computers and electronics in agriculture · and albedo feedback; however the key feature was the...

Computers and Electronics in Agriculture xxx (2012) xxx–xxx

Contents lists available at SciVerse ScienceDirect

Computers and Electronics in Agriculture

journal homepage: www.elsevier .com/locate /compag

Noah-GEM and Land Data Assimilation System (LDAS) based downscalingof global reanalysis surface fields: Evaluations using observations from aCarboEurope agricultural site

Umarporn Charusombat a, Dev Niyogi a,⇑, Sébastien Garrigues b, Albert Olioso b, Olivier Marloie b,Michael Barlage c, Fei Chen c, Michael Ek d, Xuemei Wang e, Zhiyong Wu e

a Purdue University, 915 W. State Street, West Lafayette, IN 47907-2054, United Statesb UMR 1114 EMMAH, INRA, Domaine St Paul, Site Agroparc, 84914 Avignon cedex 9, Francec National Center for Atmosphere Research, 3450 Mitchell Lane, Boulder, CO 80301, United Statesd NOAA/NCEP 5200 Auth Rd, Rm 207 Suitland, MD 20746-4304, United Statese School of Environmental Science and Engineering, Sun Yat-Sen University, Guangzhou 510275, China

a r t i c l e i n f o

Article history:Received 16 July 2011Received in revised form 20 October 2011Accepted 8 December 2011Available online xxxx

Keywords:NoahNoah-GEMFluxnetLatent heat fluxCarboEuropeLand Data Assimilation System

0168-1699/$ - see front matter � 2011 Elsevier B.V. Adoi:10.1016/j.compag.2011.12.001

⇑ Corresponding author.E-mail address: [email protected] (D. Niyogi).

Please cite this article in press as: Charusombasurface fields: Evaluations using observations fr

a b s t r a c t

This study provides the first assessment of the Noah and Noah-GEM (photosynthesis-based Gasexchange Evapotranspiration Model) land surface model using observations from the Avignon, FranceCarboEurope agricultural site during 2006 and 2007. Noah and Noah-GEM are integrated within a LandData Assimilation System (LDAS) framework. The LDAS fields of soil moisture, temperature field, andsurface and subsurface water and energy budget terms are useful for meteorological model initial con-ditions, and agricultural applications. The models were integrated using 1 km grid spacing with mete-orological forcing from the Japanese global reanalysis (JRA). Consistent with results compiled over theUS Southern Great Plains, the Noah and Noah-GEM based model performance was comparable for sor-ghum and wheat cropland. Both models had a relatively better performance during the low LAI plantgrowth stage however the performance deteriorated during peak green conditions and the bias betweenthe observed and modeled latent heat flux was consistently higher by 100 W m�2. To further diagnosethis bias, a series of experiments were undertaken by considering observed biweekly dynamic leaf areaindex (LAI), vegetation height, roughness length (z0), and albedo changes. These experiments were con-ducted using Noah-GEM because of similar results between Noah and Noah-GEM and also becauseNoah-GEM has an explicit C3 and C4 photosynthesis model. The results were compared with the defaultmodel run as well as in situ surface flux and soil moisture/temperature observations. Prescribing onsitecharacteristics led to modest improvements in the model fields, however the model still could not cap-ture the peak growing heat flux values of sensible heat for both C3 and C4 plants. Additional experi-ments were undertaken to investigate the inconsistencies in model parameterization. These includeexperiments with a CO2-based transpiration and thermal roughness formulation in surface-layer phys-ics; the surface coupling coefficient through the ‘‘Zilitinkevich constant’’; effect of soil texture andmodel spin-up time. Based on the study results and the experiments, we conclude that a high resolutionLDAS/Noah setup can be driven using global reanalysis fields producing reasonably good results whenevaluated against point observations. The model performance was enhanced after using dynamic LAIand albedo feedback; however the key feature was the tuning of the model structure through couplingand modifying Vmax as a function of LAI. These results highlight the need for improvements in the tur-bulent surface layer and plant physiological modules, and model deficiencies cannot be overcome byonsite biophysical data alone.

� 2011 Elsevier B.V. All rights reserved.

1. Introduction

A number of agricultural and environmental applications re-quire spatiotemporal soil moisture/soil temperature fields for the

ll rights reserved.

t, U., et al. Noah-GEM and Landom a CarboEurope agricultural

surface and subsurface. Soil parameters however are difficult tomonitor and the sensors require routine maintenance, have a local-ized footprint, and are expensive. Satellite remote-sensed productscan be an alternative to in situ soil moisture/temperature measure-ments (Jackson, 1993). However remote-sensed products also havelimitations such as they typically have a coarser resolution and canprovide the skin/surface conditions with a limited ability to

Data Assimilation System (LDAS) based downscaling of global reanalysissite. Comput. Electron. Agric. (2012), doi:10.1016/j.compag.2011.12.001

http://dx.doi.org/10.1016/j.compag.2011.12.001

mailto:[email protected]


http://www.sciencedirect.com/science/journal/01681699

http://www.elsevier.com/locate/compag


2 U. Charusombat et al. / Computers and Electronics in Agriculture xxx (2012) xxx–xxx

understand root level changes. As a result, Land Data AssimilationSystems (LDAS) are an emerging approach (Houser et al., 1998).LDAS can be configured with a gridded domain and can assimilateheterogeneous observations into a uniform field for surface andsubsurface soil variables. In this paper, we focus on the Noah landsurface model-based LDAS applications over a European croplandsite.

The Noah-based LDAS is often used for providing accuratebiophysical and soil moisture/temperature fields that are neededto predict the weather and regional meteorological processes(Chen and Dudhia, 2001; Holt et al., 2006). The LDAS fields can alsobe important input parameters for crop models such as theDecision Support System for Agrotechnology Transfer (DSSAT). Inaddition, the Noah-based LDAS is used to estimate and monitordrought stress and is thus an important component for agronomicappreciations (Thorp et al., 2008).

This paper concerns the testing of the performance of defaultNoah and Noah-GEM (photosynthesis-based Gas exchange Evapo-transpiration Model) over field sites using observations availablefrom a French CarboEurope flux site (Reichstein et al., 2005) andwww.carboeurope.org. The interesting feature of this field site isthe rotation of different crop types using C3 (wheat) and C4 (sor-ghum) photosynthesis pathways which then affects the waterand energy flux processes (Farquhar and Sharkey, 1982).This isthe first evaluation of the Noah and Noah-GEM system over theFrench agriculture field site. The majority of model evaluationsfor Noah and Noah-GEM have been over the US using the Ameri-flux and DOE Atmospheric Radiation Measurement (ARM) specialfield experimental data (e.g., International H2O Project-IHOP,2002, LeMone et al., 2008). The tests over a French CarboEurope

Fig. 1. (a) Location of study domain in Avignon, France, (b) USGS land use map (1–24) forthe study site.

Please cite this article in press as: Charusombat, U., et al. Noah-GEM and Landsurface fields: Evaluations using observations from a CarboEurope agricultural

flux site are of interest for a number of reasons. First, the modelneeds to be tested over different sites as globally as possible andthe region is important for energy and water cycle studies. Second,the tests are over C3 and C4 vegetation, which allows the testingand verification of model physiological modules. Third, many ofthe default parameterizations and constants used in the land sur-face model are based on observations made over France as partof the HAPEX-MOBILHY program (André et al., 1986; Noilhan andPlanton, 1989). Therefore, it is presumed that testing the modelunder an almost default environment will allow for greater under-standing of the uncertainties and help to guide future modelenhancements. The model is typically implemented at the conti-nental and global scale for meteorological and climatic studies.This test will provide insights on the model performance at the lo-cal scale using standard parameters and global atmosphericforcing.

The following section provides a summary of the study site andmodel setup involving Noah and Noah-GEM and its integrationusing the Land Data Assimilation System (LDAS) (Chen et al.,2007). Section 3 discusses the model results for the default config-uration and outlines the results from a number of sensitivity stud-ies and the experiments undertaken. Section 4 presents the studyconclusions.

