Assessment of Noah land surface model with various
runoff parameterizations over a Tibetan river
Donghai Zheng1 , Rogier Van Der Velde1 , Zhongbo Su1 , Jun Wen2 , and Xin Wang2
1
Faculty of Geo-Information Science and Earth Observation, University of Twente, Enschede, Netherlands,2Key Laboratory of Land Surface Process and Climate Change in Cold and Arid Regions, Northwest Institute of Eco-Environment and Resources, Chinese Academy of Sciences, Lanzhou, China
Abstract
Runoff parameterizations currently adopted by the (i) Noah-MP model, (ii) Community Land Model (CLM), and (iii) CLM with variable infiltration capacity hydrology (CLM-VIC) are incorporated into the structure of Noah land surface model, and the impact of these parameterizations on the runoff simulations is investigated for a Tibetan river. Four numerical experiments are conducted with the default Noah and three aforementioned runoff parameterizations. Each experiment is forced with the same set of atmospheric forcing, vegetation, and soil parameters. In addition, the Community Earth System Model database provides the maximum surface saturated area parameter for the Noah-MP and CLM parameterizations. A single-year recurrent spin-up is adopted for the initialization of each model run to achieve equilibrium states. Comparison with discharge measurements shows that each runoff parameterization produces significant differences in the separation of total runoff into surface and subsurface components and that the soil water storage-based parameterizations (Noah and CLM-VIC) outperform the groundwater table-based parameterizations (Noah-MP and CLM) for the seasonally frozen and high-altitude Tibetan river. A parameter sensitivity experiment illustrates that this underperformance of the groundwater table-based parameterizations cannot be resolved through calibration. Further analyses demonstrate that the simulations of other surface water and energy budget components are insensitive to the selected runoff parameterizations, due to the strong control of the atmosphere on simulated land surfacefluxes induced by the diurnal dependence of the roughness length for heat transfer and the large water retention capacity of the highly organic top soils over the plateau.1. Introduction
Large uncertainties exist among quantification of the climate change impact on the runoff production and water availability in high-altitude Himalayan and adjacent Tibetan river basins [Immerzeel et al., 2010; Immerzeel et al., 2013; Lutz et al., 2014], due to an imperfect understanding of present-day hydrology. The dependence of large populations downstream on the freshwater supply from the High Asia’s river upstream, e.g., Yangtze, Yellow, Mekong, Brahmaputra, and Ganges Rivers, underlines the need for better quanti fica-tions of the available water resources under various climate change scenarios. This task requires a thorough understanding of the present-day hydrology.
Previous studies [e.g., Immerzeel et al., 2010; Andermann et al., 2012; Lutz et al., 2014] utilized hydrological models to quantify the runoff regime across the Tibetan Plateau, which generally neglected the soil freeze-thaw process that may be particularly important for the Tibetan Plateau, as substantial areas are subject to seasonally frozen ground and/or underlain with permafrost region [Guo and Wang, 2013]. Model structures of land surface models (LSMs) do include the freeze-thaw process and have also been often applied to Tibetan study areas [Xue et al., 2013; Zhang et al., 2013]. However, advances in the understanding of land sur-face process over the Tibetan Plateau [e.g., Yang et al., 2005; Yang et al., 2009; Chen et al., 2011; Zheng et al., 2014] have not been incorporated in the deployed LSMs. For instance, the diurnally varying roughness length for heat transfer (z0h) has been proven to be essential for simulating turbulent heatfluxes and surface tem-perature [Yang et al., 2009; Chen et al., 2011; Zheng et al., 2014], and the vertical soil heterogeneity associated with root system and organic content is imperative for calculating soil heat and water transfer [Yang et al., 2005; van der Velde et al., 2009; Chen et al., 2013; Zheng et al., 2015c]. The omission of these processes forms crucial sources of uncertainty in LSM simulations in terms of the coupled surface water and energy budgets. Recently, Zheng et al. [2016] incorporated the abovefindings into the Noah LSM and illustrated that a com-plete treatment of both warm and cold season hydrometeorological processes is imperative for a correct
Journal of Geophysical Research: Atmospheres
RESEARCH ARTICLE
10.1002/2016JD025572
Key Points:
• Runoff parameterizations adopted by Noah, Noah-MP, CLM, and CLM-VIC are evaluated for the seasonally frozen Yellow River source region • Soil water storage-based runoff
schemes perform much better than groundwater table-based
parameterizations for the Tibetan river • Surface water and energy budget
simulations are insensitive to the selected runoff scheme
Correspondence to:
D. Zheng, d.zheng@utwente.nl
Citation:
Zheng, D., R. Van Der Velde, Z. Su, J. Wen, and X. Wang (2017), Assessment of Noah land surface model with various runoff parameterizations over a Tibetan river, J. Geophys. Res. Atmos., 122, 1488–1504, doi:10.1002/2016JD025572. Received 24 JUN 2016
Accepted 14 JAN 2017
Accepted article online 19 JAN 2017 Published online 2 FEB 2017
©2017. The Authors.
This is an open access article under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs License, which permits use and distri-bution in any medium, provided the original work is properly cited, the use is non-commercial and no modifications or adaptations are made.
simulation of the runoff regime over a seasonally frozen Tibetan river. However, the Noah LSM ignores the interactions between soil water and groundwater as well as the topography-driven lateral surface and sub-surfaceflow that impact the runoff. These deficiencies have been resolved in the newly developed Noah-MP LSM (Noah LSM with multiparameterization options) [Niu et al., 2011] through implementation of a simple groundwater module [Niu et al., 2007] and a simple TOPMODEL-based runoff parameterization [Niu et al., 2005] that implicitly accounts for the lateralflow. Gulden et al. [2007] showed the robustness of the explicit representation of groundwater in simulating the runoff dynamics.
Although the TOPMODEL-based parameterization explicitly represents the topographic effects on soil moist-ure heterogeneity and runoff production, Li et al. [2011] have shown that similar TOPMODEL-based formula-tion adopted by the Community Land Model (CLM) [Oleson et al., 2013] cannot correctly simulate the measured runoff dynamics in a baseflow-dominated mountainous catchment. They also demonstrated that the runoff generation scheme implemented in the variable infiltration capacity (VIC) model [Liang et al., 1994] is superior and generally valid under various climatic conditions. As a result, the VIC runoff parameterization has been included as an alternative hydrological option in the latest version of CLM (CLM-VIC) [Li et al., 2011; Oleson et al., 2013]. This diversity in the runoff modeling approaches adopted by LSMs demonstrate already by itself the need for further investigation especially for underexplored regions such as the Tibetan Plateau. Model comparison studies have been historically undertaken by the land surface and climate modeling community to explore differences in LSM performance, such as the Project for Intercomparison of Land Surface Parameterization Schemes (PILPS) [Henderson-Sellers et al., 1993, 1995], Global Soil Wetness Project (GSWP) [Dirmeyer et al., 1999, 2006], and Global Land-Atmosphere Coupling Experiment (GLACE) [Koster et al., 2004, 2011]. Since large differences exist among different LSMs because of different model physics, structure, and parameter choices, they are not able to meaningfully attribute intermodel differ-ences in predictive ability to individual model components [Clark et al., 2011, 2015]. Alternative approaches have been thus developed to systematically evaluate alternative parameterization options for various processes within a unified modeling framework, such as the Joint UK Land Environment Simulator (JULES) [Best et al., 2011], Noah-MP [Niu et al., 2011], and the Structure for Unifying Multiple Modeling Alternatives (SUMMA) [Clark et al., 2015].
We investigate in this study the performance of the aforementioned runoff parameterizations within a single modeling framework in simulating runoff as well as other surface water and energy budget com-ponents across a Tibetan catchment. The investigation is conducted in the source region of the Yellow River (SRYR) for the period 2001–2010. The version of the Noah LSM described in Zheng et al. [2016] is adopted and modified to include the three selected runoff parameterizations which are currently adopted by (i) Noah-MP [Niu et al., 2011], (ii) CLM [Oleson et al., 2013], and (iii) CLM-VIC [Li et al., 2011].
