Improving Land Surface Hydrological Simulations in China Using CLDAS Meteorological Forcing Data

+ Author Affiliations + Find other works by these authors
  • Corresponding author: Chunxiang SHI, shicx@cma.gov.cn
  • Funds:

    Supported by the National Natural Science Foundation of China (91437220 and 41405083) and Fund Project for the Excellent Youth Scholars of the Education Department of Hunan Province in China (18B494)

  • doi: 10.1007/s13351-019-9067-0

PDF

  • The accuracy of land surface hydrological simulations using an offline land surface model (LSM) depends largely on the quality of the atmospheric forcing data. In this study, Global Land Data Assimilation System (GLDAS) forcing data and the newly developed China Meteorological Administration Land Data Assimilation System (CLDAS) forcing data are used to drive the Noah LSM with multiple parameterizations (Noah-MP) and to explore how the newly developed CLDAS forcing data improve land surface hydrological simulations over mainland China. The monthly soil moisture (SM) and evapotranspiration (ET) simulations are then compared and evaluated against observations. The results show that the Noah-MP driven by the CLDAS forcing data (referred to as CLDAS_Noah-MP) significantly improves the simulations in most cases over mainland China and its eight river basins. CLDAS_Noah-MP increases the correlation coefficient (R) values from 0.451 to 0.534 for the SM simulations at a depth range of 0–10 cm in mainland China, especially in the eastern monsoon area such as the Huang–Huai–Hai Plain, the southern Yangtze River basin, and the Zhujiang River basin. Moreover, the root-mean-square error is reduced from 0.078 to 0.068 m3 m−3 for the SM simulations, and from 12.9 to 11.4 mm month−1 for the ET simulations over mainland China, especially in the southern Yangtze River basin and Zhujiang River basin. This study demonstrates that, by merging more in situ and remote sensing observations in regional atmospheric forcing data, offline LSM simulations can better simulate regional-scale land surface hydrological processes.
  • 加载中
  • Fig. 1.  Spatial distributions of the 6-yr (2008–13) averaged volumetric SM (m3 m−3) at a depth range of 0–10 cm derived from the two simulations, the in situ site measurements, and the locations of the eight basins in mainland China: (a) GLDAS_Noah-MP, (b) CLDAS_Noah-MP, (c) the in situ site observations, and (d) the location of the eight basins in China. I: Songhuajiang River basin, II: Haihe River basin, III: Yellow River basin, IV: Huaihe River basin, V: Heihe River basin, VI: Tarim River basin, VII: Yangtze River basin, and VIII: Zhujiang River basin.

    Fig. 2.  Time series of the monthly volumetric SM at a depth range of 0–10 cm from in situ observations, GLDAS_Noah-MP, and CLDAS_Noah-MP during the period of 2008–13 in the eight major basins of China.

    Fig. 3.  Spatial distributions of the difference in R and RMSE (m3 m−3) between the modeled SM and the in situ site observations. (a) R difference: CLDAS_Noah-MP minus GLDAS_Noah-MP and (b) RMSE difference: CLDAS_Noah-MP minus GLDAS_Noah-MP.

    Fig. 4.  Spatial distributions for the 4-yr (2008–11) averaged ET (mm yr−1) derived from the two simulations and the observation-based Obs_MTE product in mainland China: (a) GLDAS_Noah-MP, (b) CLDAS_Noah-MP, and (c) Obs_MTE.

    Fig. 5.  Time series of the monthly ET from the observation-based Obs_MTE product, GLDAS_Noah-MP, and CLDAS_Noah-MP during the period of 2008–11 in the eight major basins of China.

    Fig. 6.  Spatial distributions of the difference in R between the modeled ET and the observation-based Obs_MTE product. (a) R of GLDAS_Noah-MP, (b) R of CLDAS_Noah-MP, and (c) R difference: CLDAS_Noah-MP minus GLDAS_Noah-MP.

    Fig. 7.  Spatial distributions of the difference in RMSE (mm month−1) between the modeled ET against the observation-based Obs_MTE product. (a) RMSE of GLDAS_Noah-MP, (b) RMSE of CLDAS_Noah-MP, and (c) RMSE difference: CLDAS_Noah-MP minus GLDAS_Noah-MP.

    Table 1.  Mean R, MBE (m3 m−3), RMSE (m3 m−3), and ubRMSE (m3 m−3) values between the simulated and measured SM at a 0–10-cm depth range from 2008 to 2013 over the eight basins and mainland China

    GLDAS_Noah-MPCLDAS_Noah-MP
    RRMSEMBEubRMSERRMSEMBEubRMSE
    Songhuajiang River
     basin (38 stations)
    0.387 (78.9%)0.0810.0530.0380.381 (66.7%, –1.6%)0.070 (–13.6%)0.0140.034 (–10.5%)
    Haihe River
     basin (28 stations)
    0.473 (85.2%)0.0600.0390.0380.533 (89.3%, +12.7%)0.049 (–18.3%)0.002 0.039 (+2.6%)
    Heihe River
     basin (10 stations)
    0.212 (30.0%)0.072–0.0330.0390.253 (40.0%, +19.0%)0.079 (+9.7%)–0.0520.039 (0)
    Tarim River
     basin (8 stations)
    0.256 (42.9%)0.076–0.0140.0350.257 (42.9%, +0.4%)0.095 (+25%)–0.0490.037 (+5.7%)
    Yellow River
     basin (84 stations)
    0.493 (81.5%)0.0820.0500.0380.622 (95.1%, +26.2%)0.065 (–20.7%)0.0130.034 (–10.5%)
    Huaihe River
     basin (25 stations)
    0.552 (96%)0.0690.0300.0360.638 (96%, +15.6%)0.061 (–11.6%)–0.0080.035 (–2.8%)
    Yangtze River
     basin (43 stations)
    0.501 (87.8%)0.0880.0450.0380.641 (95.3%, +27.9%)0.072 (–18.2%)0.0120.034 (–10.5%)
    Zhujiang River
     basin (7 stations)
    0.596 (100%)0.1150.0780.0360.740 (100%, +24.2%)0.087 (–25.2%)0.0320.034 (–5.6%)
    China (308 stations)0.451 (79.6%)0.0780.0390.0390.534 (86.2%, +18.4%)0.068 (–12.8%)0.0030.037 (–5.1%)
    Note: For the R values, the first number in the parentheses is the percentage of stations with a correlation that is statistically significant at the p = 0.05 level. The second number in the parentheses is the percentage increase relative to the paired experiment. For the RMSEs and ubRMSEs, the numbers in the parentheses are the percentages of reduction relative to the paired experiment.
    Download: Download as CSV

    Table 2.  Mean R and RMSE (mm month−1) values between the simulated and the observation-based Obs_MTE ET product from 2008 to 2011 over the eight river basins and mainland China

    GLDAS_Noah-MPCLDAS_Noah-MP
    RRMSERRMSE
    Songhuajiang River basin0.97510.60.962 (–1.3%)11.0 (+3.8%)
    Haihe River basin0.97310.90.962 (–1.1%)11.5 (+5.5%)
    Heihe River basin0.7338.30.718 (–2.0%)7.7 (–7.2%)
    Tarim River basin0.77210.00.706 (–8.5%)10.7 (+7.0%)
    Yellow River basin0.9608.90.954 (–0.6%)9.2 (+9.4%)
    Huaihe River basin0.93515.60.914 (–2.2%)16.9 (+8.3%)
    Yangtze River basin0.95914.70.961 (+2.1%)11.7 (–20.4%)
    Zhujiang River basin0.91821.50.923 (+0.5%)16.8 (–21.9%)
    China0.90812.90.901 (–0.7%)11.4 (–11.6%)
    Note: The numbers in the parentheses are the percentage of variations relative to the paired experiment.
    Download: Download as CSV
  • [1]

