remote sensing

Article
A One-Source Approach for Estimating Land Surface
Heat Fluxes Using Remotely Sensed Land
Surface Temperature
Yongmin Yang 1,2 , Jianxiu Qiu 3, *, Hongbo Su 4,5, *, Qingmei Bai 6 , Suhua Liu 4,7 , Lu Li 4,7 ,
Yilei Yu 8,9 and Yaoxian Huang 3
1 State Key Laboratory of Simulation and Regulation of Water Cycle in River Basin, Beijing 100038, China;
yangym@iwhr.com
2 Research Center on Flood and Drought Disaster Reduction of the Ministry of Water Resources,
China Institute of Water Resources and Hydropower Research, Beijing 100038, China
3 Guangdong Provincial Key Laboratory of Urbanization and Geo-simulation,
School of Geography and Planning, Sun Yat-sen University, Guangzhou 510275, China; ymszr@163.com
4 Key Laboratory of Water Cycle & Related Land Surface Processes, Institute of Geographic Sciences and
Natural Resources Research, Beijing 100101, China; liulin557@163.com (S.L.); a_sunbeam@126.com (L.L.)
5 Department of Civil, Environmental and Geomatics Engineering, Florida Atlantic University, Boca Raton,
FL 33431, USA
6 Xi’an Meteorological Bureau, Xi’an 710016, China; melissa1027@163.com
7 Graduate University of Chinese Academy of Sciences, Beijing 100049, China
8 Institute of Wetland Research, Chinese Academy of Forestry, Beijing 100091, China; yuyilei1222@126.com
9 Beijing Key Laboratory of Wetland Services and Restoration, Beijing 100091, China
* Correspondence: qiujianxiu@mail.sysu.edu.cn (J.Q.); suh@fau.edu (H.S.); Tel.: +86-208-411-2512 (J.Q.)

Academic Editors: Qiusheng Wu, Charles Lane, Melanie Vanderhoof, Chunqiao Song, Zhaoliang Li and
Prasad S. Thenkabail
Received: 22 October 2016; Accepted: 28 December 2016; Published: 6 January 2017

Abstract: The partitioning of available energy between sensible heat and latent heat is important
for precise water resources planning and management in the context of global climate change.
Land surface temperature (LST) is a key variable in energy balance process and remotely sensed LST
is widely used for estimating surface heat fluxes at regional scale. However, the inequality between
LST and aerodynamic surface temperature (Taero ) poses a great challenge for regional heat fluxes
estimation in one-source energy balance models. To address this issue, we proposed a One-Source
Model for Land (OSML) to estimate regional surface heat fluxes without requirements for empirical
extra resistance, roughness parameterization and wind velocity. The proposed OSML employs both
conceptual VFC/LST trapezoid model and the electrical analog formula of sensible heat flux (H) to
analytically estimate the radiometric-convective resistance (rae ) via a quartic equation. To evaluate the
performance of OSML, the model was applied to the Soil Moisture-Atmosphere Coupling Experiment
(SMACEX) in United States and the Multi-Scale Observation Experiment on Evapotranspiration
(MUSOEXE) in China, using remotely sensed retrievals as auxiliary data sets at regional scale.
Validated against tower-based surface fluxes observations, the root mean square deviation (RMSD)
of H and latent heat flux (LE) from OSML are 34.5 W/m2 and 46.5 W/m2 at SMACEX site and
50.1 W/m2 and 67.0 W/m2 at MUSOEXE site. The performance of OSML is very comparable to
other published studies. In addition, the proposed OSML model demonstrates similar skills of
predicting surface heat fluxes in comparison to SEBS (Surface Energy Balance System). Since OSML
does not require specification of aerodynamic surface characteristics, roughness parameterization
and meteorological conditions with high spatial variation such as wind speed, this proposed method
shows high potential for routinely acquisition of latent heat flux estimation over heterogeneous areas.