2. Experimental

2.1. Site description

Measurements were performed at the CarboEurope site inAvignon, southeastern France, located at 43� 550 0000 N, 4� 520

the study domain used in the LDAS runs, and (c) site picture of the flux station over


http://www.carboeurope.org


U. Charusombat et al. / Computers and Electronics in Agriculture xxx (2012) xxx–xxx 3

4700 E (Fig. 1). The site is located in a semi-urban area and in atypical Mediterranean climate (annual climatic mean of 14 �C fortemperature and 680 mm for precipitations). However during the2004–2007 period the yearly average rainfall was 450 mm.Wheat, which shows a C3 carbon photosynthesis pathway, andsorghum, which exhibits a C4 carbon pathway, were grown onsilty clay loam soil. Sorghum was irrigated. We have selectedthe period from 1 January to 30 June 2006 with wheat and 1

Fig. 2. Model configuration and setup us

Table 1Summary of experiments.

#Run Model Year

1, 2 Noah 2005–2006, 20073, 4 Noah-GEM 2005–2006, 20075, 6 Noah-GEM 2005–2006, 20077, 8 Noah-GEM 2005–2006, 20079, 10 Noah-GEM 2005–2006, 200711, 12 Noah-GEM 2005–2006, 200713, 14 Noah-GEM 2005–2006, 200715, 16 Noah-GEM 2005–2006, 200717, 18 Noah-GEM 2005–2006, 200719, 20 Noah-GEM 2005–2006, 200721, 22 Noah-GEM 2005–2006, 200723, 24 Noah-GEM 2005–2006, 200725, 26 Noah-default 2005–200627, 28 Noah-default 2005–200629, 30 Noah-default 2005–200631, 32 Noah-default 2005–2006


May–30 October 2007 with sorghum for the model analysis.The difference between C3 and C4 is the carbon fixation pro-cesses; the C3 plants incorporate carbon from atmospheric CO2

within three carbon compounds, while C4 plants incorporate itwithin four carbon compounds. C4 plants also use specific mech-anisms for concentrating CO2 at the carboxilation site, so thatthey more efficiently use CO2 than C3 at present atmosphericCO2 concentrations.

ed in the study (⁄changes biweekly).

Crop Experiments

Wheat, Sorghum DefaultWheat, Sorghum DefaultWheat, Sorghum Biweekly LAIWheat, Sorghum Biweekly LAI + albedoWheat, Sorghum Biweekly LAI + albedo + roughness length (z0)Wheat, Sorghum Czil = 0.001Wheat, Sorghum Czil = 0.01Wheat, Sorghum Czil = 1Wheat, Sorghum Czil = 10�0.4h

Wheat, Sorghum Vmax � LAIWheat, Sorghum Vmax � LAI/2Wheat, Sorghum Vmax/LAIWheat 1 year Spin-upWheat 1 month Spin-upWheat 1 week Spin-upWheat Silty clay loam




At the Avignon site, continuous monitoring of crop and soilprocesses is undertaken, including energy balance, water and car-bon surface-atmosphere exchanges, crop structure (leaf area index,vegetation height), crop biomass, soil moisture, soil temperature,surface temperature (thermal infrared measurements), and canopyreflectance. Energy fluxes were measured continuously using theeddy covariance technique at 30 min intervals using a Young81000 3D sonic anemometer and a LiCOR Li7500 open path

Fig. 3. Time series of daily surface fluxes from observations (solid black); observed-ccompared with Noah (blue) and Noah-GEM (red) during the 2006 (wheat) and 2007 (sorthis figure legend, the reader is referred to the web version of this article.)


CO2/H2O analyser. We used ‘‘Level 4’’ i.e., quality-checked andgap-filled data from the CarboEurope database (Reichstein et al.,2005). Moreover, footprint analysis suggests that, on average,85–90% of the fluxes originated from the fields. Information onthe accuracy of the CarboEurope data can be found in Mauderet al. (2008). In the CarboEurope context, the data were used beforefor analyzing CO2 surface-atmosphere exchanges and productionof full crop rotation (e.g., Kutsch et al., 2010).

orrected with net radiation (dash black) and corrected with Bowen ratio (green)ghum) growing season (Exp. 1–3). (For interpretation of the references to colour in




2.2. Model description

In this section we briefly outline the Noah, Noah-GEM, andLDAS model system used in this study.

2.2.1. Noah land surface modelThe community Noah land surface model (LSM) is a part of the

Weather Research and Forecasting (WRF) modeling suite. TheNoah land model (Chen et al., 1996; Ek et al., 2003) is the soilvegetation atmosphere transfer (SVAT) component of the WRFadvanced research version (ARW, Skamarock et al., 2005), theWRF-Chem (Grell et al., 2005), Regional Climate Region-WRF(Leung et al., 2006), Hurricane WRF-HWRF (Gopalakrishnan et al.,2010), and Advance Hurricane WRF ARW (Davis et al., 2008). It isalso a critical component of different Land Data Assimilation Sys-tems such as the US North American Land Data Assimilation Sys-tem NLDAS (Mitchell et al., 2004), global LDAS (Rodell et al.,2004), NASA Land Information System (LIS, Kumar et al., 2008)and NCAR High Resolution LDAS (HRLDAS, Chen et al., 2007). TheNoah model has a number of modifications such as the WRF Noah

Table 2Mean degree of agreement (d) between model and observations (LE, latent heat flux; SH,temperature at surface level).

Experiments 2006

LE SH NR G SM

Noah default 0.87 0.84 0.95 0.78 0.99Noah-GEM 0.83 0.84 0.95 0.78 0.99Noah-adjust 0.89 0.93Noah-GEM-LAI 0.85 0.84 0.95 0.78 0.99Noah-GEM-z0 0.85 0.84 0.95 0.78 0.99Noah-GEM-Albedo 0.86 0.85 0.95 0.79 0.99Noah-GEM-C0.01 0.82 0.78 0.98Noah-GEM-C1 0.66 0.73 0.98Noah-GEM-Cdyn 0.80 0.74Noah-GEM-Vmax � LAI 0.91 0.86 0.99Noah-GEM-Vmax � LAI/2 0.89 0.86Noah-GEM-Vmax/LAI 0.84 0.97Noah spin-up 5 years 0.89 0.84 0.99Noah spin-up 1 years 0.89 0.84 0.99Noah spin-up 1 month 0.88 0.83 0.99Noah spin-up 1 week 0.87 0.83 0.99

Table 3Mean fractional bias (FB) between model and observations (LE (W m�2), latent heat flux; SHSM (m3 m�3), soil moisture; ST (K), soil temperature at surface level).

Experiments 2006

LE SH NR G SM

Noah default 0.35 �0.70 0.18 �2.22 �0.1Noah-GEM 0.48 �1.02 0.19 �2.20 �0.15Noah-GEM-adjust 0.20 0.09Noah-GEM-LAI 0.45 �0.96 0.19 �2.21Noah-GEM-z0 0.38 �0.79 0.18 �2.21Noah-GEM-Albedo 0.39 �0.46 0.24 �2.22 �0.09Noah-GEM-C0.0001 0.28 0.11Noah-GEM-C0.01 0.33 �0.33 0.13Noah-GEM-C1 0.72 �0.86 0.27Noah-GEM-Cdyn 0.66 �0.76Noah-GEM-Vmax � LAI 0.24 0.82 �0.22Noah-GEM-Vmax � LAI/2 0.32 �0.13Noah-GEM-Vmax/LAI �0.75 �0.44Noah spin-up 5 years 0.34 �0.66 �0.11Noah spin-up 1 years 0.34 �0.66 �0.11Noah spin-up 1 month 0.22 �0.06 �0.10Noah spin-up 1 week 0.31 �0.53 �0.13


(Chen and Dudhia, 2001; Ek et al., 2003), Noah-GEM (Niyogi et al.,2009), and Noah-MP (Niu et al., 2011).