    Albergel, C., P. de Rosnay, G. Balsamo, et al., 2012: Soil moisture analyses at ECMWF: Evaluation using global ground-based in situ observations. J. Hydrometeor., 13, 1442–1460. doi: 10.1175/JHM-D-11-0107.1.
    [2]

    Beaudoing, H., and M. Rodell, 2016: GLDAS Noah Land Surface Model L4 3 hourly 0.25 × 0.25 degree V2.1. Goddard Earth Sciences Data and Information Services Center, Greenbelt, Maryland, USA.
    [3]

    Bi, H. Y., J. W. Ma, W. J. Zheng, et al., 2016: Comparison of soil moisture in GLDAS model simulations and in situ observations over the Tibetan Plateau. J. Geophys. Res. Atmos., 121, 2658–2678. doi: 10.1002/2015jd024131.
    [4]

    Chen, F., and J. Dudhia, 2001: Coupling an advanced land surface–hydrology model with the Penn State–NCAR MM5 modeling system. Part I: Model implementation and sensitivity. Mon. Wea. Rev., 129, 569–585. doi: 10.1175/1520-0493(2001)129<0569:CAALSH>2.0.CO;2.
    [5]

    Chen, F., Z. Janjić, and K. Mitchell, 1997: Impact of atmospheric surface-layer parameterizations in the new land-surface scheme of the NCEP mesoscale Eta model. Bound.-Layer Meteor., 85, 391–421. doi: 10.1023/A:1000531001463.
    [6]

    Chen, J., B. Z. Chen, T. A. Black, et al., 2013: Comparison of terrestrial evapotranspiration estimates using the mass transfer and Penman-Monteith equations in land surface models. J. Geophys. Res. Biogeosci., 118, 1715–1731. doi: 10.1002/2013JG002446.
    [7]

    Chen, Y., J. Z. Xia, S. L. Liang, et al., 2014: Comparison of satellite-based evapotranspiration models over terrestrial ecosystems in China. Remote Sens. Environ., 140, 279–293. doi: 10.1016/j.rse.2013.08.045.
    [8]

    Dirmeyer, P. A., X. Gao, M. Zhao, et al., 2006: GSWP-2: Multimodel analysis and implications for our perception of the land surface. Bull. Amer. Meteor. Soc., 87, 1381–1398. doi: 10.1175/BAMS-87-10-1381.
    [9]

    Dorigo, W. A., A. Xaver, M. Vreugdenhil, et al., 2013: Global automated quality control of in situ soil moisture data from the International Soil Moisture Network. Vadose Zone J., 12, doi: 10.2136/vzj2012.0097.
    [10]

    He, J., and K. Yang, 2011: China Meteorological Forcing Dataset. Cold and Arid Regions Science Data Center, Lanzhou, doi: 10.3972/westdc.002.2014.db.
    [11]

    Jia, B., J. Liu, and Z. Xie, 2015: Evaluation of a multi-satellite soil moisture product and the Community Land Model 4.5 simulation in China. Hydrol. Earth Syst. Sci. Discuss., 12, 5151–5186. doi: 10.5194/hessd-12-5151-2015.
    [12]

    Jia, B. H., J. G. Liu, Z. H. Xie, et al., 2018: Interannual variations and trends in remotely sensed and modeled soil moisture in China. J. Hydrometeor., 19, 831–847. doi: 10.1175/JHM-D-18-0003.1.
    [13]

    Jung, M., M. Reichstein, and A. Bondeau, 2009: Towards global empirical upscaling of FLUXNET eddy covariance observations: Validation of a model tree ensemble approach using a biosphere model. Biogeosciences, 6, 2001–2013. doi: 10.5194/bg-6-2001-2009.
    [14]

    Jung, M., M. Reichstein, P. Ciais, et al., 2010: Recent decline in the global land evapotranspiration trend due to limited moisture supply. Nature, 467, 951–954. doi: 10.1038/nature09396.
    [15]

    Jung, M., M. Reichstein, H. A. Margolis, et al., 2011: Global patterns of land–atmosphere fluxes of carbon dioxide, latent heat, and sensible heat derived from eddy covariance, satellite, and meteorological observations. J. Geophys. Res. Biogeosci., 116, G00J07. doi: 10.1029/2010JG001566.
    [16]

    Koster, R. D., M. J. Suarez, and M. Heiser, 2000: Variance and predictability of precipitation at seasonal-to-interannual timescales. J. Hydrometeor., 1, 26–46. doi: 10.1175/1525-7541(2000)001<0026:VAPOPA>2.0.CO;2.
    [17]

    Koster, R. D., P. A. Dirmeyer, Z. C. Guo, et al., 2004: Regions of strong coupling between soil moisture and precipitation. Science, 305, 1138–1140. doi: 10.1126/science.1100217.
    [18]

    Lei, H. M., and D. W. Yang, 2010: Interannual and seasonal variability in evapotranspiration and energy partitioning over an irrigated cropland in the North China Plain. Agric. For. Meteor., 150, 581–589. doi: 10.1016/j.agrformet.2010.01.022.
    [19]

    Li, J. D., F. Chen, G. Zhang, et al., 2018: Impacts of land cover and soil texture uncertainty on land model simulations over the central Tibetan Plateau. J. Adv. Model. Earth Syst., 10, 2121–2146. doi: 10.1029/2018MS001377.
    [20]

    Li, X., G. D. Cheng, S. M. Liu, et al., 2013: Heihe Watershed Allied Telemetry Experimental Research (HiWATER): Scientific objectives and experimental design. Bull. Amer. Meteor. Soc., 94, 1145–1160. doi: 10.1175/BAMS-D-12-00154.1.
    [21]

    Liu, J. G., and Z. H. Xie, 2013: Improving simulation of soil moisture in China using a multiple meteorological forcing ensemble approach. Hydrol. Earth Syst. Sci., 17, 3355–3369. doi: 10.5194/hess-17-3355-2013.
    [22]

    Liu, J. G., B. H. Jia, Z. H. Xie, et al., 2016: Ensemble simulation of land evapotranspiration in China based on a multi-forcing and multi-model approach. Adv. Atmos. Sci., 33, 673–684. doi: 10.1007/s00376-016-5213-0.
    [23]

    Liu, J. G., B. H. Jia, Z. H. Xie, et al., 2018: Improving the simulation of terrestrial water storage anomalies over China using a Bayesian model averaging ensemble approach. Atmos. Oceanic Sci. Lett., 11, 322–329. doi: 10.1080/16742834.2018.1484656.
    [24]

    Liu, S. M., Z. W. Xu, Z. L. Zhu, et al., 2013: Measurements of evapotranspiration from eddy-covariance systems and large aperture scintillometers in the Hai River Basin, China. J. Hydrol., 487, 24–38. doi: 10.1016/j.jhydrol.2013.02.025.
    [25]

    Liu, Y. Y., W. A. Dorigo, R. M. Parinussa, et al., 2012: Trend-preserving blending of passive and active microwave soil moisture retrievals. Remote Sens. Environ., 123, 280–297. doi: 10.1016/j.rse.2012.03.014.
    [26]

    Ma, N., Y. S. Zhang, Y. H. Guo, et al., 2015: Environmental and biophysical controls on the evapotranspiration over the highest alpine steppe. J. Hydrol., 529, 980–992. doi: 10.1016/j.jhydrol.2015.09.013.
    [27]