Remote Sens. 2017, 9, 43; doi:10.3390/rs9010043 www.mdpi.com/journal/remotesensing

Remote Sens. 2017, 9, 43 2 of 25

Keywords: One-Source Model for Land (OSML); evapotranspiration; land surface heat flux; land
surface temperature; SMACEX; MUSOEXE

1. Introduction
Land surface heat fluxes are essential components of water and energy cycles interactions in
the hydrosphere, atmosphere and biosphere [1–6]. The energy balance and its partitioning between
sensible heat flux (H) and latent heat flux (LE) are extremely important for understanding global
climate change and land-atmosphere interaction [7]. Satellite remote sensing provides a promising
way to obtain information on both LE and H fluxes at larger spatiotemporal scales than models depend
on in-situ meteorological observations alone [8–10]. Considerable progress has been made in the
development of global flux products from space observations [11–16].
Remotely sensed land surface temperature (LST) is widely used in evapotranspiration (ET)
modeling, and it is a key variable for partitioning available energy into H and LE [9,17]. LST is
often referred to as “radiative surface temperature” [18]. Because of the inequality between LST and
aerodynamic surface temperature (Taero ), one-source and two-source surface energy balance models
have been proposed with various expressions for aerodynamic resistance, in order to estimate the
sensible heat loss based on difference between surface and air temperature [19–23]. Much effort has
been made to compensate for the difference between LST and Taero in one-source models, most of
which was concentrated on making corrections by: (1) adjusting temperature based on empirical
relationship between LST and Taero [19,21,23]; and (2) adjusting the aerodynamic resistance roughness
length for heat or the kB−1 parameter, or by including empirical “extra resistance” (rex ) factors [24,25].
For instance, Chehbouni et al. [20] and Matsushima [22] proposed methods to interpret the relationship
between LST and Taero , and make further correction on LST based on this relationship. However, due to
the high empirical dependency of such models and the non-uniqueness in the relationship between LST
and Taero , these approaches generally produced significant errors when applied to partially-vegetated
landscapes [26,27]. Other works were based on adjustments of the roughness length for heat or the
kB−1 parameter, or by including empirical “extra resistance” factors [23,25,28,29]. However, the high
empirical dependency of the model parameterization on aerodynamic characteristics introduces great
uncertainty in H estimation [9,10,17,24]. Extensive reviews on the methodologies of compensation
for the difference between LST and Taero can be found in the related works [17,22,24,30]. Two-source
models, which eliminate the need for extra resistance and treat H as the sum of sensible heat fluxes from
both soil surface and vegetation, were developed by Norman et al. [26]. However, various aerodynamic
resistances need to be parameterized in two-source models, and the soil surface aerodynamic resistance
to heat transfer is the most difficult parameter to be characterized. Extensive reviews of satellite-based
methodologies on surface heat fluxes estimation can be found in the related works [10,17,30].
Lacking information on regional wind speed and aerodynamic surface characteristics currently
poses a major challenge for regional surface heat fluxes estimation [17,31]. In particular, wind velocity at
regional scale is highly variable and is difficult to estimate from in-situ observations [32,33]. In addition,
estimation of the aerodynamic resistance usually requires stability corrections, as well as determination of
parameters including suitable roughness lengths and zero displacement level [25,34–36]. Boegh et al. [31]
pointed out that aerodynamic resistance is usually estimated from local data, although it must be
noted that areal average of roughness length from point-scale observation is highly unreliable due to
its nonlinearity. Several studies have been carried out in an attempt to estimate surface energy fluxes
without having to calculate various aerodynamic resistances. Such attempt includes the proposal of the
spatial contextual models based on the LST and vegetation fractional cover (VFC) space (hereinafter
referred to as VFC/LST space), and a key advantage of this category of approaches is their relative
independence from site-specific tuning of resistances-related parameters [36–41]. However, a large
number of pixels with a wide range of soil wetness and fractional cover are required in such models.