The structure of this paper is as follows: Section 2 provides a description of the Noah model physics and the selected runoff parameterizations. Section 3 introduces the in situ measurements, as well as the experimental design and model setup. Section 4 presents the evaluation of selected runoff parameterizations in simulating runoff and other surface water and energy budget components, as well as the sensitivity test. Further, con-clusions with a summary of thefindings in this study are drawn in section 5.
2. Model Physics
The version of the Noah LSM described in Zheng et al. [2016] is adopted, which includes a diurnally vary-ing roughness length for heat transfer (z0h), an asymptotic function for root water uptake, vertical
hetero-geneous soil thermal and hydraulic properties, and a frozen ground parameterization, all of which have been modified to better represent the SRYR environment. Model physics of the augmented version are briefly introduced in Appendix A, and readers are referred to existing literatures [Zheng et al., 2014; Zheng et al., 2015b, 2015c, 2016] for more information. The default as well as the three selected runoff parameterizations is described below.
2.1. Noah Runoff Parameterization
Surface runoff (Rs, in m s1) in the Noah LSM consists of infiltration excess runoff from the unfrozen part
of the model grid and direct runoff from the impermeable frozen part (fimp, -) [Schaake et al., 1996; Koren
Rs¼ fimpPxþ 1 fimp
P2x
Pxþ Wd½1 exp Kð dtΔtÞ
=Δt (1)
whereΔt is the time step (s), Wdis the soil water deficit within the soil column (m), Pxis the precipitation
fall-ing on the ground (m), Kdtis an empirical parameter taken as 3.0 d1, and the impermeable frozen fraction is
as in Zheng et al. [2016] assuming a gamma probability distribution for the soil ice content (Wice, m) within
root zone: fimp¼ ev Xα i¼1 vαi Γ α i þ 1ð Þ (2) v¼ αWcr Wice (3) Wice¼ X nroot i¼1 θice; i Δzi (4)
whereΔz is the soil depth (m), nroot is the number of soil layers within the root zone (-), θiceis the soil ice
content (m3m3), Wcris the critical ice content above which the frozen ground is impermeable taken as
0.15 m,α is a shape parameter of the gamma probability distribution (-) taken as 3, and v is the upper limit of an integral function representing the spatial variability of frozen depth [Koren et al., 1999].
Subsurface runoff (Rb, in m s1) is formulated as gravitational free drainage from the model bottom:
Rb¼ slope K θ liq;4 (5)
whereθliq,4is the liquid soil water content of the bottom (fourth) layer (m3m3), K(θliq,4) is the hydraulic
con-ductivity (m s1) that is estimated by equation (A4), and slope is a slope index between 0 and 1 depending on the slope of the model grid.
2.2. Noah-MP Runoff Parameterization
The Noah-MP [Niu et al., 2011; Yang et al., 2011; Cai et al., 2014; Zheng et al., 2015a] is an augmented version of the Noah LSM and includes improved representation of biophysical and hydrological processes as well as implementing a multiparameterization framework for selected processes. An unconfined aquifer is added below the 2 m bottom of the Noah soil column to allow exchange of water between the unsaturated zone and the groundwater reservoir. Changes in water stored in the aquifer (Wa, m) is determined as the residual
of the recharge, Q (m s1), minus discharge (i.e., subsurfaceflow): dWa
dt ¼ Q Rb (6)
where Q is parameterized following Darcy’s law:
Q¼ Kbotz∇ fðmicψbot zbotÞ
z∇ zbot
(7) where z∇is the water table depth (m); fmicis the fraction of micropore content in the bottom soil layer (-)
taken as 0.2;ψbotand Kbotare the matric potential (m) and hydraulic conductivity (m s1) of the bottom soil
layer, respectively; and zbot(1.5 m in Noah LSM) is the midpoint of the bottom soil layer. The detailed
calcula-tion of z∇andψbotis given in Niu et al. [2007].
The simplified TOPMODEL-based runoff scheme [Niu et al., 2005] is adopted to simulate surface runoff and subsurfaceflow, both of which are parameterized as exponential function of z∇:
Rs¼ fimpPxþ 1 fimp
fsatPx
=Δt (8)
fsat¼ fmaxexp 0:5 f z∇ z0bot
(9) Rb¼ 1 f 0 imp
Rb; max exp λ f z∇ z0bot
(10) where fsatis the fraction of surface saturated area (-), fmaxis the maximum surface saturated area (-), f is a
decay factor taken as 6.0 m1, z0botis the soil column depth taken as 2 m, Rb,maxis the maximum subsurface
Niu et al. [2011]. In this study, the fraction of impermeable frozen area in the bottom (fourth) soil layer f0impis also estimated by the gamma distribution (equations (2) and (3)) calculated with soil ice content from the bottom layer.
2.3. CLM Runoff Parameterization
The CLM [Oleson et al., 2013] is implemented as the land component of the Community Earth System Model (CESM), which employs, similar to Noah-MP, a TOPMODEL-based runoff scheme for simulating surface and subsurface runoff:
fsat¼ fmax exp 0:5 fð over z∇Þ (11)
Rb¼ 1 f
0
imp
Rb; max exp fð drai z∇Þ (12)
where foveris the surface runoff decay factor taken as 0.5 m1, fdraiis the subsurface runoff decay factor taken
as 2.5 m1, and Rb,maxis taken as 5.5 × 106m s1within CLM. The calculation of Rsis identical to equation (8),
and the groundwater parameterization implemented in the Noah-MP (equations (6) and (7)) is also adopted here to calculate the z∇. It can be found that the runoff parameterizations adopted by the Noah-MP and CLM share similar formulations while implementing different values for model parameters, e.g., a surface runoff decay factor of 6.0 versus 0.5 m1and maximum subsurface runoff of 5.0 × 103versus 5.5 × 106m s1, respectively.
2.4. CLM-VIC Runoff Parameterization
Li et al. [2011] have shown that the simple TOPMODEL-based runoff scheme is not able to capture the runoff dynamics of a baseflow-dominated mountainous catchment, which can be resolved by implementing the VIC runoff generation scheme [Liang et al., 1994]. Consequently, the VIC runoff parameterization has been implemented as an alternative option within the latest version of CLM (CLM-VIC) [Oleson et al., 2013]. Although the soil column is typically divided into three layers with variable soil depths in the original VIC, the soil depths arefixed in the CLM-VIC parameterization to be consistent with the CLM soil model [Li et al., 2011]. In this study, the total soil depth is set as 2 m to be consistent with the Noah LSM. In line with the CLM-VIC implementation, the four soil layers of the Noah LSM are aggregated into three layers, with thick-nesses of 0.1, 0.9, and 1.0 m, for modeling surface and subsurface runoff. The Noah soil layers for simulation of the soil heatflow are kept intact.
In CLM-VIC, surface runoff computation follows the VIC formulation [Oleson et al., 2013]:
Rs¼ px 1 fimp Wm;t Wt =Δt; i0þ px≥ im px 1 fimp Wm;t Wt Wm;t 1i0þ px im 1þb " # ( ) =Δt; i0þ px< im 8 > > < > > : (13)
where Wm,tis the maximum water storage of top two soil layers (m), Wtis the total soil water storage in the
top two layers at the beginning of a time step (m), i0is the current soil water holding capacity (m), imis the
maximum soil water holding capacity (m), and b is the shape parameter controlling the spatial distribution of i.