    Ma, N., G.-Y. Niu, Y. L. Xia, et al., 2017: A systematic evaluation of Noah-MP in simulating land–atmosphere energy, water, and carbon exchanges over the continental United States. J. Geophys. Res. Atmos., 122, 12245–12268. doi: 10.1002/2017JD027597.
    [28]

    Ma, N., J. Szilagyi, Y. S. Zhang, et al., 2019: Complementary-relationship-based modeling of terrestrial evapotranspiration across China during 1982–2012: Validations and spatiotemporal analyses. J. Geophys. Res. Atmos., 124, 4326–4351. doi: 10.1029/2018JD029850.
    [29]

    Ma, Y., S. Kang, L. Zhu, et al., 2008: Tibetan observation and research platform: Atmosphere–land interaction over a heterogeneous landscape. Bull. Amer. Meteor. Soc., 89, 1487–1492. doi: 10.1175/2008bams2545.1.
    [30]

    Mao, J. F., W. T. Fu, X. Y. Shi, et al., 2015: Disentangling climatic and anthropogenic controls on global terrestrial evapotranspiration trends. Environ. Res. Lett., 10, 094008. doi: 10.1088/1748-9326/10/9/094008.
    [31]

    Mitchell, K. E., D. Lohmann, P. R. Houser, et al., 2004: The multi-institution North American Land Data Assimilation System (NLDAS): Utilizing multiple GCIP products and partners in a continental distributed hydrological modeling system. J. Geophys. Res. Atmos., 109, D07S90. doi: 10.1029/2003JD003823.
    [32]

    Moradkhani, H., K.-L. Hsu, H. Gupta, et al., 2005: Uncertainty assessment of hydrologic model states and parameters: Sequential data assimilation using the particle filter. Water Resour. Res., 41, W05012. doi: 10.1029/2004WR003604.
    [33]

    Niu, G.-Y., Z.-L. Yang, K. E. Mitchell, et al., 2011: The community Noah land surface model with multiparameterization options (Noah-MP): 1. Model description and evaluation with local-scale measurements. J. Geophys. Res. Atmos., 116, D12109. doi: 10.1029/2010JD015139.
    [34]

    Oleson, K. W., G.-Y. Niu, Z.-L. Yang, et al., 2008: Improvements to the Community Land Model and their impact on the hydrological cycle. J. Geophys. Res. Biogeosci., 113, G01021. doi: 10.1029/2007JG000563.
    [35]

    Qian, T. T., A. G. Dai, K. E. Trenberth, et al., 2006: Simulation of global land surface conditions from 1948 to 2004. Part I: Forcing data and evaluations. J. Hydrometeor., 7, 953–975. doi: 10.1175/JHM540.1.
    [36]

    Robock, A., C. A. Schlosser, K. Y. Vinnikov, et al., 1998: Evaluation of the AMIP soil moisture simulations. Global Planet. Change, 19, 181–208. doi: 10.1016/S0921-8181(98)00047-2.
    [37]

    Robock, A., K. Y. Vinnikov, G. Srinivasan, et al., 2000: The Global Soil Moisture Data Bank. Bull. Amer. Meteor. Soc., 81, 1281–1300. doi: 10.1175/1520-0477(2000)081<1281:TGSMDB>2.3.CO;2.
    [38]

    Rodell, M., P. R. Houser, U. Jambor, et al., 2004: The Global Land Data Assimilation System. Bull. Amer. Meteor. Soc., 85, 381–394. doi: 10.1175/BAMS-85-3-381.
    [39]

    Sheffield, J., G. Goteti, and E. F. Wood, 2006: Development of a 50-year high-resolution global dataset of meteorological forcings for land surface modeling. J. Climate, 19, 3088–3111. doi: 10.1175/JCLI3790.1.
    [40]

    Shen, Y., P. Zhao, Y. Pan, et al., 2014: A high spatiotemporal gauge-satellite merged precipitation analysis over China. J. Geophys. Res. Atmos., 119, 3063–3075. doi: 10.1002/2013JD020686.
    [41]

    Shi, C. X., Z. H. Xie, H. Qian, et al., 2011: China land soil moisture EnKF data assimilation based on satellite remote sensing data. Sci. China Earth Sci., 54, 1430–1440. doi: 10.1007/s11430-010-4160-3.
    [42]

    Shi, C. X., L. P. Jiang, T. Zhang, et al., 2014: Status and plans of CMA land data assimilation system (CLDAS) project. Geophysical Research Abstracts, 16, EGU2014-5671.
    [43]

    Shi, X. Y., J. F. Mao, P. E. Thornton, et al., 2013: Spatiotemporal patterns of evapotranspiration in response to multiple environmental factors simulated by the Community Land Model. Environ. Res. Lett., 8, 024012. doi: 10.1088/1748-9326/8/2/024012.
    [44]

    Sun, S. B., B. Z. Chen, Q. Q. Shao, et al., 2017: Modeling evapotranspiration over China’s landmass from 1979 to 2012 using multiple land surface models: Evaluations and analyses. J. Hydrometeor., 18, 1185–1203. doi: 10.1175/JHM-D-16-0212.1.
    [45]

    Vinukollu, R. K., J. Sheffield, E. F. Wood, et al., 2012: Multimodel analysis of energy and water fluxes: Intercomparisons between operational analyses, a land surface model, and remote sensing. J. Hydrometeor., 13, 3–26. doi: 10.1175/2011JHM1372.1.
    [46]

    Wang, A. H., and X. B. Zeng, 2011: Sensitivities of terrestrial water cycle simulations to the variations of precipitation and air temperature in China. J. Geophys. Res., 116, D02107. doi: 10.1029/2010JD014659.
    [47]

    Wang, A. H., X. B. Zeng, and D. L. Guo, 2016: Estimates of global surface hydrology and heat fluxes from the Community Land Model (CLM4.5) with four atmospheric forcing datasets. J. Hydrometeor., 17, 2493–2510. doi: 10.1175/JHM-D-16-0041.1.
    [48]

    Wang, K. C., and R. E. Dickinson, 2012: A review of global terrestrial evapotranspiration: Observation, modeling, climatology, and climatic variability. Rev. Geophys., 50, RG2005. doi: 10.1029/2011RG000373.
    [49]

    Wei, J. F., P. A. Dirmeyer, and Z. C. Guo, 2008: Sensitivities of soil wetness simulation to uncertainties in precipitation and radiation. Geophys. Res. Lett., 35, L15703. doi: 10.1029/2008GL034494.
    [50]

    Xia, Y. L., K. E. Mitchell, M. B. Ek, et al., 2012: Continental-scale water and energy flux analysis and validation for North American Land Data Assimilation System project phase 2 (NLDAS-2): 2. Validation of model-simulated streamflow. J. Geophys. Res. Atmos., 117, D03110. doi: 10.1029/2011JD016051.
    [51]

    Xia, Y. L., J. Sheffield, M. B. Ek, et al., 2014: Evaluation of multi-model simulated soil moisture in NLDAS-2. J. Hydrol., 512, 107–125. doi: 10.1016/j.jhydrol.2014.02.027.
    [52]

    Xia, Y. L., T. W. Ford, Y. H. Wu, et al., 2015: Automated quality control of in situ soil moisture from the North American Soil Moisture Database using NLDAS-2 products. J. Appl. Meteor. Climatol., 54, 1267–1282. doi: 10.1175/JAMC-D-14-0275.1.
    [53]

    Xia, Y. L., B. A. Cosgrove, K. E. Mitchell, et al., 2016: Basin-scale assessment of the land surface water budget in the National Centers for Environmental Prediction operational and research NLDAS-2 systems. J. Geophys. Res. Atmos., 121, 2750–2779. doi: 10.1002/2015JD023733.
    [54]