In order to mediate the difference between Ts and Taero . 2. The alternative approach is to solve LE without the need for estimation of the aerodynamic resistance. ◦ C). H is generally calculated using the electrical analog: Taero − Ta Ts − Ta Ts − Ta H = ρCp = ρCp = ρCp (2) ra ra + rex rae where ρ is the air density (kg/m3 ).e. Section 6 provides a conclusion. [31] termed rae as the “atmospheric resistance” [31]. s/m) by Norman et al. Section 4 reports the LE and H fluxes validation results from OSML at two experimental regions with contrasting climatic features and different surface complexities. Taero is the aerodynamic temperature (◦ C) which is rather difficult to estimate over heterogeneous surface and cannot be directly obtained from remote sensing retrievals. The key issue of one-source model is to accurately estimate H. LE is the latent heat flux (W/m2 ) and G is the soil heat flux (W/m2 ). namely. which integrates the LST into the Penman–Monteith equation for estimating the terrestrial surface energy balance fluxes. [31] proposed a novel method that eliminates the need for introducing the parameter of excess resistance and used three equations to estimate atmospheric resistance and surface resistance. However.29]. kB−1 is often empirically determined by local calibration and the erratic values of kB−1 for sparse vegetation can . 43 3 of 25 They also involve subjectivity in the determination of so-called “dry edge” and “wet edge” [42]. named Surface Temperature Initial Closure (STIC). which served as a benchmark for evaluating OSML in this study. Boegh et al. The objective of this study is twofold: (1) to develop a One-Source Model for Land (OSML) for estimating land surface fluxes which circumvents the difficulties of parameterizing extra resistance and acquiring regional-scale wind velocity. OSML performance is also compared with SEBS in estimated surface fluxes. Boegh et al.44].1 and 2..3 introduces the commonly used one-source model (SEBS (Surface Energy Balance System)). on the contrary. [31]). 9.Remote Sens. [43] proposed a new method. Ta is the air temperature (◦ C). hs. In one-source model rex is often calculated using the kB−1 approach [25. which can be estimated using remote sensing retrievals [25. Remotely sensed land surface temperature (Ts . Theoretical Background of One-Source Model The energy balance equation can be formulated for the whole canopy-soil system: Rn = H + LE + G (1) where Rn is the net radiation flux (W/m2 ). Sections 2.1. The sum of ra and rex is termed as radiometric-convective resistance (rae .2 present a theoretical background and detailed description of the OSML model. The variable ra is aerodynamic resistance (s/m). In one-source model. The difference between Ts and Taero may differ by several degrees. and Cp is the specific heat of air at constant pressure (J/kg·◦ C). is more easily obtained and is used as a replacement for Taero .max in Boegh et al. and then LE is calculated as the residual term in the energy balance equation. Section 5 discusses the advantages and limitations of this work. However. Model Description 2. this method needs a modification of the surface humidity parameter (i. and (2) to evaluate the performance of the proposed OSML model by validation with in-situ surface fluxes observations and comparison with SEBS in terms of estimating both surface fluxes and radiometric-convective resistance. and Section 2. H is the sensible heat flux (W/m2 ). an extra resistance (rex ) is added in the denominator (Equation (2)). Mallick et al. [26]. For instance. 2017. Soil Moisture-Atmosphere Coupling Experiment (SMACEX) site and Multi-Scale Observation Experiment on Evapotranspiration (MUSOEXE) site. Section 3 introduces the remotely-sensed data sets used for driving OSML and the eddy covariance system measurements used for assessing the model.

