitation over the Yellow River basin for the past several decades, and annual precipitation anomaly in 2002 and 2003 appears opposite and positive value, well consistent with the result in Fig. 2(a). For river discharge estimation, the monthly TWSC were calculated and the time series were showed in Fig. 2(b).

monthly amplitude of R1eR4 is about 22.33 mm, it means that the discharge has been less affected by the ET from four LSMs GLDAS models. The discharge estimated from the formula above is so called GRACE based discharge for convenience description below.

3.2.

Fig. 3 e a-Time series of the Yellow River basin precipitation anomaly and TWSA from GRACE; b-Time series of the Yellow River discharge from GRACE estimation and precipitation from GPCP.

Discharge over the Yellow River basin

The river discharge can be estimated from the water balance equation while the basin is considering as a closed drainage system. The Yellow River basin originates in the northeast Tibet Plateau, and snow melting in this area makes it one of the main runoff formation areas. Therefore, while calculating the discharge of the whole Yellow River basin, the snow melting from the river source should be considered. The TNH gauging station is a control hydrological station on the Yellow River source and the catchment area is 12.19  104 km2. The discharge observed in TNH station represents the total discharge deduced by the regional precipitation and the snow melting. Thus, the discharge of ith month over the whole Yellow River basin (excluding the river source area) can be calculated as: RðiÞ ¼ PðiÞ þ RTNH ðiÞ  ETðiÞ  TWSCðiÞ

(4)

In the formula above, the TWSC is recovered from GRACE monthly gravity field models from equations (1) and (3), the P is provided by the GPCP model, and the ET is the GLDAS model with four LSMs. The results R1eR4 corresponding to Noah, Mosaic, VIC and CLM as well as the time series of monthly precipitation from GPCP are illustrated in Fig. 3. The positive anomaly of the TWSA in Fig. 3(a) is well consistent with the strong precipitation and the correlation coefficient between GPCP precipitation and TWSC is 0.51. There is a scope of timing delay between precipitation anomaly and TWSA from Fig. 3(a). It can be inferred that precipitation is one of the main reason deducing the terrestrial water storage changes in this basin. The ET from four LSMs GLDAS models show slightly difference, which cause the differences of R1eR4 in Fig. 3. While removing the mean value of the four discharge results from each time series, the root-mean-square of the residual are 1.66, 1.98, 3.03 and 3.63 respectively. Whereas the mean

The GLDAS models based four LSMs also provided the surface and sub-surface runoff. Fig. 4 present the time series of discharge based GRACE and each GLDAS models, and the correlation and slope between each group. It can be seen from Fig. 4 (a1ed1) that the amplitudes of each discharge of GLDAS model are less than the GRACE based discharge, and similar conclusion could be derived from each slope in Fig. 4 (a2ed2). The correlation coefficient between GRACE based discharge and the discharge from GLDAS model based Mosaic and CLM LSMs are 0.855 and 0.891 respectively, which reveals strong consistency. The two results appears similar seasonal variations and amplitude changes, except for the absolute amplitude of GRACE based discharge are nearly twice as much as from GLDAS based Mosaic and CLM LSMs. The correlation coefficient between GRACE based discharge and the discharge from GLDAS model based VIC is 0.26, and the results from VIC hardly reveal the discharge of the basin. It can be concluded that the sum of surface and sub-surface runoff data from GLDAS model based Mosaic and CLM LSMs can partially represent the river discharge over the Yellow River basin.

3.3.

Discharge of sub-stream

The Yellow River basin were divided into three sub-stream based on the feature of the regional climate and topography. The upper reaches constitute a segment starting from its source and ending at Hekou Town (in Inner Mongolia), where the TDG gauging station located. This segment has a total length of 3472 km and total basin area of 386,000 square

g e o d e s y a n d g e o d y n a m i c s 2 0 1 6 , v o l 7 n o 3 , 1 8 7 e1 9 3

191

Fig. 4 e Time series of the Yellow River discharge from GRACE estimation and the GLDAS models. a,b,c,d represent the Noah, Mosaic, VIC and CLM respectively.