    Yan, C. H., W. L. Zhao, Y. Wang, et al., 2017: Effects of forest evapotranspiration on soil water budget and energy flux partitioning in a subalpine valley of China. Agric. For. Meteor., 246, 207–217. doi: 10.1016/j.agrformet.2017.07.002.
    [55]

    Yang, F., H. Lu, K. Yang, et al., 2017: Evaluation of multiple forcing data sets for precipitation and shortwave radiation over major land areas of China. Hydrol. Earth Syst. Sci., 21, 5805–5821. doi: 10.5194/hess-21-5805-2017.
    [56]

    Yang, K., J. Qin, L. Zhao, et al., 2013: A multiscale soil moisture and freeze–thaw monitoring network on the third pole. Bull. Amer. Meteor. Soc., 94, 1907–1916. doi: 10.1175/BAMS-D-12-00203.1.
    [57]

    Yang, Z.-L., G.-Y. Niu, K. E. Mitchell, et al., 2011: The community Noah land surface model with multiparameterization options (Noah-MP): 2. Evaluation over global river basins. J. Geophys. Res. Atmos., 116, D12110. doi: 10.1029/2010JD015140.
    [58]

    Yu, G. R., X. F. Wen, X. M. Sun, et al., 2006: Overview of ChinaFLUX and evaluation of its eddy covariance measurement. Agric. For. Meteor., 137, 125–137. doi: 10.1016/j.agrformet.2006.02.011.
    [59]

    Zhao, P., X. D. Xu, F. Chen, et al., 2018: The third atmospheric scientific experiment for understanding the Earth–atmosphere coupled system over the Tibetan Plateau and its effects. Bull. Amer. Meteor. Soc., 99, 757–776. doi: 10.1175/BAMS-D-16-0050.1.
    [60]

    Zhao, P., Y. Q. Li, X. L. Guo, et al., 2019: The Tibetan Plateau surface–atmosphere coupling system and its weather and climate effects: The Third Tibetan Plateau Atmospheric Science Experiment. J. Meteor. Res., 33, 375–399. doi: 10.1007/s13351-019-8602-3.
    [61]

    Zheng, H., and Z.-L. Yang, 2016: Effects of soil-type datasets on regional terrestrial water cycle simulations under different climatic regimes. J. Geophys. Res. Atmos., 121, 14387–14402. doi: 10.1002/2016jd025187.
    [62]

    Zreda, M., W. J. Shuttleworth, X. Zeng, et al., 2012: COSMOS: The Cosmic-ray Soil Moisture Observing System. Hydrol. Earth Syst. Sci., 16, 4079–4099. doi: 10.5194/hess-16-4079-2012.
  • 加载中
通讯作者: 陈斌, bchen63@163.com
  • 1. 

    沈阳化工大学材料科学与工程学院 沈阳 110142

  1. 本站搜索
  2. 百度学术搜索
  3. 万方数据库搜索
  4. CNKI搜索

Improving Land Surface Hydrological Simulations in China Using CLDAS Meteorological Forcing Data

    Corresponding author: Chunxiang SHI, shicx@cma.gov.cn
  • 1. School of Mathematics and Computational Science, and Key Laboratory of Intelligent Control Technology for Wuling-Mountain Ecological Agriculture in Hunan Province, Huaihua University, Huaihua 418008, China
  • 2. National Meteorological Information Center, China Meteorological Administration, Beijing 100081, China
  • 3. Key Laboratory of Regional Climate–Environment Research for Temperate East Asia, Institute of Atmospheric Physics, Chinese Academy of Sciences, Beijing 100029, China
  • 4. Department of Geological Sciences, The John A. and Katherine G. Jackson School of Geosciences, The University of Texas at Austin, Austin, TX 78712, USA
Funds: Supported by the National Natural Science Foundation of China (91437220 and 41405083) and Fund Project for the Excellent Youth Scholars of the Education Department of Hunan Province in China (18B494)

Abstract: The accuracy of land surface hydrological simulations using an offline land surface model (LSM) depends largely on the quality of the atmospheric forcing data. In this study, Global Land Data Assimilation System (GLDAS) forcing data and the newly developed China Meteorological Administration Land Data Assimilation System (CLDAS) forcing data are used to drive the Noah LSM with multiple parameterizations (Noah-MP) and to explore how the newly developed CLDAS forcing data improve land surface hydrological simulations over mainland China. The monthly soil moisture (SM) and evapotranspiration (ET) simulations are then compared and evaluated against observations. The results show that the Noah-MP driven by the CLDAS forcing data (referred to as CLDAS_Noah-MP) significantly improves the simulations in most cases over mainland China and its eight river basins. CLDAS_Noah-MP increases the correlation coefficient (R) values from 0.451 to 0.534 for the SM simulations at a depth range of 0–10 cm in mainland China, especially in the eastern monsoon area such as the Huang–Huai–Hai Plain, the southern Yangtze River basin, and the Zhujiang River basin. Moreover, the root-mean-square error is reduced from 0.078 to 0.068 m3 m−3 for the SM simulations, and from 12.9 to 11.4 mm month−1 for the ET simulations over mainland China, especially in the southern Yangtze River basin and Zhujiang River basin. This study demonstrates that, by merging more in situ and remote sensing observations in regional atmospheric forcing data, offline LSM simulations can better simulate regional-scale land surface hydrological processes.

    • While covering about 30% of the earth’s surface, land plays a vital role in modulating the global water cycle, energy cycle, and the carbon cycle (Koster et al., 2000, 2004; Oleson et al., 2008; Wang et al., 2016). Obtaining accurate high-resolution spatiotemporal land surface hydrological information improves weather forecasts and seasonal climate predictions, and allows us to better monitor extreme events such as drought and floods (Robock et al., 1998; Koster et al., 2000, 2004; Albergel et al., 2012; Wang et al., 2016).

      Currently, three major methods are used to obtain land surface states and fluxes: in situ observations, satellite retrievals, and land surface model (LSM) simulations, where each has its own advantages and disadvantages. In situ measurements can provide high temporal resolution and accurate land surface hydrological records, but they have sparse spatial coverage and short durations. For example, in situ soil moisture (SM) observations are only available at limited stations for certain periods (Robock et al., 2000; Zreda et al., 2012; Dorigo et al., 2013; Xia et al., 2015). In the past few decades, Chinese scientists have successfully monitored SM and evapotranspiration (ET) by implementing field experiments such as ChinaFLUX (Yu et al., 2006; Lei and Yang, 2010; Ma et al., 2015; Yan et al., 2017), HiWATER (Li et al., 2013), the Haihe Experiment (Liu et al., 2013), and the Third Atmospheric Scientific Experiment over the Tibetan Plateau (Ma et al., 2008; Yang et al., 2013; Zhao et al., 2018, 2019). Satellite-retrieval products can characterize the spatial variability of land surface hydrological variables well, but they cannot provide relatively long-term climate data (Liu et al., 2012; Chen et al., 2014; Mao et al., 2015).

      LSMs, due to their process-based structure and their ability to describe continuous variations both in time and space, have been widely used to provide estimates of mutually consistent land surface hydrological variables (Robock et al., 1998; Oleson et al., 2008; Vinukollu et al., 2012; Wang and Dickinson, 2012; Chen et al., 2013; Shi et al., 2013; Xia et al., 2014, 2016; Ma et al., 2017; Jia et al., 2018).