[45]. [45] and Boegh et al. [31]. Ep is the potential evaporation and is computed using the Priestley–Taylor equation: ∆ Ep = 1. [31]. Ta . For a given pixel. and b is the temperature difference between Point M and the dry edge. Full reviews of the kB−1 approaches on one-source model can be found in Paul et al. (Tc − Ta )B = [(rae (Rn − G))/(ρa Cp )][γ(1 + rcx /rae )/(∆ + γ(1 + rcx /rae ))] (11) − [VPD/(∆ + γ(1 + rcx /rae ))] where rcp is the canopy resistance for full cover well-watered vegetation (s/m). the four theoretical extreme points can be derived based on the values of Rn .Remote Sens. [24] and Boegh et al. Values of rcp and rcx could be parameterized following Jarvis-type representation of canopy resistance as follows [46–49]: rcp = rcmin F1−1 F2−1 F3−1 F− 1 cp /LAI (12) rcx = rcmin F1−1 F2−1 F3−1 Fcx −1 /LAI (13) . [45] introduced a water deficit index (WDI) based on the VFC/LST trapezoid space and extended the application of the crop water stress index (CWSI) from fully. H can be estimated using the following approach: H = (Rn − G) − LE = (Rn − G) − (1 − WDI)Ep (3) a WDI = (4) a+b a = (Ts − Ta ) − fc (Ts − Ta )C + (fc − 1)(Ts − Ta )D (5) b = (1 − fc )(Ts − Ta )A + fc (Tc − Ta )B − (Ts − Ta ) (6) where a is the temperature difference between Point M (Figure 1) and the cold edge. That is. Development of One-Source Model for Land (OSML) In this study.to partially-vegetated surface areas.26 (Rn − G) (7) ∆+γ where ∆ is the slope of the saturation vapor pressure curve at air temperature Ta (kPa·◦ C−1 ). vapor pressure deficit (VPD) and rae following the equations from Moran et al. 9. and rcx is the canopy resistance for full cover vegetation with no available water (s/m). for the dry bare soil of Point A. 43 4 of 25 lead to large uncertainties of one-source model when applied to heterogeneous surface. Moran et al. the difference takes the form of: (Ts −Ta )D = [(rae (Rn − G))/(ρa Cp )][γ/(∆ + γ)] − [VPD/(∆ + γ)] (9) For the well-watered full-cover vegetation of Point C. 2.2. G. 2017. the parameterization scheme of rae is inspired by the work of Moran et al. (Ts − Ta )C = [(rae (Rn − G))/(ρa Cp )][γ(1 + rcp /rae )/(∆ + γ(1 + rcp /rae ))] (10) − [VPD/(∆ + γ(1 + rcp /rae ))] For the water-stressed full-cover vegetation of Point B. and γ is the psychrometric constant (kPa·◦ C−1 ). evaporation equals zero and the difference between LST and Ta can be derived as follows: (Ts − Ta )A = rae (Rn − G)/ρa Cp (8) For the saturated bare soil of Point D where the surface resistance equals zero. Based on the WDI.

F2 and F3 are taken from Dickinson et al. 43 5 of 25 describes the influence of vapor pressure deficit on canopy resistance and F3 describe the influence of air temperature on canopy resistance. cd is the vapor pressure deficit causing stomatal closure. G. Fcp takes the value 1+ f of 1. Remote Sens. where rcmaxand R SL is a limitvalue is the maximum valueofofthe 100 canopy for crops. A sketch of the trapezoid in the vegetation fractional cover and LST (VFC/LST) space. 43 5 of 25 Remote Sens.ofEstimation vapor pressure schemes deficit for Fon1. (Figure 2). [46].0 F1 = f = 0. Figure 1. Figure 1.0. The estimated rae are used to are performed using pixel-based LST.0016T a RS↓ is the downward shortwave (17) ↓ radiation. A sketch of the trapezoid in the vegetation fractional cover and LST (VFC/LST) space. Equations (2)–(11) are combined to solve rae (please see Appendix A for details). stomatal c ofis100 a limit value W/m about vp d for crops. and consequently.0. 2017.551 +Sf ↓ (15) (14) RSL f + (rcmin /rLAI cmax ) ↓ F2 = 1. [50] as:factors Fcp and Fcx take into account the effect of water stress on canopy resistance. G. Noihan and Planto [48] and Wigmosta et al. Estimation schemes for F1 . and VPD. the vapor pressure downward deficit shortwave causing radiation.55 (15) R↓SL LAI F3 = 0. 2017.08T 2 resistance. compute H. The factors Fcp and Fcx take into account the effect of water where stressron and rcmax cmincanopy is the minimum resistance.08Ta − 0. LE can be solved as the residual term in the energy balance equation (Equation (1)). cvpd is about 4 kPa. The Equations (2)–(11) are combined to solve rae (please see Appendix A for details).0 − VPD/cvpd (16) where rcmax is the maximum value Fof 3 =the canopy 0.05. Ta. . Noihan of and air temperature Planto [48] andon canopy Wigmostaresistance.0 (16) f = 0. The computations computations are performed using pixel-based LST. F2canopy resistance and F3 are taken fromand Dickinson F3 describeetthe influence al. The model inputs (from meteorological observations and remote sensing retrievals). The et al. Fcpfactor takes Fthe 1 describes value ofthe 1. LE can be solved as the residual term in the energy balance equation and consequently.05. For canopy resistance for full coverF1vegetation= with no available water. The vegetation. The estimated rae are used to compute H. LAI is leaf area cover well-watered index. Rc↓Sd isisthe W/m2resistance. [46]. a − 0. Rn . ↓ 2 and RSL is closure.influence For of canopy the photosynthetically resistance for full active cover radiation vegetation on with canopy no resistance. The model inputs (from meteorological observations and remote sensing retrievals). For canopy resistance for full cover well-watered vegetation. 9. 4 kPa. available water. [50] as: R ↓ 2. Rn. and VPD. 9. Ta .0016Ta2 (17) F2 = 1. For canopy value of the canopy resistance for fullresistance. Fcx takes the value of (14) f + (rcmin / rcmax ) 0. (Equation (1)).0 −RVPD / cvpd S 2. FF2 describes cx takes the the influence value of 0. intermediate intermediate variables and the main procedures for implementation of OSML are summarized in a variables and the main procedures for implementation of OSML are summarized in a flow chart flow chart (Figure 2).