The Noah model is a community LSM involving the operationalresearch and academic community. It is built on the original Ek andMahrt (1991) Oregon State University (OSU) land model and hasfour soil layers of increasing thickness (0.1, 0.3, 0.6, and 1.0 m)and a single implicit vegetation layer. The model is complex andhas various processes to represent a multi-scale water and energybalance for a variety of land surface and soil conditions. Noah cal-culates prognostic changes in the variables including energy flux(sensible heat, latent heat, ground heat flux, and net radiation), soilmoisture, soil temperature, water storage in the canopy and snowon the ground as a dynamic hydrological and surface energy feed-back. The ground heat flux (G) is estimated by the diffusion equa-tion from soil temperature (T) and thermal conductivity (K) (detailsin Pan and Mahrt, 1987). The sensible heat flux (H) is calculatedfrom an aerodynamic formula which is a function of specific heat,air, and skin temperature and drag coefficient (CH). The importantfeature of the Noah model is the estimation of the water vapor flux(LE) from soil evaporation and plant transpiration. The bare soilevaporation and vegetation feedback within the Noah land surface

sensible heat flux; NR, net radiation; G, ground heat flux; SM, soil moisture; ST, soil

2007

ST LE SH NR G SM ST

0.99 0.90 0.81 0.97 0.48 0.97 0.990.99 0.90 0.81 0.97 0.45 0.97 0.99

0.92 0.860.99 0.90 0.83 0.97 0.44 0.97 0.990.99 0.91 0.84 0.97 0.44 0.97 0.990.99 0.91 0.84 0.97 0.44 0.97 0.990.99 0.88 0.79 0.97 0.99

0.72 0.72 0.930.90 0.84

0.99 0.92 0.85 0.97 0.990.91 0.840.89 0.81 0.94

0.990.990.990.99

(W m�2), sensible heat flux; NR (W m�2), net radiation; G (W m�2), ground heat flux;

2007

ST LE SH NR G SM ST

�0.005 0.21 �0.43 0.23 0.65 �0.22 0.001�0.005 0.23 �0.41 0.23 0.62 �0.26 0.001

0.07 �0.110.29 �0.12 0.23 0.630.28 �0.14 0.22 0.62

�.005 0.27 �0.16 0.20 0.61 �0.25 0.00070.23 �0.22

�.004 0.23 �0.21 �0.03 0.0030.76 0.41 �0.50.35 0.12

�.004 0.24 �0.13 �0.22 0.00080.28 �0.170.17 �0.05 �0.44

0.0040.0040.0040.004




model are represented following a Penman-based energy balanceapproach to calculate the potential evaporation (Betts et al., 1997).

The transpiration from plants is a function of the stomatal resis-tance (Chen et al., 1996) which is estimated by the Jarvis schememodel (Niyogi and Raman, 1997).

Rc ¼Rcmin

LAI � F1F2F3F4ð1Þ

where Rcmin is the minimum resistance and LAI is the leaf area in-dex. The multiplicative factors represent the stress effect of solarradiation (F1), vapor pressure deficit (F2), air temperature (F3), andsoil moisture (F4) on the plant.

The model is setup over the domain of interest using USGS 30-sglobal 24-category vegetation data and 16 categories of hybridSTATSGO/FAO (5-min grid spacing soil texture, Chen and Dudhia,2001). The vegetation and soil characteristics are based on look-up tables. The high resolution terrain and topography was down-loaded from the USGS website (http://seamless.usgs.gov/) andinterpolated using the WRF Preprocessing System (more detailsregarding the process and the model can be found in Chen andDudhia, 2001; Ek et al., 2003).

2.2.2. Noah-GEMThe Noah-GEM model was developed and tested by Niyogi et al.

(2009) to enhance the vegetation transpiration feedback within theNoah model. The model employs a photosynthesis transpirationscheme using the Ball-Berry photosynthesis–transpiration ap-proach (Ball et al., 1987) as an alternate to the Jarvis scheme whichis more commonly used in many weather and climate models inthe Noah model (Eq. (1)). Thus Noah-GEM only replaces the tran-spiration component; stomatal resistance is calculated as

1=Rc ¼ mAn

Cshs þ b ð2Þ

where An is the net photosynthesis rate, hs is relative humidity atthe canopy surface, and Cs is the CO2 concentration at the canopy

Fig. 4. Day (a and b) and nighttime (c and d) latent heat and sensible heat flux values fordynamic roughness length (z0) (green), and dynamic albedo (blue) for 1 January 2006–3reader is referred to the web version of this article.)


surface. The terms m and b are the species-specific gas exchangeconstants (Collatz et al., 1991). Details regarding the formulationsand constants can be found in Niyogi et al. (2009) and an updateis available in Wu et al. (2011). The model has explicit C4 and C3modules based on the photosynthesis model detailed in Collatzet al. (1991, 1992). The limiting carbon assimilation rate in C3and C4 plants are calculated separately as a function of maximumcatalytic Rubisco capacity (Vm), the photosynthetically active radia-tion or light limitation, and the phosphoenolpyruvate (PEP) carbox-ylase limitation, along with the estimates of quantum efficiency forcarbon dioxide uptake, leaf-scattering coefficient and surface atmo-spheric pressure. The estimation of carbon assimilation rates is usedfor calculating transpiration rates and the stomatal/canopy resis-tance term (Farquhar and Sharkey, 1982).

The major photosynthesis parameter is the catalystic Rubiscocapacity for the leaf (Vm) which is parameterized as a function ofVmax values, soil moisture, and temperature as

Vm ¼ Vmaxf ðTÞf ðw2Þ ð3Þ

where f(w2) is the soil moisture content term and f ðTÞ is thetemperature term. The temperature term has been calculateddifferently between C3 and C4 plants depending on the functionof the vegetation stress factor (S1, S2, S3, and S4) following Sellerset al. (1996). Noah-GEM is integrated within Noah and uses thebroader Noah hydrometeorological framework.

2.3. High resolution Land Data Assimilation System (LDAS)

The Noah and Noah-GEM are run as a part of the LDAS, whichwas configured and centered over the Avignon field site(43.92� N 4.88� E) in France. The model was set up within 1 km gridspacing based on 50 � 50 grid points. HRLDAS input required themeteorological and surface forcing variables which were retrievedfrom Japan global reanalysis (JRA) fields. The JRA is a recent high-quality reanalysis, which incorporates many observational datainto the assimilation system (Onogi et al., 2007). JRA-25 has a

observations (black) compared with model runs: Noah-GEM (red), Noah-GEM with0 June 2006. (For interpretation of the references to colour in this figure legend, the


http://seamless.usgs.gov/



horizontal resolution around 120 km and 40 vertical layers to thetop of atmosphere. Note that, an alternate approach is to use theGlobal Land Data Assimilation System (GLDAS) forcing, but oneof the study objectives is also to determine if the JRA Climate DataAssimilation System (JCDAS) analysis could be used as an input toLDAS. The JRA-25 is considered a good reference dataset and hasbeen used to validate precipitation and wind profiles over Europeas described in Bosilovich et al. (2008) and Hatsushika et al.(2006). Also the impact of using forcing variables with Avignondata can be found at http://www.biogeosciences-discuss.net/8/2467/2011/bgd-8-2467-2011.html.

For LDAS implementation, the 6-hourly JRA forecasted air tem-perature, specific humidity, wind speed, surface pressure, precipi-tation, and downward shortwave and longwave radiation wereinterpolated into an hourly input as required by the model driver(Chen et al., 2007). The output data from LDAS includes energybalance components, soil moisture, and soil temperature and wasdownscaled into 1 km grid spacing (Fig. 2) The LDAS was run in

a

c

1 Jan - 30 Jun 2006

1 Jan - 30 Jun 2006

Ho

- H

m (

W m

-2)

Ho

- H

m (

W m

-2)

120

100

80

60

40

20

0

0.0 1.0 2.0 3.0 4.0 5.0 6.0

120

100

80

60

40

20

0

0.0 0.2 0.4 0.6 0.8 1.0

LAI

Roughness length

r = 0.22

r = 0.36

Wheat

Wheat

Fig. 5. The difference between the observed (Ho) and modeled (Hm) sensible heat flux foobserved (SMo) and modeled (SMm) soil moisture during the 2006 growing season.


offline mode from 1 January 2002 to 30 October 2007 with defaultspin up run for 4 years.