      However, LSM simulations of land surface hydrological variables still contain large errors (Dirmeyer et al., 2006; Wang and Zeng, 2011; Wang and Dickinson, 2012; Liu and Xie, 2013; Liu et al., 2016, 2018; Wang et al., 2016). The quality of offline LSM simulations depends largely on the accuracy of the atmospheric forcing data (Wei et al., 2008; Wang and Zeng, 2011; Liu and Xie, 2013; Wang et al., 2016). Such atmospheric forcing data are generally produced by merging in situ measurements, remote sensing observations, and reanalysis data, which include air temperature, specific humidity, surface pressure, wind speed, radiation, and precipitation values (Qian et al., 2006; Sheffield et al., 2006). Wang and Zeng (2011) pointed out that simulations resulting from the global reanalysis-based meteorological forcing dataset developed by Sheffield et al. (2006) produced too low SM values in northwestern China and too high SM values in northeastern China. Instead, these simulated values could be significantly improved through using an atmospheric forcing dataset obtained by combining regional in situ temperature and precipitation measurements in China. At regional scales, remote sensing observations and in situ measurements are available for LSM use. Therefore, much effort has been made to develop improved regional atmospheric forcing datasets through merging in situ and remote sensing observations. For example, for North America, there is a regional high-resolution space/time meteorological forcing dataset called the North American Land Data Assimilation System (Mitchell et al., 2004; Xia et al., 2012); for mainland China, there are two regional high-resolution space/time meteorological forcing datasets. One was established and is regularly updated by the Institute of Tibetan Plateau Research, Chinese Academy of Sciences (ITPCAS; He and Yang, 2011), and the other is the China Meteorological Administration Land Data Assimilation System (CLDAS) atmospheric forcing dataset developed by Shi et al. (2014).

      In this study, two experiments are conducted to explore how the newly developed regional CLDAS meteorological forcing dataset improves land surface hydrological simulations over mainland China. The two experiments employ the Noah LSM with multiple parameterization options (Noah-MP; Niu et al., 2011) driven by the Global Land Data Assimilation System (GLDAS) forcing data and the newly developed regional CLDAS forcing data over mainland China.

      This paper is organized as follows. Section 2 briefly describes Noah-MP, GLDAS, CLDAS, the validation data, the experimental design, and the analysis method. In Section 3, we compare and evaluate the two simulations of SM and ET against observations, which are followed by discussions in Section 4. Finally a summary and our conclusions are given in Section 5.

    2.   Models, data, and methodology
    • The LSM used in this study is Noah-MP version 3.6 (Niu et al., 2011; Yang et al., 2011), which was developed from the original Noah LSM version 3.0 (Chen et al., 1997; Chen and Dudhia, 2001) with several major improved physics schemes. In addition, Noah-MP provides multiple parameterization options for selected processes, such as dynamic phenology, canopy stomatal resistance, runoff, and an SM factor for transpiration. In this study, we use the default options following EXP6 in Yang et al. (2011) for different parameterizations. For further details of Noah-MP, please refer to Niu et al. (2011) and Yang et al. (2011).

    • Two sets of meteorological forcing data, which were developed by different institutions, were used to drive the Noah-MP over mainland China in this study. The first is the GLDAS version 2.1 (GLDAS v2.1) global forcing dataset developed and updated jointly by NASA and NCEP (Rodell et al., 2004; Beaudoing and Rodell, 2016; https://disc.gsfc.nasa.gov/datasets/GLDAS_NOAH025_3H_V2.1/summary?keywords=GLDAS). The GLDAS v2.1 dataset covers the period of 2000–18 with a spatial resolution of 0.25° × 0.25° in longitude and latitude and a 3-h temporal resolution. It combines model, remote sensing, and observation datasets, where the source dataset contains the following: (1) NCEP Global Data Assimilation System atmospheric analysis fields; (2) spatially and temporally disaggregated Global Precipitation Climatology Project precipitation fields; and (3) observation-based downward shortwave and longwave radiation fields.

      The other forcing dataset is CLDAS version 2.0 (CLDAS v2.0), which was established and is updated by the China Meteorological Administration (CMA; Shi et al., 2014). The CLDAS v2.0 dataset contains hourly 0.0625° meteorological variables of mainland China from 2008 to 2018, which merges surface in situ measurements from more than 2400 national automatic stations and approximately 40,000 regional automatic weather stations of the CMA with the ECMWF 0.125° 3-hourly numerical analysis/predicted products to produce near-surface air temperature, specific humidity, and wind speed fields. The precipitation field is determined from four datasets, which are in situ surface precipitation observations, 0.0625° hourly East Asian Multi-Satellite Integrated Precipitation products, fusion precipitation products from Fengyun-2 (FY-2) satellite (Shen et al., 2014), and numerical analysis/predicted precipitation products from the Global Forecast System with a temporal resolution of 3 h and a spatial resolution of 0.25° × 0.25° in longitude and latitude. The shortwave radiation field was retrieved from the FY-2C/E series of geostationary meteorological satellites (Shi et al., 2011).

      Yang et al. (2017) evaluated precipitation from the ITPCAS, GLDAS, CLDAS, and the gridded analysis based on CMA gauge observations (CN05.1) over mainland China during 2008–14, and showed that CLDAS provides a more realistic precipitation approximation than the other datasets. They also assessed the shortwave radiation estimates from the ITPCAS, GLDAS, and CLDAS forcing datasets over mainland China during 2008–14, and their results showed that CLDAS slightly outperforms GLDAS. Compared to the GLDAS forcing dataset, the CLDAS forcing dataset contains more in situ and remote sensing observations of China, so it is more accurate over mainland China.

    • In this study, in situ SM measurements in China and observation-based gridded ET data were employed to evaluate and compare the two simulations.

    • We obtained in situ SM data from the CMA National Meteorological Information Center (NMIC; http://data.cma.cn). The SM values were collected every 10 days (i.e., on the 8th, 18th, and 28th of every month) from 778 agricultural meteorological stations located in farmland in China during the growing season (March to October) from 1992 to 2013. The original data were measured using a gravimetric technique at soil depths of 10, 20, 50, 70, and 100 cm. Here, we selected in situ SM observations at soil depths of 0–10 cm from 308 stations during the period 2008–13, which were used as post-processed monthly volumetric water content data via the quality-control method proposed by Jia et al. (2015).

    • Due to a lack of direct observations of ET, we used the global observation-based 0.5° × 0.5° monthly ET product to compare and assess our two ET simulations. This observation-based ET dataset was developed by Jung et al. (2009, 2010, 2011) (https://www.bgc-jena.mpg.de/geodb/projects/Data.php), covering the period of 1982–2011. It was generated by upscaling the point-wise eddy covariance flux measurements from the FLUXNET sites based on the geospatial information obtained from satellite remote sensing and surface meteorological data using a machine learning approach called the model tree ensembles (MTE) algorithm (hereafter referred to as Obs_MTE). Many previous studies used the Obs_MTE dataset to assess and compare with other ET products at regional and global scales (Jung et al., 2010; Shi et al., 2013; Liu et al., 2016; Sun et al., 2017).

    • In this study, two simulations were conducted to explore how the newly developed regional CLDAS meteorological forcing dataset improves the land surface hydrological simulations over mainland China.

      The GLDAS and CLDAS forcing datasets were used to drive Noah-MP using the default land surface parameter information, hereafter referred to as GLDAS_Noah-MP and CLDAS_Noah-MP, respectively. These two cases were compared to each other to highlight the influence of the meteorological forcing data on offline LSM simulations. The two simulations were run at a spatial resolution of 0.25° × 0.25° with a time step of 3 h in mainland China. In order to achieve an equilibrium state in Noah-MP, we first spun up the simulations for more than 60 years.