and momentum frictionrespectively.θa = [ln ( ρC)p-u3∗ψ θvh ( ) + ψ h ( o h )] (19) k u *ρ C p L =z − oh kgH L L (20) ρ C p u *3 θ v(m/s). while light-gray rectangles represent intermediate variables or parameters.3. setsH[25]. respectively. In SEBS. 43 6 of 25 Remote Sens. Surface Energy Balance System (SEBS) 2.ψm ( ) + ψ m ( o m )] (18) k θ −θ = Hz o m z − d0 [ln( L z − d0 ) + ψ ( zLoh )] )−ψ ( (19) 0 a h h ku∗ ρCp zoh L L H z-d 0 z-d 0 z θ0 . 2 (m/s z). and g is d0 thezero acceleration due to gravity (m). 9. and the dark-gray rectangles represent the main procedures for solving rae and consequently estimating surface heat fluxes. 2. 2017. 9. L is the Obukhov length (m). In SEBS. ψ m and ψh are the stability correction functions for sensible heat and momentum transfer. In this iteratively is calculated study. k isObukhov L is the the vonlength Kaman’s constant. above the surface (m). Remote Sens. [25]. (m). respectively. u∗ is the friction velocity is the zero plane displacement (m). θ0 and θa are the potential temperatures at surface and at height z (K). z isLthe=height. In SEBS. H respectively. where diamond-shaped boxes represent OSML input from meteorological observations and remote sensing retrievals. the roughness height for heat transfer is calculated using a model for calculating kB−1 [25. velocity (m/s). u * is thetransfer. d0 where u is the wind speed (m/s). and zom and zoh are the(20) roughness heights for momentum and heat transfer k g(m). where diamond-shaped Figure 2.51]. bare soil conditions and mixed . SEBS servesthree by solving as anonlinear benchmark for evaluating OSML equations: performance. respectively. and g is the acceleration due to gravity (m/s2). while boxes represent OSML input from meteorological observations and remote sensing retrievals. ψm and ψh are the stability correction where u is theforwind functions speed sensible heat(m/s). θ0 and θa are the potential temperatures at surface and at height z (K). and the dark-gray rectangles light-gray rectangles represent intermediate variables or parameters. A weighted average between limiting cases of full canopy. H is calculated iteratively by solving three nonlinear equations: u∗ z − d0 z − d0 zom u = [ln( ) − ψm ( ) + ψm ( )] (18) u k z-d zom z-d L L z 0 0 u = * [ln ( ) . Surface Energy Balance System (SEBS) The SEBS model was developed for estimating regional surface heat fluxes by using satellite The SEBS remote model sensing datawas setsdeveloped forstudy. represent the main procedures for solving rae and consequently estimating surface heat fluxes. In this estimating regional SEBS serves as asurface heat for benchmark fluxes by using evaluating satellite OSML remote sensing data performance.is the height above the surface (m). 43 6 of 25 Figure 2.3. respectively. k is the von Kaman’s constant. and zom and zoh are the is the plane displacement roughness heights for momentum and heat transfer (m). Flow chart of One-Source Model for Land (OSML) implementation. 2017. Flow chart of One-Source Model for Land (OSML) implementation.