Subsurface runoff generation is based on the ARNO model concept [Todini, 1996] adopted for VIC [Cherkauer and Lettenmaier, 1999; Oleson et al., 2013]:
Rb¼ 1 f0imp DsDsmax Ws Wa;b =Δt; 0 ≤ Wa;b≤ Ws 1 f0imp DsDsmax Ws Wa;bþ Dsmax DsDsmax Ws Wa;b Ws 1 Ws 2 " # =Δt; Wa;b> Ws 8 > > > > < > > > > : (14) Wa;b¼ Wl;b Ww;b Wm;b Ww;b (15)
where for the bottom soil Wm,brepresents the maximum water storage (m), Ww,bis the water storage at
wilt-ing point (m), Wl,bis the soil liquid water storage at the beginning of a time step (m), Dsmaxis the maximum
subsurfaceflow (m s1), Dsis a fraction of Dsmax(-), and Wsis a fraction of the potential water storage
according to Cuo et al. [2013] and Zhang et al. [2013], and Dsmax is set equal to the saturated hydraulic conductivity of the bottom soil layer.
3. Study Area and Model Setup
3.1. Study Area and In Situ MeasurementsThe source region of the Yellow River (SRYR, Figure 1) is located in the northeastern part of the Tibetan Plateau, within a transition region from continuous and discontinuous permafrost to seasonally frozen ground [Jin et al., 2009]. The SRYR covers 122,000 km2above the Tangnag discharge station, which contri-butes for more than 35% to the streamflow of the Yellow River basin [Zhou and Huang, 2012].
The weather category falls under the class of wet and cold climate according to the updated Köppen-Geiger climate classification, with 75–90% of the mean annual precipitation falling in the monsoon season between June and September [Hu et al., 2011]. Loamy soils and alpine grassland are the dominant land cover, and the fraction of glacier and lake coverage is about 1%.
Monthly measurements of streamflow from the Tangnag discharge station are utilized for the validation of simulated runoff, which are available for the period of 2002–2009. Besides, in situ heat flux and profile soil moisture and temperature measurements are used to investigate the impact of various runoff parameteriza-tions on the simulated surface water and energy budgets, which are available for the southeastern Maqu sta-tion (33.88°N, 102.15°E, elevasta-tion 3430 m; Figure 1) and the northwestern Maduo stasta-tion (35.03°N, 96.38°E, elevation 4450 m). These in situ measurements are available from November 2009 to December 2010, and detailed information about the measurements and data processing is given in Zheng et al. [2016].
3.2. Experimental Design and Model Setup
Four numerical experiments are designed to assess the performance of the four runoff parameterizations, described in section 2, within the framework of the Noah LSM. The Noah LSM is run with its default runoff parameterizationfirst (section 2.1). Later on, three runs are performed by replacing the default with the para-meterization currently implemented in Noah-MP (section 2.2), CLM (section 2.3), and CLM-VIC (section 2.4). All the experiments are set up with atmospheric forcing data from the Institute of Tibetan Plateau Research, Chinese Academy of Sciences (hereafter ITPCAS) [Chen et al., 2011], soil and vegetation parameters from the China Soil Database [Shangguan et al., 2013], and Weather Research and Forecasting (WRF) data set (http:// www2.mmm.ucar.edu/wrf/users/download/get_sources_wps_geog.html). In addition, the maximum surface saturated area parameter (fmax) for Noah-MP and CLM surface runoff calculation is derived from the CESM
database (https://svn-ccsm-inputdata.cgd.ucar.edu). The model spatial and temporal resolutions are 10 km and 30 min, respectively, and the 0.1° and 3 h ITPCAS atmospheric forcing are interpolated to the model domain and time step. A single-year recurrent spin-up is conducted from 1 July 2001 to 30 June 2002 for each experiment to achieve the initial equilibrium model states, for which 30 model years are needed. The equili-brium is achieved when the following conditions are reached: |Varn+1 Varn|< 0.001|Varn|, whereby Var represents each of the model states (e.g., soil moisture/temperature and groundwater table) and n is the spin-up cycle. A single continuous 8.5 year simulation is then carried out from 1 July 2002 to 31 December 2010 for each experiment.
As in Zheng et al. [2016], the monthly streamflow (m3) measurements (section 3.1) are converted for valida-tion to the monthly area-averaged total runoff (R, mm) by dividing by the SRYR total area (km2), and the sur-face runoff (Rs) and subsurface runoff (Rb) simulated with the Noah LSM are accumulated to produce the
monthly area-averaged R. Additionally, the heatflux and profile soil temperature and moisture simulations are extracted from the grid elements where the Maduo and Maqu stations are located for the comparison with those in situ measurements.
4. Results
4.1. Areal Averaged Runoff
Figure 2 shows the monthly accumulated and area-averaged measured/simulated total runoff (R) for the period July 2002 to December 2009 produced by the Noah numerical experiments with the four different
Figure 2. Comparisons of measured and simulated (left) monthly accumulated and (right) multiyear monthly averaged total runoff (R) produced by the numerical experiments for the period July 2002 to December 2009.
runoff parameterizations (section 3.2). The monthly R for both measurements
and simulations averaged for the
7.5 year period is shown as well. Table 1 provides the respective error statistics, i.e., mean error (ME), coefficient of determination (R2), Nash-Sutcliffe efficiency (NSE), and root-mean-square error (RMSE). These results demonstrate that Noah, with its default runoff parameterization, captures the measured monthly R dynamics reason-ably well, as is also supported by the R2 and NSE values larger than 0.85. The plot and the error statistics also indicate that the runoff parameterization implemented currently by the Noah-MP leads to performance in simulating R inferior to the others. Particularly, the simulated R largely underestimates the June and July measurements (>5 mm on the average), while a significant overestimation is found for December (>7 mm on the average). Also, the performance of the CLM runoff parameterization in simulating the hydrograph is worse than the Noah’s default according to the error statistics in terms of R2, RMSE, and NSE. The CLM runoff parameterization overestimates the measured R during the transition season from spring to summer (i.e., May and June) and winter (December to February), while it largely underesti-mates the measurements in October. Replacement of the CLM runoff parameterization with the one imple-mented by the CLM-VIC yields a considerable improvement in the R simulations which also exceed the default Noah performance. The error statistics improve with respect to the default Noah simulations by about 1.5, 56.5, 9.7, and 3.5% for R2, ME, RMSE, and NSE, respectively.
In support of further analyses, Figure 3 presents the monthly averaged measured and simulated total runoff (R), surface runoff (Rs), and base flow (Rb) for each numerical experiment. The mean monthly total
Table 1. Coefficient of Determination (R2), Mean Error (ME), Root-Mean-Square Error (RMSE), and Nash-Sutcliffe Efficiency (NSE) Computed Between the Measured and Simulated Total Runoff Produced by the Numerical Experiments for the Period July 2002 to December 2009
Experiments R2 ME (mm) RMSE (mm) NSE
Noah 0.882 1.08 3.62 0.86 Noah-MP 0.704 0.78 5.45 0.69 CLM 0.761 0.64 4.90 0.75 CLM-VIC 0.895 0.47 3.27 0.89 EXPS1a 0.664 -0.02 6.20 0.60 EXPS1b 0.543 -0.25 6.95 0.50 EXPS1c 0.466 -0.35 7.35 0.44 EXPS2a 0.350 -1.09 8.20 0.31 EXPS2b 0.755 0.41 4.90 0.75 EXPS2c 0.772 0.73 6.24 0.60
Figure 3. Monthly averaged precipitation (P), measured runoff (R_obs), simulated total runoff (R_sim), surface runoff (Rs_sim), and baseflow (Rb_sim) produced using
precipitation (P) derived from the ITPCAS atmospheric forcing (section 3.2) is also shown in thefigure. The Noah model run with its default runoff parameterization produces a strongly Rb-dominated total runoff
(Figure 3a), yet the Rscontribution remains significant. Specifically, both R and Rsare governed by the strong precipitation seasonality that generally peaks in July. A time lag is noted between the Rband precipitation,
which sustains the relatively high runoff production during the postmonsoon season (i.e., October) as the monthly precipitation decreases. The time lag can be associated with the water transport through the soil col-umn. The presence of organic matter in the Tibetan top soils increases the soil porosity and retention capacity that is responsible for the delayed movement of soil water from the surface to the deeper layers enlarging the time lag.