2.4. Model runs

In this study, first the control versions of Noah and Noah-GEMwere used over the study domain to test their performance, and re-sults were compared with observations. The results are shown inSection 3.1. The next sets of experiments were conducted to fur-ther evaluate the performance of Noah-GEM by prescribingdynamical biweekly LAI, albedo, and roughness length, which wereinterpolated from in situ observations a month apart. These resultsare discussed in Section 3.2.1. A review of these experiments wasfollowed by an assessment of the sensitivity of the coupling coeffi-cient by changing the Zilitinkevich coefficient (Czil) from 0.01 to 1and 0.001 and by calculating it as Czil = 10�0.4h, where h is the bi-weekly canopy height. This coefficient, Czil, is used to calculatethe ratio of roughness length for momentum roughness length

b

d 1 Jan - 30 Jun 2006

1 Jan - 30 Jun 2006

Ho

- H

m (

W m

-2)

Ho

- H

m (

W m

-2)

120

100

80

60

40

20

0

0 0.3 0.6 0.9 0.12 0.15 0.18

120

100

80

60

40

20

0

0.160 0.180 0.200 0.220 0.240 0.260

Albedo

SMo - SMm (m3 m-3)

r = 0.15

r = -0.19

Wheat

Wheat

r different (a) LAI, (b) albedo, (c) roughness length, and (d) the difference between


http://www.biogeosciences-discuss.net/8/2467/2011/bgd-8-2467-2011.html

http://www.biogeosciences-discuss.net/8/2467/2011/bgd-8-2467-2011.html



for heat/moisture (Chen et al., 2010; Chen and Zhang, 2009). Exper-iments were also undertaken by scaling Vmax with LAI and resultsare presented in Section 3.2.2. The sensitivity of HRLDAS to spin-up time is in Section 3.2.3. Finally, the model was also tested withdifferent soil textures which were changed from loam to silty clayloam (results shown in Section 3.2.4).

The effect of model spin-up time was also evaluated by chang-ing the spin-up time from 4 years (2002) to 1 year (2005), 1 month,and 1 week. All experiments in this study are summarized inTable 1. The models and observations were compared with degreeof agreement and fractional bias. The degree of agreement (d) andfractional bias (FB) between observation and model results wascalculated using

d ¼ 1�Xn

i¼1

ð0i �miÞ2,Xn

i¼1

ð=0i=þ =mi=Þ2 ð4Þ

FB ¼ 2Pn

i¼10i

n�Pn

i¼1mi

n

� �� Pni¼10i

nþPn

i¼1mi

n

� �ð5Þ

where oi is the observation data, mi the model results, and n is thenumber of samples.

r = 0.57

a b

c d

1 Jan - 30 Jun 2006

LE

o -

LE

m (

W m

-2)

120

100

80

60

40

20

0

0.0 1.0 2.0 3.0 4.0 5.0 6.0

LAI

Wheat

1 Jan - 30 Jun 2006

LE

o -

LE

m (

W m

-2)

120

100

80

60

40

20

0

0.0 0.2 0.4 0.6 0.8 1.0

Roughness length

r = 0.64

Wheat

Fig. 6. Same as Fig. 5 but


3. Results and discussion

3.1. The comparison of Noah and Noah-GEM with observations

Fig. 3a–h shows the comparison between Noah, Noah-GEM andobservations to evaluate daily latent heat flux, sensible heat flux,net radiation, and ground heat flux. Noah default and Noah-GEMproduced comparable results that show the reanalysis data areappropriately incorporated within LDAS. The reanalysis data is ingood agreement with the observations. The degree of agreementand fractional bias values of both models compared with observa-tions and averaged over the entire season are shown in Tables 2and 3, respectively. The degree of agreement of observed and mod-eled latent heat, sensible heat, and radiation ranges from 0.81 to0.97 with the bias ranging from �1.0 to 0.23 W m�2. The groundheat flux has a low degree of agreement (0.45) and a high bias(�2.2 W m�2). For latent heat flux at the beginning of the growingseason (LAI �0–3), the models mostly underestimate latent heatflux by less than 50 W m�2 on average. However when the peakgreen growing stage began (high LAI �4–5) around 16 March2006, the mean differences between model and observations in-creased to 75–150 W m�2. Again, at the end of the growing season

1 Jan - 30 Jun 2006

LE

o -

LE

m (

W m

-2)

120

100

80

60

40

20

00.0 0.2 0.4 0.6 0.8 1.0

SMo-SMm (m3 m-3)

r = 0.22

Wheat

1 Jan - 30 Jun 2006

LE

o -

LE

m (

W m

-2)

120

100

80

60

40

20

0

0.160 0.180 0.200 0.220 0.240 0.260

Albedo

r = -0.09

Wheat

for latent heat flux.




or mature stage (LAI <3), the mean differences decreased to50 W m�2 for both C3 and C4 plants (Fig. 3a and b).

Similar results were seen for sensible heat flux (Fig. 3c and d). Atthe beginning of the growing season, both models underestimatesensible heat flux, then the models adjust the values up throughoutthe entire growing season. However, during the peak green grow-ing season, the sensible heat flux dropped abruptly resulting in thedifferences between models and observations showing an increaseof more than 50 W m�2. Neither model could capture the peak oflatent heat flux nor the drop off of sensible heat flux during thepeak green growing season between days 75–125 for wheat in2006 and 100–130 for sorghum in 2007. Thus, the model did notperform well under peak crop conditions. These issues have beenexplored further leading to the substance of the experimentsdescribed in Section 3.2 to assess the lack of performance undergrowth peak conditions.

Fig. 7. Same as Fig. 5 bu


The net radiation values are one of the problems (Fig. 3e and f)that show an almost 100 W m�2 difference between observationsand model for both C3 and C4 plants. We tried to redistributethe error to latent heat flux and sensible heat flux in the netradiation by adjusting the latent (LE) and sensible heat flux (H)with the difference of net radiation (NR) as

LEad ¼ LEo � NRo=NRm ð6Þ

Had ¼ Ho � NRo=NRm ð7Þ

The degree of agreement (Table 2) between modeled and ob-served latent heat flux increased as shown in Fig. 3a and b (dashedline during the peak green growing season). The results show a mod-est change in latent heat and sensible heat flux. Also there are manystudies showing that the eddy covariance method does not conserve

t for net radiation.




the energy balance equation (sensible + latent heat flux = netradiation � soil heat flux) (Dugas et al., 1991; Nie et al., 1992;Goulden et al., 1997; McCaughey et al., 1997; Mahrt, 1998; Twineet al., 2000).

Fig. 8. Observed and modeled latent heat flux from Noah (left column) and Noah-GEM (rstage for the 2006 (wheat) growing season.


Overall, the individual values of H and LE need to be appropri-ately adjusted to the energy balance by using the Bowen ratioclosure (Barr et al., 1994; Blanken et al., 1997). This methodassumes that the Bowen ratio (b = H/LE) was measured accurately

ight column) for (a and b) growing stage, (c and d) peak green, and (e and f) mature




by the eddy-covariance system and then used for adjusting both LEand H to preserve the energy balance. Accordingly we recalculatedlatent heat flux (LE) as (Rn � G)/1 + b and sensible heat flux (H) asb(LE), where Rn is net radiation and G is soil heat flux. Thenew adjustment also shows slightly reduced differences oflatent heat flux between modeled and observed by 10%, 20

Fig. 9. Same as Fig. 8 but for the 20


(W m�2) and increasing sensible heat flux by 5%, (10 W m�2). Over-all there is limited impact of adjusting the observational data andwas not considered further to avoid complications for modelevaluations.

Thus, it is apparent the model has inherent performance issuesparticularly for the peak green growing season, and there are also

07 (sorghum) growing season.




uncertainties in the surface characteristics used in the evaluation.For example, the site has a footprint that can stretch outside thecrop field and therefore has a 10–15% inherent uncertainty. Resultswere further evaluated in Fig. 4 for the day and nighttime period.The results show that during the daytime period, the differencesbetween observed and modeled were as high as 300 W m�2 for la-tent heat and 200 W m�2 for sensible heat flux. During the night-time, the absolute differences were lower, especially for sensibleheat flux. These results indicate that the error in the energy bal-ance during the daytime drives the differences between observedand modeled values. This would be expected since fluxes are gen-erally lower during nighttime (stable condition).