kilometres, 51.4% of the total basin area. The part of the Yellow River basin between Hekou Town and Zhengzhou (in Henan province) constitutes the middle reaches, and the controlling hydrological station is HYK station. There are large amount of mud and sand discharged into the river, and make the Yellow River the most sediment-laden river in the world. The middle reaches are 1206 km long, with a basin area of 344,000 square kilometres, 45.7% of the total basin area. The distribution of the sub-stream and the location of the gauging stations are shown in Fig. 1. The lower reaches of the Yellow River basin constitute from the rest part from HYK station and the basin area is only 23,000 square kilometres. It is too small to detect from GRACE mission and the low reaches are not discussed in this paper. Fig. 5 shows the GPCP precipitation and GRACE based discharge over the sub-streams. It is can be seen that the average precipitation increased gradually from upper to middle stream, which is coherent with the variations of spatial distribution, i.e., with increased from northeast to southwest Yellow River basin [20]. There are similar variations for the two sub-streams in Fig. 5. In order to analyze the influence on river discharge, we estimated the correlation coefficients between discharge and precipitation, and soil moisture respectively. The correlation coefficient

between discharge and precipitation are 0.89 and 0.92 respectively for the upper and middle stream. To some

Fig. 5 e Comparing the time series of precipitation and discharge over the three sub-stream of the Yellow River basin.

192

g e o d e s y a n d g e o d y n a m i c s 2 0 1 6 , v o l 7 n o 3 , 1 8 7 e1 9 3

extent, this is due to the precipitation used while calculating the river discharge from the water balance equation. The smaller coefficient in upper stream could reveal that the precipitation influence on upper river discharge is smaller than middle stream areas. Fig. 6 presents the time series of discharge from GRACE estimation and the two control hydrological stations observations for sub-stream. There are two peaks (March and August) for each year from the time series of the TDG station observations over the upper stream (Fig. 6(a)). It is possibly due to the catchment area of TDG station includes the river source and the snow melting in spring is part of main source for the Yellow River basin. And the GRACE based discharge estimated from water balance equation only detect the discharge deduced by the regional precipitation. Compared with the upper stream, there are similar seasonal variations in both discharge series of the middle stream. It is concluded that precipitation is the main content for the discharge in the middle of the Yellow River basin and besides precipitation, snow melting also contributes dominantly to the water discharge in the upper Yellow River basin.

Fig. 6 e Comparing monthly discharge over the two substream from GRACE estimation and in situ observations.

4.

Conclusion

Combining the GRACE monthly gravity field models, global hydrological models from GLDAS and the precipitation from GPCP, the water discharge over the Yellow River basin are estimated based on the water balance equation. Based on comparison with the discharge of GLDAS model, it can be concluded that the sum of surface and sub-surface runoff data from GLDAS model based Mosaic and CLM LSMs can partially represent the river discharge over the Yellow River basin. Finally, the agreement between the GRACE based discharge variations and the gauging station measurement reveals that

the discharge estimation from water balance equation could reflect the discharge from precipitation. While estimating the total basin discharge, the water source such as snow and permafrost melting should be considered.

Acknowledgement This research was jointly funded by the National 973 Project China (2013CB733302) and National Natural Science Foundation of China (41504014, 41474019). The authors would like to thank CSR providing the RL05 gravity models and the GES DISC providing the GLDAS models.