The Noah-MP runoff parameterization also produces a strongly Rb-dominated total runoff (Figure 3b), and
similar time lags in response to precipitation are observed for both Rsand Rb. Apart from the organic matter effect, the explanation for the time lags can be also related to the large value for the runoff decay factor f (=6) that facilitates infiltration and a slow release from the groundwater reservoir delaying the Rs and Rb
production by the simple TOPMODEL-based runoff formulations (equations (8)–(10)). As such, the Noah-MP underestimates the June and July measurements. Both Noah-Noah-MP and CLM runoff parameterizations (Figures 3b and 3c) overestimate the measured R during winters, which may indicate that the impermeable frozen factor (f0imp) does not sufficiently represent the ice effect on groundwater dynamics and base flow production (equation (10) or (12)). In contrast, the CLM runoff parameterization adopts a much smaller value for surface runoff decay factor fover(=0.5, equation (11)), leading to a Rs-dominated total runoff during the
monsoon season. Since less water is stored and available for the Rbproduction during the postmonsoon sea-son, a large underestimation of the measured R is obtained for October.
The CLM-VIC runoff parameterization yields comparable runoff dynamics as the default Noah and measure-ments (Figures 3d and 3a), with a strongly Rb-dominated total runoff production during monsoons and winters, whereby the Rscontribution is of the same order of magnitude as the Rbcomponent during the
thawing period from March to June. In summary, the runoff dynamics measured in the seasonally frozen and high-altitude SRYR can be much better reproduced by the Noah LSM when both surface and subsurface runoff contributions are parameterized as functions of the soil water storage (including both liquid water and ice storages, e.g., Noah or CLM-VIC) instead of the groundwater table as is done by the Noah-MP and CLM parameterizations. This implies that the freeze-thaw effect on the soil water storage plays an impor-tant role on the runoff production in the Tibetan rivers, something that may not have been fully investi-gated in many previous studies [e.g., Immerzeel et al., 2010; Andermann et al., 2012; Lutz et al., 2014]. Moreover, the one-dimensional (1-D) unconfined aquifer module adopted by the Noah-MP and CLM is an oversimplification of the complex groundwater system residing on the Tibetan Plateau and its interac-tion with surface water and unsaturated zone [Ge et al., 2008, 2011].
4.2. Spatial Variation of Runoff
Figure 4 shows maps of annual mean R produced for the SRYR over the period July 2002 to December 2010 using the four numerical experiments with different runoff parameterizations. All the four runoff parameter-izations produce comparable spatial R distributions, which generally follows the spatial precipitation distri-bution as shown in Zheng et al. [2016]. Areas with high precipitation form the runoff source regions, such as the central high mountain area and the southeastern part of the SRYR. The reason for the similar spatial R distribution produced by each numerical experiment will be further elaborated in section 4.3.
Figure 5 presents the annual mean Rb as a fraction of R produced by the four numerical experiments. Both the Noah and Noah-MP runoff parameterizations produce a strongly Rb-dominated total runoff
across the entire SRYR, with the Rbfractions generally larger than 0.7 (Figures 5a and 5b). Within these high precipitation and runoff production regions as shown in Figure 4, the CLM runoff parameterization produces a less Rb-dominated total runoff with the Rbfraction larger than 0.5, and the Rscomponent even
dominates the runoff production in certain regions (Figure 5c). The Rbcontrol on the total runoff is
stron-ger with the CLM-VIC parameterization (Figure 5d), and the Rbfraction is on average larger than 0.5 across
the SRYR, which is smaller than the fractions found for the Noah and Noah-MP parameterizations. In sum-mary, the Rbcomponent generally dominates the total R production simulated with all runoff
4.3. Surface Water and Energy Budgets
Tables 2 and 3 list the RMSE and R2computed between the monthly mean measured and simulated latent heatflux (LE), soil temperature (Ts), and moisture (θ) profiles produced by the four numerical experiments for the period November 2009 to December 2010 for Maqu and Maduo stations, respectively. All the numer-ical experiments generally produce reasonable simulations of LE,θ, and Tsaccording to the RMSE and R2
sta-tistics; e.g., the RMSE for the LE simulations are about 10 W m2and 0.023–0.046 m3m3and 1.43–2.55°C for theθ and Tssimulations, respectively. Overall, the four numerical experiments produce comparable LE, Ts,
andθ results supported by the small differences among the RMSE and R2values computed for each variable. For instance, the maximum differences in terms of RMSE are about 0.24 W m2, 0.08°C, and 0.008 m3m3for the LE, Ts, andθ, respectively. This indicates that different runoff parameterizations have minor impacts on
the simulated LE, Ts, andθ.
The impacts of the runoff parameterization on the simulated annual surface water and energy budgets of the SRYR for the period July 2002 to December 2010 are illustrated by Figure 6, in which the individual water and energy budget components are expressed as the ratios with respect to mean annual precipitation sums for the water budget and the mean annual net radiation (Rn) for the energy budget, both of which are derived
from the ITPCAS atmospheric forcing. In general, the annual mean evapotranspiration (ET) is the dominant component of the surface water budget with the fraction larger than 65%, and only small differences (<2.5%) are noted among the simulations. The runoff parameterizations mainly affect the separation of the total runoff into Rsand Rb, and the Rbcomponent generally dominates the production of R. On the other hand, the four numerical experiments produce very similar surface energy budgets with small differences in the partitioning of Rninto latent (LE), sensible (H), and ground heatfluxes (G0), and the LE and H components
are of equal magnitude. This confirms the little impact that the runoff parameterization has on the simula-tions of surface heatfluxes.
Further, Figure 7 presents the simulated monthly ET averaged for the SRYR over the period July 2002 to December 2010, and Figure 8 shows the maps of mean annual ET across the SRYR obtained by the four
Figure 4. Maps of annual mean total runoff across the SRYR produced by (a) Noah, (b) Noah-MP, (c) CLM, and (d) CLM-VIC numerical experiments for the period July 2002 to December 2010.
numerical experiments. Figures 7 and 8 confirm the small differences among ET simulations produced by different runoff parameterizations on both temporal and spatial scales. Spatially, the ET production largely follows the topography (see Figure 1) with highest annual ET sums of more than 500 mm in the southeastern part of the SRYR, which has also been previously reported in Zhou and Huang [2012]. Obviously, the ET production over time is in line with the seasons with a peak of 70 mm in July and a dip of less than 10 mm in January.