There is also a lag between the observed and modeled peak withthe model missing the peak green growing season latent heat flux.This feature is examined in the following section. The difference ofsensible heat flux between observations and simulations did notshow a strong correlation with the LAI, albedo, roughness length,and soil moisture (Fig. 5a–d). Fig. 6a and c shows the higher rough-ness length and LAI, and higher latent heat flux variability. The dif-ference in soil moisture does not show a significant impact on thedifference of both latent and sensible heat fluxes (Figs. 5 and 6cand d). The difference of observed and modeled net radiation in-creased when the difference between observed and modeled soilmoisture increased (Fig. 7d). The difference of net radiation has asmall correlation with LAI, albedo, and roughness length (Fig. 7a–c). We also clustered the data into three different periods; duringpeak growing season, and pre- and post peak green growing sea-son. The correlation, the degree of agreement, and the fractional

Fig. 10. (a) Rainfall and soil moisture, (b) sensible heat flux and (c)


bias between observations and model simulations were analyzed.Figs. 8 and 9a and b show the correlation between observed andmodeled latent heat flux is relatively high (0.69–0.79), as is the de-gree of agreement (0.85–0.89) for both Noah and Noah-GEM forwheat and sorghum. The corresponding biases were as low as0.05–0.3 W m�2. During the peak green of the growing stage, thecross correlation is higher (0.8), but the values are more scatteredand reduced the linear relationship between observed and mod-eled and increases the bias to 0.63–0.75 W m�2 (Figs. 8 and 9cand d). At the end of the growing season (mature stage), the modelestimated higher values of latent heat flux than was found inobservations for both Noah and Noah-GEM (Fig. 8e and f) for wheatand sorghum (Fig. 9e and f). The degree of agreement, the bias, andthe correlation decreased at the end of the growing season (maturestage) for wheat.

3.2. Further assessment

Diagnostics of the results and consultation with field agrono-mists raises the possibility that about 20 mm of irrigation was ap-plied intermittently as needed to keep the soil moisture from goingbelow the wilting point. Since the Noah model forces the soil mois-ture to remain within the wilting and field capacity range, the ef-fect of this remedial irrigation (applied once during 2006towards harvest and three times during 2007 during the earlyplanting period), is expected to be minimal. Nonetheless we con-ducted an experimental run by adding the specified irrigation for5 May 2006 and 16, 21, and 25 May 2007; the dates that irrigation

latent heat flux (C3 crop) for 2006 after considering irrigation.




was likely prescribed by the agronomists. We also forced the soilmoisture to approach the wilting point before applying irrigation.The corresponding time series (Fig. 10a–c) showed a modest irriga-tion effect which increased soil moisture by 10% but as a feedbackbetween different processes increased latent heat flux to 50 W m�2

at the end of the growing season. In this case, these results showthat minor irrigation does not significantly affect the energy bal-ance because unlike the field conditions, the model configurationnever allowed the soil moisture to go over saturation or below10% of wilting point. Therefore, when remedial irrigation was pre-scribed, the model has an inherently limited muted response whencompared to observations. Therefore, here the irrigation effect

Fig. 11. Difference between (a and b) latent heat flux, (c and d) sensible heat flux, and (echanges in albedo, roughness, and LAI during the 2006 and 2007 growing season (Exp.


alone is not the reason for the models’ inability to correlate withobservations.

3.2.1. The effect of vegetation characteristicsIn this section we discuss the experiment to further diagnose

Noah and Noah-GEM model performance over the study siteby comparing default Noah-GEM, Noah-GEM with dynamic(time-varying) LAI, roughness length, and albedo. Discussion inthis section is based on physiological changes including prescrip-tion of onsite time-varying LAI, height, and albedo. The experi-ments that were undertaken try to understand dynamicvegetation not currently considered in the model. Charusombat

and f) net radiation for default Noah-GEM and Noah-GEM with prescribed dynamic4–6).




et al. (2010) showed that for a forest site, LAI is an importantparameter to help with dynamical phenological changes. In thispaper, two additional parameters, roughness length (z0) andsurface albedo were altered using monthly observational datainterpolated to biweekly values. Fig. 11a–h shows the model dif-ferences between default Noah-GEM, Noah-GEM with dynamicLAI, Noah-GEM with dynamic roughness length (z0), andNoah-GEM with dynamic albedo. Results indicate that theperformance was slightly improved by incorporating weeklyupdated dynamic LAI based on field observations. The effect ofdynamic LAI causes the latent heat flux to improve by10–20 W m�2 for wheat and 5–10 W m�2 for sorghum duringthe pre- and peak season with a higher degree of agreementand lower fractional bias (Tables 2 and 3). After the peak greengrowing season, the simulated values reduced rapidly to�30 W m�2 from the Noah GEM default after adding LAI forwheat during days 145–160 and days 225–270 for sorghum. Sim-ilar results were seen for sensible heat flux. Dynamically updat-ing the LAI based on observations slightly decreased sensibleheat flux by only 2–5% during the peak green growing seasonand with an increase of 10% more during the beginning andend of the growing season. The effect of adding z0 and albedoare even smaller and the effect is sometimes opposite and com-pensatory of LAI changes. For instance, in the initial stage, chang-ing z0 and albedo reduced the differences of modeled andobserved latent heat flux in the pre- and post-peak season whilecorresponding to the effect of LAI changes; the height effect does

Fig. 12. (a and b) Latent heat and (c and d) sensible heat flux for Noah GEM wi


not show much difference. The effect was similarly noted in thesensible heat flux which reduced by 10–20 W m�2 during thepre- and peak green growing season (Fig. 11c and d) as expectedin both wheat and sorghum. During the peak season, the sensibleheat flux reduced to 60 W m�2. The differences in wheat aremuch larger than the sorghum, yet the differences steadily in-creased for the first growing period starting at �40 W m�2 andincreasing to 80 W m�2 within 50 days when it is greener. Interms of net radiation, it was also interesting that similarpatterns such as for sensible heat flux starting at �40 W m�2

continuously increased to 120 W m�2 for wheat. For sorghum,however, the net radiation started at 40 W m�2 and increasedto 160 W m�2 during peak LAI and fell back to 40 W m�2 at theend of the growing season. After assimilating LAI values, thenet radiation slightly decreased by only 0–10 W m�2 for wheat(C3) and slightly increased by almost 20 W m�2 for sorghumduring the pre-growing season (Fig. 11e and f). Ground heat flux(figure not shown) had no difference in modeled values afterapplying dynamic LAI, albedo, and roughness length. This indi-cates a need for a comprehensive dynamic formulation ratherthan for only one parameter based on assimilation or temporalprescription. Findings indicate definite improvements as ex-pected with incorporation of the onsite vegetation dynamic butdid not eliminate the differences that were seen during thegrowing period when compared with observations. Therefore,additional experiments were conducted to address the issuewithin the model structure.

th dynamic Czil and Czil = 0.01, 0.1, and 1 during 2006 and 2007 (Exp. 7–9).