    • In this study, in situ measurements and observation-based gridded products were employed to validate and compare the two simulations over a variety of spatial and temporal scales. Four metrics, i.e., the mean bias error (MBE), correlation coefficient (R), root-mean-square error (RMSE), and unbiased RMSE (ubRMSE), were calculated from the two simulations and observations for mainland China as:

      $$\hspace{-93pt} {\rm{MBE}} = \frac{{{y_i} - {x_i}}}{n}, $$ (1)
      $$ R = \frac{{\displaystyle\mathop \sum \nolimits_{i = 1}^n \left({{y_i} - \bar y} \right)\left({{x_i} - \bar x} \right)}}{{\sqrt {\displaystyle\mathop \sum \nolimits_{i = 1}^n {{\left({{x_i} - \bar x} \right)}^2}} \sqrt {\displaystyle\mathop \sum \nolimits_{i = 1}^n {{\left({{y_i} - \bar y} \right)}^2}} }}, $$ (2)
      $$\hspace{-44pt} {\rm{RMSE = }}\sqrt {\frac{{\displaystyle\sum\nolimits_{i = 1}^n {{{\left({{y_i} - {x_i}} \right)}^2}} }}{n}}, $$ (3)
      $$\hspace{-28pt} {\rm{ubRMSE}} = \sqrt {{\rm{RMS}}{{\rm{E}}^2} - {\rm{MB}}{{\rm{E}}^2}}, $$ (4)

      where xi is a reference data point, yi is the corresponding simulated data point, n is the sample size, and $\bar x $ (${\bar y} $) is the mean value of x (y) among the n values. In this paper, the in situ observed SM and the observation-based gridded data of ET were used as references.

    3.   Results
    • SM and ET are two important land surface hydrological variables. In this study, we have compared and assessed the two simulated monthly values from Noah-MP against observations to determine to what extent CLDAS improves the simulations.

    • In this subsection, two SM simulations at a 0–10-cm depth range from 2008 to 2013 were assessed by using in situ SM measurements from 308 stations in China. First, the simulated volumetric SM value from the grid cell closest to the relevant observation station was compared with the observed value. Furthermore, the 308 SM sites were grouped into eight subregions based on the eight major river basins in China (Figs. 1c, d), and the statistics from these 308 stations over mainland China and all stations located within the eight major river basins were used to avoid partial representation of the SM site measurements in the simulated grid.

      Figure 1.  Spatial distributions of the 6-yr (2008–13) averaged volumetric SM (m3 m−3) at a depth range of 0–10 cm derived from the two simulations, the in situ site measurements, and the locations of the eight basins in mainland China: (a) GLDAS_Noah-MP, (b) CLDAS_Noah-MP, (c) the in situ site observations, and (d) the location of the eight basins in China. I: Songhuajiang River basin, II: Haihe River basin, III: Yellow River basin, IV: Huaihe River basin, V: Heihe River basin, VI: Tarim River basin, VII: Yangtze River basin, and VIII: Zhujiang River basin.

      Figure 1 displays the spatial distributions of the 6-yr (2008–13) averaged volumetric SM derived from the two simulations (Figs. 1a, b), the in situ observations (Fig. 1c) in the 0–10-cm soil layer depth range, and the locations of the eight basins in mainland China (Fig. 1d). The observed SM values (Fig. 1c) display strong regional variations and there are evident northeast–northwest and southeast–northwest wet-to-dry patterns. The two SM simulations all generally captured the spatial patterns and dry–wet centers of the SM in most cases, although GLDAS_Noah-MP overestimated the SM over most of the eastern monsoon areas of China; however, CLDAS_Noah-MP agrees more closely with the observed spatial patterns, especially in the southern and northeastern regions of China.

      To quantitatively examine the performance of the two SM simulations driven by the GLDAS and CLDAS forcing datasets, we compared the simulated and observed monthly SM time series averaged over the eight major basins of mainland China. Figure 2 displays the time series of the observed and modeled monthly volumetric SM at a depth range of 0–10 cm averaged over the eight major basins of mainland China from 2008 to 2013. The two SM simulations match the observed SM values well in most cases, which indicates that the two simulations generally captured the seasonal cycle and temporal evolution of the observed SM reasonably well, at least in terms of the mean SM over the eight major basins of China. However, GLDAS_Noah-MP overestimated the amplitude, and simulated much higher SM values than the observed values over all the six major basins in the eastern monsoon area. In contrast, the simulated values from CLDAS_Noah-MP generally agreed more closely with the observed values over most of the river basins. The magnitudes and patterns of SM shown in Fig. 2 are generally consistent with the amount of precipitation in these areas (figure omitted). We note that there is more precipitation from GLDAS than CLDAS over all the six major basins in the eastern monsoon area of China. This indicates that precipitation plays a vital role in the SM simulations, especially in the surface soil layer (0–10 cm). However, the two simulations simulated lower SM values in the Heihe River basin and the Tarim River basin, which are located in the northwestern arid area of China. This was probably caused by using the sparse site measurements from agricultural meteorological stations located in farmland. The uncertainties from the observations and scale mismatch between the simulations and observations may be the major sources of bias in the SM simulations over the two basins, which will be further discussed in Section 4.

      Figure 2.  Time series of the monthly volumetric SM at a depth range of 0–10 cm from in situ observations, GLDAS_Noah-MP, and CLDAS_Noah-MP during the period of 2008–13 in the eight major basins of China.

      To quantify the performance of the SM simulations, we compared their R and RMSE values against the site observations in mainland China. Figure 3 shows the differences in R (RD) and RMSE (RMSED) of the two simulated SM results against the in situ measurements. The simulated SM values from CLDAS_Noah-MP were better over most of China compared with those from GLDAS_Noah-MP, where the former had higher R values for 218 of the 306 stations, as shown in Fig. 3a (71.2%, RD > 0). Furthermore, CLDAS_Noah-MP showed smaller RMSE values for 193 of 306 stations, as seen in Fig. 3b (63.1%, RMSED < 0). The most pronounced increase of R (RD > 0.2) was seen over most of the North China Plain area (i.e., the Huang-Huai-Hai Plain).

      Figure 3.  Spatial distributions of the difference in R and RMSE (m3 m−3) between the modeled SM and the in situ site observations. (a) R difference: CLDAS_Noah-MP minus GLDAS_Noah-MP and (b) RMSE difference: CLDAS_Noah-MP minus GLDAS_Noah-MP.

      To further quantify the performances of the SM simulations, and avoid the partial representativeness of the site measurements, we used the statistics from 308 stations over mainland China and the stations located in the eight major river basins (rather than emphasized individual stations). Since we only used sparse SM measurements to validate the SM simulations, there may be large inherent systematic biases between the site measurements and LSM gridded simulations. Thus, we present the ubRMSE and MBE values of the simulated SMs against the site measurements in this subsection. Table 1 summarizes the mean R, MBE, RMSE, and ubRMSE values of the SM simulations against the site measurements in mainland China and its eight major basins. The two simulations of SM showed positive MBE values over most of the basins except for the Heihe River basin and the Tarim River basin, where the CLDAS_Noah-MP displayed smaller mean errors and was closer to the observations. The results from CLDAS_Noah-MP showed a significant increase of R from 0.451 to 0.534 (~18.4%), a significant reduction of MBE from 0.039 to 0.003 m3 m–3 (~92.3%), a significant reduction of RMSE from 0.078 to 0.068 m3 m–3 (~12.8%), and a slightly reduction of ubRMSE from 0.039 to 0.037 m3 m–3 (~5.1%) for all 308 stations of mainland China. This indicates that the systematic bias is an important contribution to the total bias in terms of the differences between GLDAS_Noah-MP and CLDAS_Noah-MP. The percentage of stations with a correlation that was statistically significant at the p = 0.05 level increased from 79.6% for GLDAS_Noah-MP to 86.2% for CLDAS_Noah-MP, which further indicates that CLDAS_Noah-MP displayed better performance compared to GLDAS_Noah-MP. For the eight major basins in China, CLDAS_Noah-MP showed significantly improved simulated SM values in the eastern monsoon area such as the Yellow River basin (R increased by ~26.2%, RMSE reduced by ~20.7%, and ubRMSE reduced by ~10.5%), the Huaihe River basin (R increased by ~15.6%, RMSE reduced by ~11.6%, and ubRMSE reduced by ~2.8%), and the Haihe River basin (R increased by ~12.7% and RMSE decreased by ~18.3%) in the North China Plain area, the southern Yangtze River basin (R increased by ~27.9%, RMSE reduced by ~18.2%, and ubRMSE reduced by ~10.5%), and the Zhujiang River basin (R increased by ~24.2%, RMSE reduced by ~25.2%, and ubRMSE reduced by ~5.6%).