In summary, the partitioning of the annual mean P into ET and R as well as the Rninto LE, H, and G0is
insen-sitive to the selected runoff parameterizations. The reason for this can be explained as follows. The surface heatflux and soil temperature simulations over the Tibetan Plateau are mostly controlled by the atmospheric forcing (e.g., radiation, wind speed, and pressure) and the parameterization of the diurnally varying rough-ness length for heat transfer [Yang et al., 2009; Chen et al., 2011; Zeng et al., 2012; Zheng et al., 2015b]. The presence of organic matter in the Tibetan top soils increases the soil porosity and retention capacity that con-trols the water transport and redistribution in the soil column [Yang et al., 2005; Chen et al., 2012; Zheng et al., 2015c]. As such, the runoff parameterization, drawing its water from the deep soil layers in a Rb-dominated
runoff regime, has minor impact on the simulated states andfluxes representing the exchanges at the
Table 2. Root-Mean-Square Error (RMSE) and Coefficient of Determination (R2) Computed Between the Measurements Collected From the Maqu Station and the Simulated Latent Heat Flux (LE), Soil Temperature (Ts), and Liquid Soil Water (θliq) for Depths of 5 cm and 25 cm Produced by the Numerical Experiments for the Period November 2009 to December 2010
Experiments
RMSE R2
LE (W m2) Ts5(°C) Ts25(°C) θliq5(m3m3) θliq25(m3m3) LE Ts5 Ts25 θliq5 θliq25
Noah 9.97 1.83 1.44 0.036 0.038 0.915 0.947 0.957 0.893 0.811
Noah-MP 9.95 1.83 1.43 0.036 0.038 0.916 0.947 0.958 0.894 0.816
CLM 10.19 1.88 1.46 0.041 0.046 0.918 0.947 0.956 0.908 0.812
CLM-VIC 9.99 1.85 1.44 0.037 0.040 0.917 0.947 0.957 0.900 0.816
Figure 5. Maps of annual mean baseflow as fraction of the total runoff for the SRYR produced by (a) Noah, (b) Noah-MP, (c) CLM, and (d) CLM-VIC numerical experi-ments for the period July 2002 to December 2010.
land-atmosphere interface. Regardless of the similarities in simulating surface water and energy budgets, the choice for a specific runoff parameterization is of crucial importance for the separation of R into Rsand Rb.
4.4. Sensitivity Test
The previous sections demonstrate that the soil water storage-based runoff parameterizations (Noah and CLM-VIC) perform better than the groundwater table-based parameterizations (Noah-MP and CLM). It should be noted that the Noah LSM has been modified to better represent the SRYR environment [Zheng et al., 2016] and the parameter values adopted for the CLM-VIC parameterization are from studies focused on the SRYR as well [Cuo et al., 2013; Zhang et al., 2013], while default parameter values for global applications are adopted for both Noah-MP and CLM parameterizations. These default values may, however, not be suited for local applications. Additional experiments are carried out to further investigate the impact of parameter choices on the runoff simulation with the groundwater table-based parameterizations (Noah-MP and CLM). Only the CLM runoff parameterization is studied here, but as both Noah-MP and CLM runoff parameterizations implement a similar TOPMODEL-based formulation (section 2), thefindings should be transferable to the Noah-MP parameterization as well.
In the CLM runoff parameterization, foverand fdraiin equations (11) and (12) have a significant impact on the
groundwater table depth and runoff production; fovercontrols the amount of infiltration, and fdraidefines the
shape of the recession curve. For instance, Figure 3 shows that the CLM parameterization with default fover
value of 0.5 produces a larger Rsand less infiltration than the others. Hence, three additional experiments
are carried out to demonstrate the effect of foveron infiltration by increasing its value to 1.0 (EXPS1a), 1.5 (EXPS1b), and 2.0 (EXPS1c). Each experiment adopts the default fdraivalue (i.e., 2.5) and is also initialized using
the single-year recurrent spin-up to achieve the equilibrium model states as described in section 3.2. Figure 9 shows the monthly area-averaged Rs, Rb, and R resulting from EXPS1a–S1c, whereby the error
statis-tics computed between the measured and simulated R are included in Table 1. The default CLM and Noah simulations are also added in Figure 9 for comparison purposes. The plots confirm that the Rsdecreases with
the increase of fovervalue (Figure 9a), leading to more infiltration into soil column and drainage from soil bot-tom, resulting in shallower groundwater table depth. Hence, the Rbis larger (Figure 9c) and simulated total R
Table 3. Root-Mean-Square Error (RMSE) and Coefficient of Determination (R2) Computed Between the Measurements Collected from the Maduo Station and the Simulated Soil Temperature (Ts) and Liquid Soil Water (θliq) for Depths of
5 cm and 25 cm Produced by the Numerical Experiments for the Period November 2009 to December 2010
Experiments
RMSE R2
Ts5(°C) Ts25(°C) θliq5(m3m3) θliq25(m3m3) Ts5 Ts25 θliq5 θliq25
Noah 2.41 2.52 0.038 0.025 0.985 0.985 0.911 0.915
Noah-MP 2.41 2.47 0.038 0.023 0.985 0.987 0.895 0.918
CLM 2.43 2.55 0.042 0.026 0.985 0.984 0.896 0.916
CLM-VIC 2.42 2.48 0.038 0.024 0.985 0.986 0.896 0.917
Figure 6. Comparisons of the ratios of different surface (a) water budget components to mean annual precipitation and (b) energy budget components to averaged net radiation produced by the numerical experiments for the period July 2002 to December 2010.
Figure 7. Time series of simulated (left) monthly accumulated and (right) multiyear monthly averaged evapotranspiration (ET) produced by the numerical experi-ments for the period July 2002 to December 2010.
Figure 8. Maps of annual mean evapotranspiration (ET) across the SRYR produced by (a) Noah, (b) Noah-MP, (c) CLM, and (d) CLM-VIC numerical experiments for the period July 2002 to December 2010.
production is more evenly distributed across the season. In comparison with the measurements, the simula-tions performed with higher fovervalues yield larger overestimations of the measured R during winters and larger underestimations during the warm season (July–October). As such, alteration of the fovervalue cannot
improve performance of the R simulations for the SRYR with a groundwater table-based runoff parameteriza-tion (Figure 9e and Table 1).
Three other experiments are carried out to investigate the impact of the fdraiby varying its values to, e.g., 1.5 (EXPS2a), 3.5 (EXPS2b), and 4.5 (EXPS2c), whereby the fovervalue is specified as 1.0 because it (EXPS1a)
pro-duces a Rscomparable to the default Noah parameterization. Again, the resulting simulations are shown in
Figure 9, and the corresponding error statistics are added to Table 1. The decrease of fdraivalue (EXPS2a)
increases Rb production during winters with respect to EXPS1a (Figure 9d) and lowers the Rsyear-round
(Figure 9b). On the other hand, the increase of fdraivalue (EXPS2b–S2c) enhances the seasonality of the Rb production, e.g., a decrease during winters versus increase during summers, and increases the Rsyear-round.
Overall, the simulation performed with a larger fdraivalue captures better the low winter runoff production but exaggerates the summer production due to combined effect of a large Rb and Rs leading to even
Figure 9. Comparisons of the simulated monthly averaged (a, b) surface runoff, (c, d) baseflow, and (e, f) total runoff pro-duced by the sensitivity experiments for the period July 2002 to December 2009.
worse error statistics as compared to both Noah-MP and CLM parameterization with default parameter values. In conclusion, the deficiency in R production simulations produced with the groundwater table driven parameterization (e.g., CLM) cannot be mitigated through calibration of foverand fdrai.
5. Conclusion
We investigate in this paper the impact of various runoff parameterizations for simulating runoff in the source region of the Yellow River (SRYR) on the Tibetan Plateau for the period 2001–2010. The Noah land surface model (LSM) described in Zheng et al. [2016] is adopted and modified to incorporate the selected runoff para-meterizations that are currently implemented by (i) Noah-MP, (ii) CLM, and (iii) CLM-VIC. Accordingly, four numerical experiments are conducted, namely, a control model run with the default runoff parameterization (hereafter Noah) and three runs each with one of the selected parameterizations (hereafter Noah-MP, CLM, and CLM-VIC). Each experiment adopts soil and vegetation parameters from the WRF data set and China Soil Database and atmospheric forcing from the ITPCAS data set. A single-year recurrent spin-up is used for the model initialization to achieve the equilibrium states. In addition, the CESM database provides the maximum surface saturated area parameter for the Noah-MP and CLM runoff parameterizations.