3.2.2. The effect of the surface coupling and Vmax

A number of follow-up experiments were conducted with twoyielding relatively significant results and are discussed here. Aninteresting feature recently reported by LeMone et al. (2008) andChen and Zhang (2009) is the change in land atmosphere interac-tion using the coupling term (zom/zoh). The two studies concludedthat the accurate specification of the Zilitinkevich coefficient (Czil),which substantially modulates the ratio zom/zoh and hence the sur-face coupling strength, is important for simulating energy balanceprocesses between the land surface and atmosphere. In this study,we conducted four simulations to test changes in the Czil prescrip-tion. Czil was changed from 0.1, which is a default value in the Noahmodel, to 0.0001, 0.01, and 1. In addition, Czil was changed dynam-ically as a function of biweekly canopy height (Czil = 10�0.4h,Chenand Zhang, 2009). Results show that when the Czil values increased,the latent heat flux decreased by 10–30 W m�2 for both C3 and C4plants (Fig. 12a and b). The sensible heat flux also increased byabout 20 W m�2 during the pre-growing season and remainedsteady for the growing season, however values dropped by40 W m�2 at the end of the growing stage for C3 crops (Fig. 12c).Therefore the effect of increasing Czil is to decrease the surfaceexchange coefficient, which also increases the surface-air temper-ature gradient thereby yielding a greater sensible heat flux. Thischange in Czil involves an interaction between the surface skintemperature and the surface turbulence (exchange coefficient)

Fig. 13. (a and b) Latent heat flux and (c and d) sensible heat flux for Noah GEM


for sensible heat flux, with the latent heat flux also affected bythe canopy conductance, vegetation-species, and a function stageof growth such as LAI, greenness fraction, and plant physiologicaleffects. However, increasing Czil values reduced the sensible heatflux during the peak green growing season. Also, when Czil valueswere smaller, the latent heat flux increased and conversely de-creased the sensible heat flux. However when reducing the Czil toa value less than 0.01, the latent heat flux would not increaseand the sensible heat flux would not decrease any further. Interest-ingly, when adding the dynamic Czil, the latent heat flux rapidly in-creases to around 30 W m�2 and the sensible heat flux dropped to40 W m�2 during the peak green growing season. As a result, thedegree of agreement and bias with observations improved to 0.9for sorghum (Tables 2 and 3). However, for wheat, the dynamic Czil

did not rapidly change the latent or sensible heat flux during thepeak green growing season but helped reduce the latent heat fluxby 20 W m�2 throughout the season and increased the sensibleheat flux during the pre- and post-growing season. The dynamicCzil formulation decreased the bias of the model from observationsto �0.76 W m�2 during 2006 and 0.12 W m�2 during 2007 forwheat and sorghum (Table 3). Overall the models were improvedafter adjusting thermal roughness length (Chen et al., 2010).

The physiological feature at the core of the leaf feedback isthe Vmax term. We changed Vmax scaling according to the leafarea index. The values were tested for both Vmax � LAI and

with carboxylation rate and LAI scaling during 2006 and 2007 (Exp. 10–12).




Vmax � LAI/2 and the Vmax/LAI results showed a dramaticimprovement in this experiment. This was tested to see if Vmax

scaled directly or inversely to the LAI. Fig. 13a–d shows the dy-namic Vmax with LAI increased latent heat flux and decreasedsensible heat flux during the peak green growing stage by40 W m�2 for both wheat and sorghum. This was an importantchange that increased the correlation. The degree of agreementwith observations of latent heat flux increased to 0.91 and0.92 for wheat and sorghum and the bias reduced to0.24 W m�2 (Tables 2 and 3). Similar results were observed forsensible heat flux. These two experiments confirm that not onlythe physical representation of the crop needs to be modified inthe model but also the surface layer turbulence, and thus cou-pling with the (lower) atmosphere of the crop needs to be incor-porated in the dynamic vegetation parameters for it to beeffective in improving the model performance.

3.2.3. The sensitivity of HRLDAS to spin-up timeThe last experiment tests the effect of spin-up time for running

the model. The recommended spin-up time is between 12 and18 months in order to remove the memory of initial soil moisture(Chen et al., 2007). We tested the model driven by the reanalysisforcing with 1 week, 1 month, 1 year, and 4 years spin-up time.The results show no significant difference in estimating latentand sensible heat flux if the spin-up time is more than 1 year usingthe reanalysis forcing. It would be expected that spin-up could takemuch longer in regions with sparse vegetation, such as an aridclimate where the hydrological cycle is slower, and opposite for re-gions that are greener/wetter.

Fig. 14. Noah simulated (a) latent heat flux, (b) sensible heat flux, (c) soil moisture (0–101 week for 2006 (Exp. 13–16).


Fig. 14a and b shows latent heat flux was negligibly higher(around 1–10 W m�2) and sensible heat flux was lower (1–2 W m�2) if the spin-up time was only 1 month. Interestingly,when the spin-up time was just 1 week long, the latent heat fluxshowed differences of only 1–2 W m�2 and no differences at allfor sensible heat flux. The reason may be that initial soil moisturefortuitously matched the estimated soil moisture of the 1 yearspin-up time as shown in Fig. 14c in the reanalysis field. There isalso no difference in soil temperature for different spin-up timeperiods (Fig. 14d). In this study, the spin-up time effect on estimat-ing the energy balance is minimal, and could because of reanalysisdata used to run the model in the downscaling mode and needs tobe confirmed in follow-up studies. However, the spin-up of soilstates is usually a much more complicated issue, where vegetationand soil type, greenness fraction/plant density (LAI), and precipita-tion all play an important role.

3.2.4. The sensitivity of HRLDAS to soil moisture and textureFig. 15a–h shows that Noah-GEM estimated soil moisture is

slightly higher than that in the Noah model for every soil layer.The model slowly responded to the reducing soil moisture duringthe peak green growing period for deeper soil layers (Fig. 15d–f).The degrees of agreement between modeled and observed wheatand sorghum are 0.96 and 0.99. The degree of agreement decreasedand the bias increased when the soil depth increased, however,overall the model overestimated soil moisture by around 0.05–0.10 m3/m3. Our results correspond with Ingwersen et al. (2011),showing that Noah and Noah-GEM could not capture the soil waterdynamics that have a distinct gradient with soil depth. Differences

cm), and (d) soil temperature (0–10 cm) with spin-up time at 1 month, 1 year, and



Fig. 15. Observed and modeled soil moisture for different soil layers for the 2006 and 2007 growing season.


between simulations and observations still existed even thoughthe dynamic (time-varying) albedo and LAI were prescribed(Fig. 16a–f). The soil moisture values were lower after adding thedynamic albedo but higher when prescribed with dynamic LAI. Dif-ferences between the model simulations and observations mayalso be due to the mismatch between modeled soil type (loam)versus that observed in parts of the field (silty clay loam). There-fore the next simulations were conducted by changing the soil typeto silty clay loam in place of loam. With this change, the models


show good agreement for soil temperature from the beginning ofthe growing stage until the end of the growing season. Overall,after adding dynamic or observed LAI, albedo, thermal roughnesscoefficient, and Vmax into the model, the soil temperature wasinsignificantly changed (figure not shown).

Charusombat et al. (2010) tested the sensitivity of soil texturein Noah-GEM showing that soil texture and conductivity had littleimpact on stomatal conductance (by using deposition velocity as asurrogate). In this study, similar results were obtained with latent



Fig. 16. Observed and modeled soil moisture (0–10 cm) for (a and b) albedo, (c and d) Vmax scaling, and (e and f) coupling changes.


heat flux slightly increasing by 1–5 W m�2, sensible heat fluxdecreasing slightly, and soil moisture increasing by 0.05–0.15 m3/m3 (Fig. 17a–f).

4. Conclusions

This study further evaluates the importance of vegetation rep-resentation in the Noah and Noah-GEM land surface model. Thestudy focused on the performance of Noah-GEM to estimate theenergy balance and soil moisture for rotation crop C3 and C4 plantsat an agricultural field site with a Mediterranean climate in Avi-gnon, France. A number of simulations were conducted and resultsfrom over 32 experiments conducted by using the LDAS Noahmodel with 1 km grid spacing from 2006 to 2007 are discussed.The domain was configured with a 50 � 50 km grid spacing overthe Avignon site using 6-hourly Japanese global reanalysis dataas forcing variables. The experiments included integrating dynamicLAI, z0, and albedo. The main results show that Noah and Noah-GEM are comparable and have overall moderate correlation withonsite observations.