      GLDAS_Noah-MPCLDAS_Noah-MP
      RRMSEMBEubRMSERRMSEMBEubRMSE
      Songhuajiang River
       basin (38 stations)
      0.387 (78.9%)0.0810.0530.0380.381 (66.7%, –1.6%)0.070 (–13.6%)0.0140.034 (–10.5%)
      Haihe River
       basin (28 stations)
      0.473 (85.2%)0.0600.0390.0380.533 (89.3%, +12.7%)0.049 (–18.3%)0.002 0.039 (+2.6%)
      Heihe River
       basin (10 stations)
      0.212 (30.0%)0.072–0.0330.0390.253 (40.0%, +19.0%)0.079 (+9.7%)–0.0520.039 (0)
      Tarim River
       basin (8 stations)
      0.256 (42.9%)0.076–0.0140.0350.257 (42.9%, +0.4%)0.095 (+25%)–0.0490.037 (+5.7%)
      Yellow River
       basin (84 stations)
      0.493 (81.5%)0.0820.0500.0380.622 (95.1%, +26.2%)0.065 (–20.7%)0.0130.034 (–10.5%)
      Huaihe River
       basin (25 stations)
      0.552 (96%)0.0690.0300.0360.638 (96%, +15.6%)0.061 (–11.6%)–0.0080.035 (–2.8%)
      Yangtze River
       basin (43 stations)
      0.501 (87.8%)0.0880.0450.0380.641 (95.3%, +27.9%)0.072 (–18.2%)0.0120.034 (–10.5%)
      Zhujiang River
       basin (7 stations)
      0.596 (100%)0.1150.0780.0360.740 (100%, +24.2%)0.087 (–25.2%)0.0320.034 (–5.6%)
      China (308 stations)0.451 (79.6%)0.0780.0390.0390.534 (86.2%, +18.4%)0.068 (–12.8%)0.0030.037 (–5.1%)
      Note: For the R values, the first number in the parentheses is the percentage of stations with a correlation that is statistically significant at the p = 0.05 level. The second number in the parentheses is the percentage increase relative to the paired experiment. For the RMSEs and ubRMSEs, the numbers in the parentheses are the percentages of reduction relative to the paired experiment.

      Table 1.  Mean R, MBE (m3 m−3), RMSE (m3 m−3), and ubRMSE (m3 m−3) values between the simulated and measured SM at a 0–10-cm depth range from 2008 to 2013 over the eight basins and mainland China

    • ET is another important land surface hydrological variable. In this subsection, the two ET simulations from 2008 to 2011 are compared and assessed by using the observation-based Obs_MTE data in China.

      Figure 4 displays the spatial distributions of the 4-yr (2008–11) averaged ET values over mainland China derived from the two simulations and Obs_MTE. The Obs_MTE data show strong regional variations and there is an evident southeast–northwest high-to-low gradient. The two simulations generally captured the spatial pattern of ET in most cases, although GLDAS_Noah-MP underestimated ET especially in the southeast areas of China; however, CLDAS_Noah-MP agreed more closely with the spatial patterns of Obs_MTE.

      Figure 4.  Spatial distributions for the 4-yr (2008–11) averaged ET (mm yr−1) derived from the two simulations and the observation-based Obs_MTE product in mainland China: (a) GLDAS_Noah-MP, (b) CLDAS_Noah-MP, and (c) Obs_MTE.

      We further compared the simulated and observation-based monthly ET time series in the eight river basins from 2008 to 2011. Both ET simulations captured the seasonal cycle and temporal evolution of the observation-based ET very well over most of the river basins (Fig. 5), and CLDAS_Noah-MP agreed more closely with the time variation of Obs_MTE over the Yangtze River basin and the Zhujiang River basin of southern China. However, there were errors in the peak values for the two ET simulations, especially in the Heihe River basin, the Tarim River basin, and the Zhujiang River basin. The biases over the Zhujiang River basin may be due to complex land surfaces, such as dense vegetation, abundant river systems, complex terrain, and coastal lines. There are large errors in the Tarim River basin and the Heihe River basin, where the Obs_MTE values are higher than those from the two simulations. The uncertainties from observations may be an important source of bias in the ET simulations over these two basins, which will be discussed further in Section 4.

      Figure 5.  Time series of the monthly ET from the observation-based Obs_MTE product, GLDAS_Noah-MP, and CLDAS_Noah-MP during the period of 2008–11 in the eight major basins of China.

      To further quantify the performance of the ET simulations, we compared the R and RMSE values and their differences (RD and RMSED). For the R value, GLDAS_Noah-MP and CLDAS_Noah-MP displayed a similar pattern, and have relatively small differences between them (Fig. 6). There are very high R values over most of the eastern monsoon areas. CLDAS_Noah-MP simulated higher R values over most of the southern and western regions in China, whereas GLDAS_Noah-MP showed slightly higher R values over most of the North China Plain area and northeastern China. For the RMSE values, GLDAS_Noah-MP and CLDAS_Noah-MP had a similar pattern (Fig. 7). Compared to GLDAS_Noah-MP, CLDAS_Noah-MP resulted in smaller RMSE values over most of China, especially in the southern regions (Fig. 7c), and thus improved the ET simulation.

      Figure 6.  Spatial distributions of the difference in R between the modeled ET and the observation-based Obs_MTE product. (a) R of GLDAS_Noah-MP, (b) R of CLDAS_Noah-MP, and (c) R difference: CLDAS_Noah-MP minus GLDAS_Noah-MP.

      Figure 7.  Spatial distributions of the difference in RMSE (mm month−1) between the modeled ET against the observation-based Obs_MTE product. (a) RMSE of GLDAS_Noah-MP, (b) RMSE of CLDAS_Noah-MP, and (c) RMSE difference: CLDAS_Noah-MP minus GLDAS_Noah-MP.

      Table 2 further shows region-averaged R and RMSE values of ET in mainland China and its eight major basins. For the R values, CLDAS_Noah-MP and GLDAS_Noah-MP had high values and small differences. GLDAS_Noah-MP displayed higher R values in northern China such as the Songhuajiang River basin, the Haihe River basin, the Heihe River basin, and the Tarim River basin, whereas CLDAS_Noah-MP simulated higher R values in southern China such as the Yangtze River and the Zhujiang River basins. For the RMSE, CLDAS_Noah-MP showed smaller values from 12.9 to 11.4 mm month–1 for all of mainland China, especially in southern China such as the Yangtze River basin (from 14.7 to 11.7 mm month–1; a decrease of ~20.4%) and the Zhujiang River basin (from 21.5 to 16.8 mm month–1; a decrease of ~21.9%). Over most of the basins in northern China, CLDAS_Noah-MP and GLDAS_Noah-MP displayed small differences, and CLDAS_Noah-MP did not perform better than GLDAS_Noah-MP. The uncertainties of the atmospheric forcing data may not be the primary source of the errors in the ET simulations of GLDAS_Noah-MP and CLDAS_Noah-MP, which will be discussed further in Section 4.