Monthly measured streamflow between July 2002 and December 2009 is utilized to assess the performance of each runoff parameterization. The soil water storage-based runoff parameterizations, i.e., Noah and CLM-VIC, are able to capture the measured total runoff (R) dynamics reasonably well supported by R2and NSE sta-tistics higher than 0.85, whereas the groundwater table-based runoff parameterizations, i.e., Noah-MP and CLM, perform worse with R2and NSE values lower than 0.77. All the runoff parameterizations except the CLM generally produce a baseflow (Rb)-dominated runoff regime, whereby the surface runoff (Rs)
contribu-tion remains important for achieving a match with the measurements. Although each of the four experiments produces similar spatial R patterns that generally follow the applied precipitationfields (P), differences are noted in the partitioning of R into Rsand Rb. The Rbcomponent generally dominates the total R production
within the high precipitation and runoff production regions. For other regions, a Rb-dominated R pattern is
also found for all the experiments except for the one with the CLM parameterization.
In situ latent heatflux (LE) and profile soil temperature (Ts) and moisture (θ) measurements are further used to
investigate the impact of various runoff parameterizations on the simulations of surface water and energy budgets. All the numerical experiments generally produce reasonable and comparable simulations of LE,θ, and Ts, which indicates that the different runoff parameterizations have a minor impact on the heatflux, soil
moisture, and temperature profile simulations. Further, it is found that the partitioning of the annual mean P into evapotranspiration (ET) and R as well as the net radiation (Rn) into LE, sensible (H), and ground heatfluxes
(G0) is insensitive to the selected runoff parameterizations. This can be related to the fact that the heat and
mass exchange at the land-atmosphere interface is controlled by the atmospheric forcing due to strong diur-nal dependence of the roughness length for heat transfer (z0h) and the large water retention capacity
facili-tated by the highly organic top soils in the SRYR. Moreover, a sensitivity experiment illustrates that the deficiency of groundwater table-based parameterizations (e.g., CLM) in simulating runoff cannot be resolved through parameter calibration.
This study demonstrates that the runoff parameterizations adopted by various LSMs produce significant dif-ferences in the separation of R into Rsand Rband that the soil water storage-based runoff parameterizations
perform much better than the groundwater table-based parameterizations in simulating the R produced in the seasonally frozen and high-altitude SRYR. Consideration of a three-dimensional (3-D) groundwaterflow system [e.g., Ge et al., 2008; Ge et al., 2011] as well as a more physical-based runoff formulation [e.g., Shi et al., 2013; Niu et al., 2014] may further improve the performance of LSMs in quantifying the total R produced in the headwaters residing on the Tibetan Plateau. As such, a better quantification of the climate change impact on the regional hydrology and available water resources from the High Asia’s rivers can be achieved.
Appendix A: Noah Model Physics
The thermal diffusion equation is utilized to describe the soil heat transport, within which a source/sink term is added to represent the phase transitions of soil water [Koren et al., 1999]:
Csðθ; θiceÞ∂T ∂t¼ ∂ ∂z κhðθ; θiceÞ∂T ∂z þ ρiceLf∂θ ice ∂t (A1)
where Csis the thermal heat capacity (J m3K1),θiceis the soil ice content (m3m3),θ is the total soil water content (m3m3), t is the time (s), T is the soil temperature (K),κhis the thermal heat conductivity (W m1 K1),ρiceis the density of ice (kg m3), z is the soil depth (m), and Lfis the latent heat of fusion (J kg1). Both Csandκhare dependent on the constituents of soil matrix (e.g.,θiceandθ), and Zheng et al. [2015b] reported recently on the inclusion of the organic matter effect.
The solution to equation (A1) is achieved using the fully implicit Crank-Nicholson scheme. The temperature at the bottom boundary (at a depth of 8 m) is generally taken as the annual mean near-surface air temperature, whereas the top boundary is confined by the ground surface temperature (Tsfc) estimated by
Tsfc¼ Taþ
S↓ S↑þ εL↓ H LE G0
4εσT3a
1
4Ta (A2)
where Tais the air temperature (K); S↓and S↑are the downward and upward shortwave radiations (W m2),
respectively; L↓is the downward longwave radiation (W m2);ε is the surface emissivity (-); σ is the Stefan-Boltzmann constant (taken as 5.67 × 108W m2K4); H is the sensible heatflux (W m2); LE is the latent heat flux (W m2); and G0is the ground surface heatflux (W m2). The scheme of diurnal thermal roughness
length for heat transfer (z0h) is implemented as in Zheng et al. [2015b] to simulate better the turbulent heat
fluxes (i.e., H and LE) and surface and soil temperatures.
The transport of unfrozen or liquid soil water is simulated with the diffusivity form of Richards’ equation with the assumption that liquid waterflow under frozen condition is analogous to that in unfrozen condition [Koren et al., 1999]: ∂θliq ∂t ¼∂z∂ Dθliq; θice ∂θliq ∂z þ∂K θliq ∂z þ S θð Þ (A3)
whereθliqis the liquid soil water content (m3m3), K is the hydraulic conductivity (m s1), D is the water
dif-fusivity (m2s1), and S is the term of sinks and sources (i.e., evapotranspiration and infiltration, m s1). The parameterization for calculating K and D updated by Zheng et al. [2016] is used to allow more transport of water within the frozen soil column:
K¼ 1 fð f rzÞKsðθ=θsÞ2bþ3 (A4)
D¼ 1 fð f rzÞDsðθ=θsÞbþ2 (A5)
ff rz¼ exp a 1 θ½ ð ice=θsÞ exp að Þ (A6)
where the subscript“s” represents the respective quantity under saturation, ffrzis the impermeable fraction, b is an empirical parameter (-) related to the pore size distribution, and a is an adjustable scale-dependent para-meter taken as 4.0.
The soil type specific hydraulic parameters (e.g., θs, Ks, and b) are obtained from the class pedotransfer func-tion (PTF), which Zheng et al. [2015c] modified to consider the organic matter effect. Specifically, the effect of organic matter on the soil water retention curve (i.e.,θsand b) is considered via the additivity hypothesis,
which estimates the hydraulic parameters as a weighted combination of the mineral and organic fractions [Zeiliguer et al., 2000; Lawrence and Slater, 2008] as:
θs¼ 1 f t;socθs;minþ ft;socθs;soc (A7)
ψs¼ 1 ft;soc
ψs;minþ ft;socψs;soc (A8)
b¼ 1 f t;socbminþ ft;socbsoc (A9)
where the ft,socis the volumetric soil carbon or organic fraction, and the hydraulic parameters of the mineral
Cosby et al. [1984], while the hydraulic properties of pure organic matter (i.e.,θs,soc,ψs,soc, and bsoc) depend on
the state of decomposition as described in Letts et al. [2000].
Further, a function is implemented that decays the Ksexponentially with soil depth [Beven, 1982], whereby
the Ksat the reference depth is estimated by the Kozeny-Carman equation based on the derivedθsand b
[Ahuja et al., 1984; Saxton and Rawls, 2006] as
Ks;z¼ Ks;ref zdð rÞ (A10)
Ks;r¼ C θ s;r θ33;r31=b (A11)
where Ks,ris the reference saturated hydraulic conductivity (m s1) at the reference depth dr(m), while Ks,zis
the estimated saturated hydraulic conductivity (m s1) at the soil depth z (m), f is the exponential profile decay factor (m1),θ33is the water content (m3m3) at33 kPa matric potential, and C is an empirically derived constant and herein taken as 1930 mm h1from Saxton and Rawls [2006].
References
Ahuja, L. R., J. W. Naney, R. E. Green, and D. R. Nielsen (1984), Macroporosity to characterize spatial variability of hydraulic conductivity and effects of land management, Soil Sci. Soc. Am. J., 48(4), 699–702, doi:10.2136/sssaj1984.03615995004800040001x.
Andermann, C., L. Longuevergne, S. Bonnet, A. Crave, P. Davy, and R. Gloaguen (2012), Impact of transient groundwater storage on the discharge of Himalayan rivers, Nat. Geosci., 5(2), 127–132, doi:10.1038/ngeo1356.