These results were thus consistent with studies conducted inthe US which confirmed that the downscaling scenario and useof reanalysis as forcing data are acceptable. However, the modelperformance deteriorated during the peak LAI or peak green grow-ing season. After integrating onsite time-varying LAI, z0, andalbedo, the model performance increased modestly. Furtherexperiments involved scaling Vmax and changing the thermalroughness coefficient. The model performance was significantlyimproved during the peak green growing stage after scaling Vmax

with LAI and using the thermal roughness coefficient. This studyalso assessed the impact of spin-up time (between 1 year, 1 month,or even 1 week) and soil texture. The results showed spin-up timeand soil texture impact is minimal when forced with the reanalysisfields at this site. The results also agree with Charusombat et al.(2010) concluding that the Noah-GEM model is not sensitive to soiltexture change as compared to vegetation change. For the future,the model generally behaves well and it would be advantageousto test it in other regions not previously tested. Future sites havebeen planned in the tropical Indian monsoon region. The down-scaling ability tested by increasing resolution of model output fromthe low resolution of input data give good confidence of model



Fig. 17. Time series for the difference between observed and modeled (a and b) latent heat flux, (c and d), and (e and f) soil moisture (0–10 cm) for different soil textures(loam and silty clay loam).


performance. Future focus should include both physical andphysiology changes in the model, such as the thermal roughnesscoefficient (Czil) and Vmax which needs to be verified further.

Acknowledgements

This research benefited through the DOE-ARM (08ER64674; Dr.R. Petty), NSF CAREER (ATM-0847472), NOAA/JCSDA Grant(NA06NES4400013), NASA Terrestrial Hydrology Program (Dr.Jared Entin), NOAA CPPA Grant NA09OAR4310193, and the ChineseNSF projects (U0833001, 40875076). We acknowledge the supportfrom NCAR Water System and BEACHON Programs. Field data wereacquired and processed in the frame of the CARBOEUROPE-IP andthe CARBOFRANCE project funded by the European FP7 Program(GOCECT-2003-505572) and the French Ministry in charge of Envi-ronment (GICC programme).


References

André, J.C., Goutorbe, J.P., Perrier, A., 1986. HAPEX-MOBILHY: a hydrologicalatmospheric experiment for the study of water budget and evaporation fluxat the climate scale. Bull. Am. Meteor. Soc. 67, 138–144.

Ball, J., Woodrow, I., Berry, J., 1987. A model predicting stomatal conductance and itscontribution to the control of photosynthesis under different environmentalconditions. In: Biggins, J. (Ed.), Progress in Photosynthesis Research, vol. 4,MartinusNijhoff, pp. 221–224.

Barr, A.G., King, K.M., Gillespie, T.J., Hartog, G., Neumann, H.H., 1994. A comparisonof Bowen ratio and eddy correlation sensible and latent heat flux measurementsabove deciduous forest. Boundary Layer Meteorol. 71, 21–41.

Betts, A.K., Chen, F., Mitchell, K., Janjic, Z., 1997. Assessment of land surface andboundary layer models in two operational versions of the NCEP Eta Model usingFIFE data. Mon. Wea. Rev. 125, 2896–2915.

Blanken, P.D., Black, T.A., Yang, P.C., Newmann, H.H., Nesic, Z., Staebler, R., denHartog, G., Novak, M.D., Lee, X., 1997. Energy balance and canopy conductanceof a boreal aspen forest: partitioning overstory and understory components. J.Geophys. Res. 102 (D24), 28915–28928.

Bosilovich, M.G., Chen, J., Robertson, F.R., Adler, R.F., 2008. Evaluation of globalprecipitation in reanalyses. J. Appl. Meteor. Climatol. 47, 2279–2299.




Charusombat, U., Niyogi, D., Kumar, A., Wang, X., Chen, F., Guenther, A., Turnipseed,A., Alapaty, K., 2010. Evaluating a new Deposition Velocity Module in the Noahland surface model. Boundary Layer Meteorol. 137, 271–290.

Chen, F. et al., 1996. Modeling of land surface evaporation by four schemes andcomparison with FIFE observations. J. Geophys. Res. 101, 7251–7268.

Chen, F., Dudhia, J., 2001. Coupling an advanced land surface-hydrology model withthe Penn State-NCAR MM5 modeling system. Part I: model implementation andsensitivity. Mon. Wea. Rev. 129, 569–585.

Chen, F., Manning, K.W., LeMone, M.A., Trier, S.B., Alfieri, J.G., Roberts, R., Wilson, J.,Tewari, M., Niyogi, D., Horst, T.W., Oncley, S.P., Basara, J.B., Blanken, P.D., 2007.Evaluation of the characteristics of the NCAR High-Resolution Land DataAssimilation System during IHOP_2002. J. Appl. Meteorol. Clim. 46, 694–713.

Chen, F., Zhang, Y., 2009. On the coupling strength between the land surface and theatmosphere: from viewpoint of surface exchange coefficients. Geophys. Res.Lett., 36, L10404. doi:10.1029/2009GL037980.

Chen, Y., Yang, K., Zhou, D., Qin, J., Guo, X., 2010. Improving the Noah Land SurfaceModel in arid regions with an appropriate parameterization of the thermalroughness length. J. Hydrometeor. 11, 995–1006.

Collatz, G.J., Ball, J.T., Grivet, C., Berry, J.A., 1991. Physiological and environmentalregulation of stomatal conductance. Agric. Forest Meteorol. 54, 561–570.

Collatz, G.J., Ribas-Carbo, M., Berry, J.A., 1992. A coupled photosynthesis – stomatalconductance model for leaves of C4 plants. Aust. J. Plant Physiol. 19, 519–538.

Davis, C. et al., 2008. Prediction of landfalling hurricanes with the AdvancedHurricane WRF model. Mon. Wea. Rev. 136, 1990–2005.

Dugas, W.A., Fritschen, L.J., Gay, L.W., Held, A.A., Matthias, A.D., Reicosky, D.C.,Steduto, P., Steiner, J.L., 1991. Bowen ratio, eddy correlation, and portablechamber measurements of sensible and latent heat flux over irrigated springwheat. Agric. For. Meteorol. 56, 1–20.

Ek, M., Mahrt, L., 1991. OSU 1-D PBL Model User’s Guide, Department ofAtmospheric Sciences. Oregon State University, Corvallis, Oregon pp. 97331–2209.

Ek, M.B., Mitchell, K.E., Lin, Y., Rogers, E., Grunmann, P., Koren, V., Gayno, G., Tarpley,J.D., 2003. Implementation of Noah land surface model advances in the NationalCenters for Environmental Prediction operational mesoscale Eta Model. J.Geophys. Res. 108, 8851.

Farquhar, G.D., Sharkey, T.D., 1982. Stomatal Conductance and Photosynthesis.Annu. Rev. Plant Physiol. 33, 317–345.

Gopalakrishnan, S., Liu, Q., Marchok, T., Sheinin, D., Surgi, N., Tuleya, R., Yablonsky,R., Zhang, X., 2010. In: L. Bernardet (Ed.), Hurricane Weather and Research andForecasting (HWRF) Model scientific documentation, pp. 75.

Goulden, M.L., Daube, B.C., Fan, S.M., Sutton, D.J., Bazzaz, A., Munger, J.W., Wofsy,S.C., 1997. Physiological responses of a black spruce forest to weather. J.Geophys. Res. 102 (D24), 28987–28996.

Grell, G., Peckham, S., Schmitz, R., McKeen, S., Frost, G., Skamarock, W., Eder, B.,2005. Fully coupled ‘online’ chemistry within the WRF model. Atmos. Environ.39, 6957–6975.

Hatsushika, H., Tsutsui, J., Fiorino, M., Onogi, K., 2006. Impact of wind profileretrievals on the analysis of tropical cyclones in the JRA-25 reanalysis. J. Meteor.Soc. Japan 84, 905.

Holt, T., Niyogi, D.S., Chen, F., Manning, K., LeMone, M., et al., 2006. Effect of land–atmosphere interactions on the IHOP 24–25 May 2002 convection case. Mon.Wea. Rev. 134, 113–133.

Houser, P.R., Shuttleworth, W.J., Famiglietti, J.S., Gupta, H.V., Syed, K.H., Goodrich,D.C., 1998. Integration of soil moisture remote sensing and hydrologic modelingusing data assimilation. Water Resour. Res. 34, 3405–3420.

Ingwersen, J., Steffens, K., Högy, P., Warrach-Sagi, K., Zhunusbayeva, D., Poltoradnev,M., Gäbler, R., Wizemann, H.-D., Fangmeier, A., Wulfmeyer, V., Streck, T., 2011.Comparison of Noah simulations with eddy covariance and soilwater measurements at a winter wheat stand. Agri. Forest Meteorol. 151,345–355.