references

[1] Hassan AA, Jin S. Lake level change and total water discharge in East Africa Rift Valley from satellite-based observations. Glob Planet Change 2014;117:79e90. [2] Papa F, Prigent C, Rossow WB. Monitoring flood and discharge variations in the large Siberian Rivers from a multi-satellite technique. Surv Geophys 2008;29:297e317. [3] Tapley BD, Bettadpur S, Ries JC, Thompson PF, Watkins MM. GRACE measurements of mass variability in the Earth system. Science 2004;305:503e6. [4] Wahr J, Smeed DA, Leuliette E, Swenson S. Seasonal variability of the red sea, from satellite gravity, radar altimetry, and in situ observations. J Geophys Res (Oceans) 2014;119:5091e104. [5] Famiglietti JS, Rodell M. Water in the balance. Science 2013;340:1300e1. [6] Reager JT, Famiglietti JS. Global terrestrial water storage capacity and flood potential using GRACE. Geophys Res Lett 2009;36:L23402. [7] Rodell M, Velicogna I, Famiglietti JS. Satellite-based estimates of groundwater depletion in India. Nature 2009;460:999e1002. [8] Syed TH, Famiglietti JS, Chen J, Rodell M, Seneviratne SI, Viterbo P, et al. Total basin discharge for the Amazon and Mississippi River basins from GRACE and a land-atmosphere water balance. Geophys Res Lett 2005;32:24404. http:// dx.doi.org/10.1029/2005GL024851. [9] Syed TH, Famiglietti JS, Chambers DP. GRACE-based estimates of terrestrial freshwater discharge from basin to continental scales. J Hydrometeorol 2009;10:22e40. http:// dx.doi.org/10.1175/2008JHM993.1. [10] Sproles EA, Leibowitz SG, Reager Jr JT, Wigington PJ, Famiglietti JS, Patil SD. GRACE storage-runoff hystereses reveal the dynamics of regional watersheds. Hydrol Earth Syst Sci 2015;19:3253e72. http://dx.doi.org/10.5194/hess-19-3253-2015. [11] Chen JL, Rodell M, Wilson CR, Famiglietti JS. Low degree spherical harmonic influences on Gravity Recovery and Climate Experiment (GRACE) water storage estimates. Geophys Res Lett 2005;32:14405. http://dx.doi.org/10.1029/ 2005GL022964. [12] Swenson S, Wahr J. Post-processing removal of correlated errors in GRACE data. Geophys Res Lett 2006;33:8402. http:// dx.doi.org/10.1029/2005GL025285. [13] Zhang ZZ, Chao BF, Lu Y, Hsu HT. An effective filtering for GRACE time-variable gravity: fan filter. Geophys Res Lett 2009;36:17311. http://dx.doi.org/10.1029/2009GL039459. [14] Ramillien G, Frappart F, Gu¨ntner A, Ngo-Duc T, Cazenave A, Laval K. Time variations of the regional evapotranspiration rate from Gravity Recovery and Climate Experiment (GRACE)

g e o d e s y a n d g e o d y n a m i c s 2 0 1 6 , v o l 7 n o 3 , 1 8 7 e1 9 3

[15]

[16]

[17]

[18]

[19]

satellite gravimetry. Water Resour Res 2006;42:10403. http:// dx.doi.org/10.1029/2005WR004331. Huffman GJ, Adler RF, Arkin P, Chang A, Ferraro R, Gruber A, et al. The global precipitation climatology project (GPCP) combined precipitation dataset. Bull Am Meteorol Soc 1997;78:5e20. Adler RF, Huffman GJ, Chang A, Ferraro R, Xie PP, Janowiak J, et al. The version-2 global precipitation climatology project (GPCP) monthly precipitation analysis (1979ePresent). J Hydrometeorol 2003;4:1147e67. Rodell M, Houser PR, Jambor U, Gottshalck J, Mitchell K, Meng CJ, et al. The global land data assimilation system. Bull Am Meteorol Soc 2004;85(3):381e94. Feng W, Zhong M, Lemoine JM, Biancale R, Hsu HT, Xia J. Evaluation of groundwater depletion in North China using the Gravity Recovery and Climate Experiment (GRACE) data and ground-based measurements. Water Resour Res 2013;49:2110e8. http://dx.doi.org/10.1002/wrcr.20192. Long D, Longuevergne L, Scanlon BR. Global analysis of approaches for deriving total water storage changes from

193

GRACE satellites. Water Resour Res 2015;51:2574e94. http:// dx.doi.org/10.1002/2014WR016853. [20] Wang N, Li DL, Zhang J. Variations and circulation pattern of drought and flood in summer at middle and upper of Yellow River. Arid Land Geogr 2012;35(5):754e63.

Qiong Li, a post doctor in School of Physics, Huazhong University of Science and Technology, majors in Satellite gravimetry. Her research interest focuses on satellite gravity field models recovery from GRACE mission and applications of GRACE temporal gravity field.