Best, M., M. Pryor, D. Clark, G. Rooney, R. Essery, C. Ménard, J. Edwards, M. Hendry, A. Porson, and N. Gedney (2011), The Joint UK Land Environment Simulator (JULES), model description—Part 1: Energy and water fluxes, Geosci. Model Dev., 4(3), 677–699, doi:10.5194/gmd-4-677-2011.
Beven, K. (1982), On subsurface stormflow: An analysis of response times, Hydrol. Sci. J., 27(4), 505–521, doi:10.1080/02626668209491129. Cai, X., Z.-L. Yang, C. H. David, G.-Y. Niu, and M. Rodell (2014), Hydrological evaluation of the Noah-MP land surface model for the Mississippi
River Basin, J. Geophys. Res. Atmos., 119, 23–38, doi:10.1002/2013JD020792.
Chen, Y., K. Yang, J. He, J. Qin, J. Shi, J. Du, and Q. He (2011), Improving land surface temperature modeling for dry land of China, J. Geophys. Res., 116, D20104, doi:10.1029/2011JD015921.
Chen, Y., K. Yang, W. Tang, J. Qin, and L. Zhao (2012), Parameterizing soil organic carbon’s impacts on soil porosity and thermal parameters for Eastern Tibet grasslands, Sci. China Earth Sci., 55(6), 1001–1011, doi:10.1007/s11430-012-4433-0.
Chen, Y., K. Yang, J. Qin, L. Zhao, W. Tang, and M. Han (2013), Evaluation of AMSR-E retrievals and GLDAS simulations against observations of a soil moisture network on the central Tibetan Plateau, J. Geophys. Res. Atmos., 118, 4466–4475, doi:10.1002/jgrd.50301.
Cherkauer, K. A., and D. P. Lettenmaier (1999), Hydrologic effects of frozen soils in the upper Mississippi River basin, J. Geophys. Res., 104, 19,599–19,610, doi:10.1029/1999JD900337.
Clark, M. P., D. Kavetski, and F. Fenicia (2011), Pursuing the method of multiple working hypotheses for hydrological modeling, Water Resour. Res., 47, W09301, doi:10.1029/2010WR009827.
Clark, M. P., et al. (2015), A unified approach for process-based hydrologic modeling: 1 Modeling concept, Water Resour. Res., 51, 2498–2514, doi:10.1002/2015WR017198.
Cosby, B. J., G. M. Hornberger, R. B. Clapp, and T. R. Ginn (1984), A statistical exploration of the relationships of soil moisture characteristics to the physical properties of soils, Water Resour. Res., 20, 682–690, doi:10.1029/WR020i006p00682.
Cuo, L., Y. Zhang, Y. Gao, Z. Hao, and L. Cairang (2013), The impacts of climate change and land cover/use transition on the hydrology in the upper Yellow River Basin, China, J. Hydrol., 502, 37–52, doi:10.1016/j.jhydrol.2013.08.003.
Dirmeyer, P. A., A. Dolman, and N. Sato (1999), The pilot phase of the Global Soil Wetness Project, Bull. Am. Meteorol. Soc., 80(5), 851–878, doi:10.1175/1520-0477(1999)080<0851:TPPOTG>2.0.CO;2.
Dirmeyer, P. A., X. Gao, M. Zhao, Z. Guo, T. Oki, and N. Hanasaki (2006), GSWP-2: Multimodel analysis and implications for our perception of the land surface, Bull. Am. Meteorol. Soc., 87(10), 1381–1397, doi:10.1175/bams-87-10-1381.
Ge, S., Q. B. Wu, N. Lu, G. L. Jiang, and L. Ball (2008), Groundwater in the Tibet Plateau, western China, Geophys. Res. Lett., 35, L18403, doi:10.1029/2008GL034809.
Ge, S., J. McKenzie, C. Voss, and Q. Wu (2011), Exchange of groundwater and surface-water mediated by permafrost response to seasonal and long term air temperature variation, Geophys. Res. Lett., 38, L14402, doi:10.1029/2011GL047911.
Gulden, L. E., E. Rosero, Z.-L. Yang, M. Rodell, C. S. Jackson, G.-Y. Niu, P. J. F. Yeh, and J. Famiglietti (2007), Improving land-surface model hydrology: Is an explicit aquifer model better than a deeper soil profile?, Geophys. Res. Lett., 34, L09402, doi:10.1029/2007GL029804. Guo, D., and H. Wang (2013), Simulation of permafrost and seasonally frozen ground conditions on the Tibetan Plateau, 1981–2010,
J. Geophys. Res. Atmos., 118, 5216–5230, doi:10.1002/jgrd.50457.
Henderson-Sellers, A., Z.-L. Yang, and R. E. Dickinson (1993), The project for intercomparison of land-surface parameterization schemes, Bull. Am. Meteorol. Soc., 74(7), 1335–1349, doi:10.1175/1520-0477(1993)074<1335:TPFIOL>2.0.CO;2.
Henderson-Sellers, A., A. J. Pitman, P. K. Love, P. Irannejad, and T. H. Chen (1995), The Project for Intercomparison of Land Surface Parameterization Schemes (PILPS): Phases 2 and 3, Bull. Am. Meteorol. Soc., 76(4), 489–503, doi:10.1175/1520-0477(1995)076<0489: TPFIOL>2.0.CO;2.
Hu, Y., S. Maskey, S. Uhlenbrook, and H. Zhao (2011), Streamflow trends and climate linkages in the source region of the Yellow River, China, Hydrol. Processes, 25(22), 3399–3411, doi:10.1002/hyp.8069.
Immerzeel, W. W., L. P. H. van Beek, and M. F. P. Bierkens (2010), Climate change will affect the Asian water towers, Science, 328(5984), 1382–1385, doi:10.1126/science.1183188.
Immerzeel, W. W., F. Pellicciotti, and M. F. P. Bierkens (2013), Rising riverflows throughout the twenty-first century in two Himalayan gla-cierized watersheds, Nat. Geosci., 6(9), 742–745, doi:10.1038/ngeo1896.
Jin, H., R. He, G. Cheng, Q. Wu, S. Wang, L. Lü, and X. Chang (2009), Changes in frozen ground in the Source Area of the Yellow River on the Qinghai–Tibet Plateau, China, and their eco-environmental impacts, Environ. Res. Lett., 4(4), 045,206, doi:10.1088/1748-9326/4/4/045206.
Acknowledgments
This study was supported by funding from the Netherlands Organization for Scientific Research (project ALW-GO/14-29), the National Natural Science Foundation of China (grants 41405079 and 41530529), and the Key Research Program of the Chinese Academy of Sciences (grant KZZD-EW-13). The glo-bal WRF geographic input data sets were obtained from WRF Users Page. The meteorological inputs were devel-oped by Data Assimilation and Modeling Center for Tibetan Multi-spheres in Institute of Tibetan Plateau Research. The China Soil Database was developed by Land-Atmosphere Interaction Research Group in Beijing Normal University. The measurements used in this study were provided by Jun Wen (jwen@lzb.ac.cn) in CAREERI/CAS. The authors would like to thank three anonymous reviewers for their con-structive comments.
Koren, V., J. Schaake, K. Mitchell, Q. Y. Duan, F. Chen, and J. M. Baker (1999), A parameterization of snowpack and frozen ground intended for NCEP weather and climate models, J. Geophys. Res., 104, 19,569–19,585, doi:10.1029/1999JD900232.
Koster, R. D., et al. (2004), Regions of strong coupling between soil moisture and precipitation, Science, 305(5687), 1138–1140, doi:10.1126/ science.1100217.
Koster, R. D., et al. (2011), The second phase of the global land–atmosphere coupling experiment: Soil moisture contributions to subseasonal forecast skill, J. Hydrometeorol., 12(5), 805–822, doi:10.1175/2011JHM1365.1.