Jackson, T.J., 1993. Measuring surface soil moisture using passive microwaveremote sensing. Hydrol. Process. 7, 139–152.

Kumar, S.V., Peters-Lidard, C.D., Eastman, J.L., Tao, W., 2008. An integrated high-resolution hydrometeorological modeling testbed using LIS and WRF. Environ.Modell. Softw. 23, 169–181.

Kutsch, W.L., Aubinet, M., Buchmann, N., Smith, P., Osborne, B., Eugster, W.,Wattenbach, M., Schrumpf, M., Schulze, E.D., Tormellieri, E., Ceschia, E.,Bernhofer, C., Béziat, P., Carrara, A., Di Tommasi, P., Grünwald, T., Jones, M.,Magliulo, V., Marloie, O., Moureaux, C., Olioso, A., Sanz, M.J., Saunders, M.,Søgaard, H., Ziegler, W., 2010. The net biome production of full crop rotations inEurope. Agric. Ecosyst. Environ. 139, 336–345. doi:10.1016/j.agee.2010.07.016.


LeMone, M.A., Tewari, M., Chen, F., Alfieri, J., Niyogi, D., 2008. Evaluation of the Noahland-surface model using data from a fair-weather IHOP_200 day withheterogeneous surface fluxes. Mon. Wea. Rev. 136, 4915–4941.

Leung, L.R., Kuo, Y., Tribbia, J., 2006. Research needs and directions of regionalclimate modeling Using WRF and CCSM. Bull. Am. Meteor. Soc. 87, 1747–1751.

Mahrt, L., 1998. Flux sampling errors for aircraft and towers. J. Atmos. OceanicTechnol. 15, 416–429.

Mauder, M., Foken, T., Clement, R., Elbers, J.A., Eugster, W., Grunwald, T., Heusinkveld,B., Kolle, O., 2008. Quality control of CarboEurope flux data – Part 2: inter-comparison of eddy-covariance software. Biogeosciences 5, 451–462.

McCaughey, J.H., Lafleur, P.M., Joiner, D.W., Bartlett, P.A., Costello, A. M., Jelinski,D.E., Ryan, M.G., 1997. Magnitudes and seasonal patterns of energy, water, andcarbon exchanges at a boreal young jack pine forest in the BOREAS northernstudy area, J. Geophys. Res. 102 (D24), 28–29, 997–1007.

Mitchell, K.E., Lohmann, D., Houser, P.R., Wood, E.F., Schaake, J.C., Robock, A.,Cosgrove, B.A., Sheffield, J., Duan, Q., Luo, L., Higgins, R.W., Pinker, R.T., Tarpley,J.D., Lettenmaier, D.P., Marshall, C.H., Entin, J.K., Pan, M., Shi, W., Koren, V.,Meng, J., Ramsay, B.H., Bailey, A.A., 2004. The multi-institution North AmericanLand Data Assimilation System (NLDAS): utilizing multiple GCIP products andpartners in a continental distributed hydrological modeling system. J. Geophys.Res.109, D07S90. doi:10.1029/2003JD003823.

Nie, D., Kanemasu, E.T., Fritschen, L.J., Weaver, H.L., Smith, E.A., Verma, S.B., Field,R.T., Kustas, W.P., Stewart, J.B., 1992. An intercomparison of surface energy fluxmeasurement systems used during FIFE 1987. J. Geophys. Res. 97, 18715–18724.

Niu, G.-Y., Yang, Z.-L., Mitchell, K.E., Chen, F., Ek, M.B., Barlage, M., Longuevergne, L.,Kumar, A., Manning, K., Niyogi, D., Rosero, E., Tewari, M., Xia, Y., 2011. Thecommunity Noah land surface model with multi-parameterization options(Noah-MP): 1. Model description and evaluation with local-scalemeasurements, J. Geophys. Res., 116, D12, D12109. doi:10.1029/2010JD015139.

Niyogi, D., Raman, S., 1997. Comparison of four different stomatal resistanceschemes using FIFE observations. J. Appl. Meteorol. 36, 903–917.

Niyogi, D., Alapaty, K., Raman, S., Chen, F., 2009. Development and evaluation of acoupled photosynthesis based gas exchange evapotranspiration model (GEM)for mesoscale weather forecasting applications. J. Appl. Meteorol. Climatol. 48,349–368.

Noilhan, J., Planton, S., 1989. A simple parameterization of land surface processesfor meteorological models. Mon. Wea. Rev. 117, 536–549.

Onogi, K., Tsutsui, J., Koide, H., Sakamoto, M., Kobayashi, S., Hatsushika, H.,Matsumoto, T., Yamazaki, N., Kamahori, H., Takahashi, K., Kadokura, S., Wada,K., Kato, K., Oyama, R., Ose, T., Mannoji, N., Taira, R., 2007. The JRA-25Reanalysis. J. Meteorol. Soc. Jpn. 85, 369–432.

Pan, H.-L., Mahrt, L., 1987. Interaction between soil hydrology and boundary-layerdevelopment. Bound.-Layer Meteor. 38, 185–202.

Reichstein, M., Falge, E., Baldocchi, D., Papale, D., Valentini, R., Aubinet, M., Berbigier,P., Bernhofer, C., Buchmann, N., Gilmanov, T., Granier, A., Grünwald, T.,Havránková, K., Janous, D., Knohl, A., Laurela, T., Lohila, A., Loustau, D.,Matteucci, G., Meyers, T., Miglietta, F., Ourcival, J.M., Rambal, S., Rotenberg, E.,Sanz, M., Seufert, G., Vaccari, F., Vesala, T., Yakir, D., 2005. On the separation ofnet ecosystem exchange into assimilation and ecosystem respiration: reviewand improved algorithm, Glob. Change Biol. 11, 1424–1439.

Rodell, M., Houser, P.R., Jambor, U., Gottschalck, J., Mitchell, K., Meng, C.-J.,Arsenault, K., Cosgrove, B., Radakovich, J., Bosilovich, M., Entin, J.K., Walker,J.P., Lohmann, D., Toll, D., 2004. The Global Land Data Assimilation System. Bull.Am. Meteor. Soc. 85, 381–394.

Sellers, P.J., Los, S.O., Tucker, C.J., Justice, C.O., Dazlich, D.A., Collatz, G.J., Randall,D.A., 1996. A revised land surface parameterization (SiB2) for atmosphericGCMs-part II: the generation of global fields of terrestrial biophysicalparameters from satellite data. J. Climate 9, 706–737.

Skamarock, W.C., Klemp, J.B., Dudhia, J., Gill, D.O., Barker, D.M., Wang, W., Powers,J.G., 2005. A description of the Advanced Research WRF version 2. NCAR Tech.Note NCAR/TN-468+STR, 88 pp.

Thorp, K.R., DeJonge, K.C., Kaleita, A.L., Batchelor, W.D., Paz, J.O., 2008. Methodologyfor the use of DSSAT models for precision agriculture decision support. Comput.Electron. Agri. 64, 276–285.

Twine, T.E., Kustas, W.P., Norman, J.M., Cook, D.R., Houser, P.R., Meyers, T.P.,Prueger, J.H., Starks, P.J., Wesely, M.L., 2000. Correcting eddy-covariance fluxunderestimates over a grassland. Agric. Forest Meteorol. 103, 279–300.

Wu, Z., Wang, X., Chen, F., Turnipseed, A.A., Guenther, A.B., Niyogi, D., Charusombat,U., Xia, B., Munger, J.W., Alapaty, K., 2011. Evaluating the calculated drydeposition velocities of reactive nitrogen oxides and ozone from two communitymodels over a temperate deciduous forest. Atmos. Environ. 45, 2663–2674.


http://dx.doi.org/10.1029/2009GL037980

http://dx.doi.org/10.1016/j.agee.2010.07.016

http://dx.doi.org/10.1029/2003JD003823

http://dx.doi.org/10.1029/2010JD015139


computers and electronics in agriculture · and albedo feedback; however the key feature was the...

Documents