      GLDAS_Noah-MPCLDAS_Noah-MP
      RRMSERRMSE
      Songhuajiang River basin0.97510.60.962 (–1.3%)11.0 (+3.8%)
      Haihe River basin0.97310.90.962 (–1.1%)11.5 (+5.5%)
      Heihe River basin0.7338.30.718 (–2.0%)7.7 (–7.2%)
      Tarim River basin0.77210.00.706 (–8.5%)10.7 (+7.0%)
      Yellow River basin0.9608.90.954 (–0.6%)9.2 (+9.4%)
      Huaihe River basin0.93515.60.914 (–2.2%)16.9 (+8.3%)
      Yangtze River basin0.95914.70.961 (+2.1%)11.7 (–20.4%)
      Zhujiang River basin0.91821.50.923 (+0.5%)16.8 (–21.9%)
      China0.90812.90.901 (–0.7%)11.4 (–11.6%)
      Note: The numbers in the parentheses are the percentage of variations relative to the paired experiment.

      Table 2.  Mean R and RMSE (mm month−1) values between the simulated and the observation-based Obs_MTE ET product from 2008 to 2011 over the eight river basins and mainland China

    4.   Discussion
    • In this study, two numerical experiments were conducted to explore how the newly developed regional CLDAS meteorological forcing data improved the land surface hydrological simulations over mainland China. Compared to GLDAS_Noah-MP, CLDAS_Noah-MP generally improved the simulated SM and ET over mainland China and its eight major river basins. However, CLDAS_Noah-MP still displayed some errors; for example, both the SM and ET simulations generally performed poorly in the Tarim River basin and the Heihe River basin. The possible causes and sources of these biases are discussed in this section.

      The main sources of biases are the measurement biases and simulation biases, where the latter mainly arise from the model parameterizations, land surface parameters, and forcing data. Thus, some limitations in this study should be noted. First, in order to compare and evaluate the two simulations of SM and ET, we used both the in situ measurements of SM and the observation-based ET products as the observational data. In this study, we assumed these observations were error-free, whereas in reality they do contain some errors; for example, the Obs_MTE ET product contains some system errors compared to the flux tower measurements (Jung et al., 2010; Sun et al., 2017; Ma et al., 2019). Over mainland China, Sun et al. (2017) found that while the Obs_MTE ET product had the lowest RMSE among the Obs_MTE product, MODIS product, and multiple LSM simulations when compared to nine representative flux tower observations, it still contained errors and overestimated the ET values. For example, Obs_MTE used the default values from some locations of the Heihe and Tarim basins, which failed to represent most of the desert areas where the ET values are very low. This may be an important reason for the poor performance of the ET simulations for the two basins for both GLDAS_Noah-MP and CLDAS_Noah-MP. Moreover, Liu et al. (2016) found a similar conclusion using four different LSMs. However, reducing the uncertainty of these measurements is beyond the scope of this current study.

      Second, uncertainties can be caused by a scale mismatch between simulations and observations. For example, the in situ SM data are point measurement values, while the model results are gridded values. This mismatch may be another reason for the differences between the observations and simulations, especially in the Heihe River basin and the Tarim basin, which are located in the northwestern arid area of China. The sparse site measurements from agricultural meteorological stations are located in farmland, where most of the agricultural lands in this arid area are affected by human activities such as irrigation. Therefore, the observed SM values are greater than the gridded SM values, resulting in large differences between the site measurements and the gridded simulations. This may be an important reason for the poor performance of the GLDAS_Noah-MP and CLDAS_Noah-MP SM simulations over the two basins.

      Third, in addition to the forcing datasets, the model parameterizations and parameter values are the other two important sources of biases in the offline LSM simulations (Moradkhani et al., 2005; Dirmeyer et al., 2006). Currently, almost all LSMs employ one-dimension (1-D) water balance equations to describe vertical land surface hydrological processes, but they neglect any hydrological processes operating in the horizontal direction. Thus, the land surface hydrological processes are not well described in current LSMs (Bi et al., 2016). Additionally, hydrological simulations using offline LSMs are sensitive to land surface parameters such as land cover and soil texture data (Zheng and Yang, 2016; Li et al., 2018). Zheng and Yang (2016) pointed out that sandier soil types used in the LSM simulations produce drier SM values, higher surface runoff, and lower ET. Li et al. (2018) found that the soil parameters used in the LSM simulations dominated the modeled water and energy fluxes over the central Tibetan Plateau. These may be the additional reasons for the differences between the observations and modeled results.

    5.   Summary and conclusions
    • LSMs are widely used to produce long-term and high-resolution land surface hydrological variables. However, the accuracy of offline LSM simulations depends on the quality of the meteorological forcing data. In this study, two offline LSM simulations using Noah-MP were conducted over mainland China to determine how CLDAS_Noah-MP driven by the newly developed CLDAS forcing data improved the land surface hydrological simulations over mainland China and its eight major river basins. Our main conclusions are stated as follows.

      (1) For the SM simulation at a 0–10 cm depth range, GLDAS_Noah-MP and CLDAS_Noah-MP generally captured the temporal variation and spatial distribution of the observed SM reasonably well. However, GLDAS_Noah-MP overestimated the SM over most of the eastern monsoon area in China, and CLDAS_Noah-MP agreed more closely with the observed SM. Compared to GLDAS_Noah-MP, CLDAS_Noah-MP significantly improved the simulated SM in mainland China [the R value increased from 0.451 to 0.534 (~18.4%), the MBE decreased from 0.039 to 0.003 (~92.3%), the RMSE reduced from 0.078 to 0.068 (~12.8%), and the ubRMSE decreased from 0.039 to 0.037 (~5.1%)], especially in the eastern monsoon areas such as the Yellow River basin, Haihe River basin, and Huaihe River basin in the North China Plain area, and southern Yangtze River basin as well as Zhujiang River basin.

      (2) The two ET simulations generally captured the temporal variation and spatial distribution of the observation-based Obs_MTE ET product reasonably well, but GLDAS_Noah-MP underestimated the ET in the southeastern regions of China. The R values showed similar spatial patterns, and had very high values and relatively small differences. For the RMSE values, CLDAS_Noah-MP significantly reduced the RMSE value from 12.9 to 11.4 mm month–1 in mainland China and especially over the southern Yangtze River basin and the Zhujiang River basin. In general, compared to GLDAS_Noah-MP, CLDAS_Noah-MP improved the ET simulation over most of mainland China, especially in southern China.

      In summary, CLDAS_Noah-MP driven by the newly developed regional CLDAS atmospheric forcing data significantly improves land surface hydrological simulations in mainland China. This indicates that, by merging more in situ and remote sensing observations in regional atmospheric forcing data, offline LSMs can better simulate regional-scale land surface hydrological processes.

     
    • Acknowledgments. The authors thank NOAA and CMA/NMIC for providing GLDAS and CLDAS meteorological forcing datasets, the CMA/NMIC for providing in situ SM observations, and the Max Planck Institute for providing ET product.

Reference (62)

Catalog

    /

    DownLoad:  Full-Size Img  PowerPoint
    Return
    Return