Lawrence, D. M., and A. G. Slater (2008), Incorporating organic soil into a global climate model, Clim. Dyn., 30(2-3), 145–160, doi:10.1007/ s00382-007-0278-1.
Letts, M. G., N. T. Roulet, N. T. Comer, M. R. Skarupa, and D. L. Verseghy (2000), Parametrization of peatland hydraulic properties for the Canadian land surface scheme, Atmosphere-Ocean, 38(1), 141–160, doi:10.1080/07055900.2000.9649643.
Li, H., M. Huang, M. S. Wigmosta, Y. Ke, A. M. Coleman, L. R. Leung, A. Wang, and D. M. Ricciuto (2011), Evaluating runoff simulations from the Community Land Model 4.0 using observations fromflux towers and a mountainous watershed, J. Geophys. Res., 116, D24120, doi:10.1029/2011JD016276.
Liang, X., D. P. Lettenmaier, E. F. Wood, and S. J. Burges (1994), A simple hydrologically based model of land surface water and energyfluxes for general circulation models, J. Geophys. Res., 99, 14,415–14,428, doi:10.1029/94JD00483.
Lutz, A. F., W. W. Immerzeel, A. B. Shrestha, and M. F. P. Bierkens (2014), Consistent increase in High Asia’s runoff due to increasing glacier melt and precipitation, Nat. Clim. Change, 4(7), 587–592, doi:10.1038/nclimate2237.
Niu, G.-Y., Z.-L. Yang, R. E. Dickinson, and L. E. Gulden (2005), A simple TOPMODEL-based runoff parameterization (SIMTOP) for use in global climate models, J. Geophys. Res., 110, D21106, doi:10.1029/2005JD006111.
Niu, G.-Y., Z.-L. Yang, R. E. Dickinson, L. E. Gulden, and H. Su (2007), Development of a simple groundwater model for use in climate models and evaluation with Gravity Recovery and Climate Experiment data, J. Geophys. Res., 112, D07103, doi:10.1029/2006JD007522. Niu, G.-Y., 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. 16, D12109, doi:10.1029/2010JD015139.
Niu, G.-Y., C. Paniconi, P. A. Troch, R. L. Scott, M. Durcik, X. Zeng, T. Huxman, and D. C. Goodrich (2014), An integrated modelling framework of catchment-scale ecohydrological processes: 1. Model description and tests over an energy-limited watershed, Ecohydrology, 7(2), 427–439, doi:10.1002/eco.1362.
Oleson, K., et al. (2013), Technical description of version 4.5 of the Community Land Model (CLM), NCAR Technical Note NCAR/TN-503+STR. Saxton, K. E., and W. J. Rawls (2006), Soil water characteristic estimates by texture and organic matter for hydrologic solutions, Soil Sci. Soc.
Am. J., 70(5), 1569–1578, doi:10.2136/sssaj2005.0117.
Schaake, J. C., V. I. Koren, Q.-Y. Duan, K. Mitchell, and F. Chen (1996), Simple water balance model for estimating runoff at different spatial and temporal scales, J. Geophys. Res., 101, 7461–7475, doi:10.1029/95JD02892.
Shangguan, W., et al. (2013), A China data set of soil properties for land surface modeling, J. Adv. Model. Earth Syst., 5(2), 212–224, doi:10.1002/jame.20026.
Shi, Y., K. J. Davis, C. J. Duffy, and X. Yu (2013), Development of a coupled land surface hydrologic model and evaluation at a critical zone observatory, J. Hydrometeorol., 14(5), 1401–1420, doi:10.1175/JHM-D-12-0145.1.
Todini, E. (1996), The ARNO rainfall-runoff model, J. Hydrol., 175(1–4), 339–382, doi:10.1016/S0022-1694(96)80016-3.
van der Velde, R., Z. Su, M. Ek, M. Rodell, and Y. Ma (2009), Influence of thermodynamic soil and vegetation parameterizations on the simulation of soil temperature states and surfacefluxes by the Noah LSM over a Tibetan plateau site, Hydrol. Earth Syst. Sci., 13(6), 759–777, doi:10.5194/hess-13-759-2009.
Xue, B.-L., et al. (2013), Modeling the land surface water and energy cycles of a mesoscale watershed in the central Tibetan Plateau during summer with a distributed hydrological model, J. Geophys. Res. Atmos., 118, 8857–8868, doi:10.1002/jgrd.50696.
Yang, K., T. Koike, B. Ye, and L. Bastidas (2005), Inverse analysis of the role of soil vertical heterogeneity in controlling surface soil state and energy partition, J. Geophys. Res., 110, D08101, doi:10.1029/2004JD005500.
Yang, K., Y. Y. Chen, and J. Qin (2009), Some practical notes on the land surface modeling in the Tibetan Plateau, Hydrol. Earth Syst. Sci., 13(5), 687–701, doi:10.5194/hess-13-687-2009.
Yang, Z.-L., et al. (2011), The community Noah land surface model with multiparameterization options (Noah-MP): 2. Evaluation over global river basins, J. Geophys. Res., 116, D12110, doi:10.1029/2010JD015140.
Zeiliguer, A. M., Y. A. Pachepsky, and W. J. Rawls (2000), Estimating water retention of sandy soils using the additivity hypothesis, Soil Sci., 165(5), 373–383.
Zeng, X., Z. Wang, and A. Wang (2012), Surface skin temperature and the interplay between sensible and ground heatfluxes over arid regions, J. Hydrometeorol., 13(4), 1359–1370, doi:10.1175/jhm-d-11-0117.1.
Zhang, L., F. Su, D. Yang, Z. Hao, and K. Tong (2013), Discharge regime and simulation for the upstream of major rivers over Tibetan Plateau, J. Geophys. Res. Atmos., 118, 8500–8518, doi:10.1002/jgrd.50665.
Zheng, D., R. van der Velde, Z. Su, M. J. Booij, A. Y. Hoekstra, and J. Wen (2014), Assessment of roughness length schemes implemented within the Noah land surface model for high-altitude regions, J. Hydrometeorol., 15(3), 921–937, doi:10.1175/jhm-d-13-0102.1.
Zheng, D., R. Van der Velde, Z. Su, J. Wen, M. J. Booij, A. Y. Hoekstra, and X. Wang (2015a), Under-canopy turbulence and root water uptake of a Tibetan meadow ecosystem modeled by Noah-MP, Water Resour. Res., 51, 5735–5755, doi:10.1002/2015WR017115.
Zheng, D., R. Van der Velde, Z. Su, X. Wang, J. Wen, M. J. Booij, A. Y. Hoekstra, and Y. Chen (2015b), Augmentations to the Noah model physics for application to the Yellow River source area. Part II: Turbulent heatfluxes and soil heat transport, J. Hydrometeorol., 16(6), 2677–2694, doi:10.1175/JHM-D-14-0199.1.
Zheng, D., R. Van der Velde, Z. Su, X. Wang, J. Wen, M. J. Booij, A. Y. Hoekstra, and Y. Chen (2015c), Augmentations to the Noah model physics for application to the Yellow River Source Area. Part I: Soil waterflow, J. Hydrometeorol., 16(6), 2659–2676, doi:10.1175/JHM-D-14-0198.1. Zheng, D., R. Van der Velde, Z. Su, J. Wen, X. Wang, M. J. Booij, A. Y. Hoekstra, S. Lv, Y. Zhang, and M. B. Ek (2016), Impacts of Noah model
physics on catchment-scale runoff simulations, J. Geophys. Res. Atmos., 121, 807–832, doi:10.1002/2015JD023695.
Zhou, D., and R. Huang (2012), Response of water budget to recent climatic changes in the source region of the Yellow River, Chin. Sci. Bull., 57(17), 2155–2162, doi:10.1007/s11434-012-5041-2.