3. emissivity. water vapor pressure and wind speed.4) (23) where Re∗ is the roughness Reynolds number. including albedo. fc is vegetation fractional cover.1. the radiometric-convective resistance of SEBS (rae_SEBS ) is calculated as: Ts −Ta rae_SEBS = ρCp (24) HSEBS where HSEBS is the sensible heat flux derived from the SEBS module. 2017. and the other was on 1 July 2002 (DOY 182) from the Enhanced Thematic Mapper plus (ETM+) on Landsat 7.1. Two Landsat scenes were used: one from the Landsat-5 Thematic Mapper (TM) on 23 June 2002 (Day of Year (DOY) 174). [52]. The extensive measurements of atmospheric states. The two Landsat scenes were obtained from the National Snow and Ice Data Center (http://nsidc. The second set includes meteorological variables. soil moisture and other land surface states on the turbulent flux exchange with the atmosphere. and kBs−1 is the value for bare soil surface. and LAI. In this study. including downward shortwave radiation and downward longwave radiation. we briefly introduced the SEBS algorithm for H estimation. The third set is atmospheric radiation fluxes. Ct∗ is the heat transfer coefficient of the soil. 3. 9.Remote Sens. the roughness height for heat transfer is calculated using a model for calculating kB−1 [25. LST. Remotely Sensed and In-Situ Observations Required for OSML Input over SMACEX The required in-situ meteorological forcing and satellite-based retrievals for OSML and SEBS are listed in Table 1. A full description of the SMACEX experiment can be found in Kustas et al. .1.org/index.46(Re∗ )1/4 − ln(7.54].1. n is the wind speed profile extinction within the canopy. the SEBS module in ILWIS (Integrated Land and Water Information System) was adopted. such as air temperature. u(h) is the horizontal wind speed at the canopy height. bare soil conditions and mixed vegetation is implemented based on vegetation fractional cover [28. The first set is satellite-based land surface variables.html). vegetation conditions. Study Area and Data Processing 3. and the detailed description of the model can be found in Su [51]. SMACEX Campaign 3. Description of SMACEX Campaign Area and Related Experiment SMACEX The Soil Moisture-Atmosphere Coupling Experiment (SMACEX) was conducted in conjunction with the Soil Moisture Experiment 2002 (SMEX02) during June and July 2002 near Ames. In this study. calculated as: kBs−1 = 2. soil properties and surface fluxes allow for rigorous analysis and the validation of land surface model and remote sensing observations in the study area [53. The land cover in the experimental area is primarily corn and soybean.2. A weighted average between limiting cases of full canopy. Iowa. fs is the fractional cover for soil. Ct is the heat transfer coefficient. In order to compare the differences between SEBS and OSML.51]: zom zoh = −1 (21) exp(kBSEBS ) kCd k uu(h∗ ) zom h −1 kBSEBS = f2 + f2c f2s + kBs−1 f2s (22) 4Ct uu(h∗ ) (1 − e −n/2 c ) Ct∗ where Cd is the drag coefficient of the foliage. 43 7 of 25 In SEBS.51]. The SMACEX was designed to provide direct measurement/remote-sensing/modeling approaches for understanding the impact of spatial and temporal variability in vegetation cover. SEBS requires three sets of input data.

Remote Sens.3NDWI)) (30) hsoybean = (0. Empirical relationships between the Landsat-derived vegetation index and LAI. LAI was derived using the following equation: LAI = (2. as suggested by Li et al. Albedo was retrieved from the visible and near-infrared bands (1–5. In order to fairly compare the performance of OSML and SEBS. The net radiation Rn is estimated as: Rn = (1 − α)Swd + εδ(εa (Ta + 273.26)(1 + 0. the same net radiation (Rn ) and soil heat flux (G) were used for the two models. [56]. The vegetation fractional cover was estimated from the normalized difference vegetation index (NDVI) following: NDVI − NDVImin k fc = 1 − ( ) (25) NDVImax −NDVImin where NDVImax and NDVImin are NDVI values for complete vegetation cover and bare soil. spectral radiance and downwelling sky radiance from the atmosphere were provided by Li et al. [58.59]. near-infrared (NIR) bands were re-sampled to be consistent in resolution with images from the thermal infrared band. 2017.15)4 ) (26) where α is the albedo. respectively. [60]. ε is the emissivity. [44]: G = Rn (fc τv +τs (1 − fc )) (27) where τv and τs are the soil heat flux ratios for full vegetation (0.6)(1 + 0.315). [55].html).005 exp(4. these empirical relationships were used to derive LAI and vegetation height for the SMACEX region. band 5) reflectances: NDWI = (NIR − SWIR)/(NIR + SWIR) (29) Crop heights for corn (hcorn ) and soybean (hsoybean ) were derived using the following equations: hcorn = (1.5NDWI)) (31) .2NDWI + 0. 9.org/index.6175 in this study. The coefficient k is a function of the leaf orientation distribution within the canopy. The land cover classification for the study area was obtained from the National Snow and Ice Data Center (http://nsidc. In this study. band 6). Details concerning the atmospheric correction and retrieval of LST were given in Li et al.14)(1 + 0.104 exp(4. The visible. [55] and Li et al.5NDWI + 0. [57] and Allen et al. [56]. The soil heat flux G is parameterized following Choudhury et al. 7) of the Landsat images following Tasumi et al. and δ is the Stefen–Boltzman constant. For both corn and soybean. Swd is the downwelling shortwave radiation (W/m2 ). 43 8 of 25 The LST was derived from the thermal infrared band (TIR. vegetation cover fraction and crop height over the watershed were given by Anderson et al. which can be computed from the near-infrared (NIR.88NDWI + 1. The atmospheric transmittance.04 exp(5. band 4) and shortwave infrared (SWIR.05) and bare soil (0. which is set as 0.1NDWI)) (28) where NDWI is the normalized difference water index.15)4 − (Ts + 273.

1.96 5.2 Wind speed (m/s) × × 6.322. band 6 based on Li et al.3.al. the measured is inherently assumed. six of them are corn sites and six soybean sites.96 834 2.html) Data NDVI Vegetation Fractional Cover Calculated Estimatedwith fromNIR and RED band NDVI Land cover Obtained Albedo from the National was retrieved from Snow and Iceand the visible Datanear-infrared Center (http://nsidc. band 6 based NIR and REDonband Li et al.2 98.71 for the soybean sites and surface energy budget [61–63].67 × Crop height × means not required.2 Atmospheric pressure Satellite-Based (kPa) Retrievals 98. Remote Sens.67 × Crop et al.2 29.html) bands (1–5. [65].1. The land use types of study area and the spatial and soil heat fluxes. 0. It is well recognized that the field measurements spatial distribution of flux towers are shown in Figure 3.2 98. [57] Anderson et al.32 LST Derived from TIR. to Kustas et al. The Figure Figure distribution 3.1 ×Crop height Displacement height (m) × × means ×not required.1 × Crop Displacement height (m) × × Anderson 0. Meteorological forcing and satellite-based observations required for OSML and SEBS for SMACEX site. observed For aratio.please referPrueger [52] and to Kustas et et al. During SMACEX. the energy balance closure was 0. [65]. [60]Center (http://nsidc. DOY174 DOY182 DOY174 DOY182 Meteorological Forcing OSML SEBS Incoming solar radiation (W/m2 ) 834 DOY174 859 DOY182 834 DOY174 859DOY182 Air Meteorological (◦ C) temperatureForcing 29.71 for the 0. [55] Satellite-Based Retrievals LAI Empirical relationship given by Anderson et al.Remote Sens.6 98. which observedthe Bowen ratio.35 98. 2017.2 859 Atmospheric pressure Air temperature (°C) (kPa) 98.96 2.[66].2× 6.2 29. pleaseduringrefer SMACEX. [60] Empirical heightrelationship given by Surface roughness Crop height (m) (m) × × ×× 0. the energy balance closure was 0. [52] and Prueger et al. where energy flux partitioning. Eddy Covariance System Measurements over SMACEX Twelve towers with eddy covariance (EC) systems were deployed in the central area of the experiment.2 29.35 Wind Vaporspeed (m/s) Pressure (kPa) × 1.org/index. Eddy Covariance System Measurements over SMACEX 3.org/index. It is well recognized that the field of heat fluxes using the eddy covariance technique often fail to show closure of the surface energy measurements of heat fluxes using the eddy covariance technique often fail to show closure of the budget [61–63]. Bowen detailed discussion of the flux For a detailed tower observations discussion of the flux and towermeasurement observations uncertainty duringuncertainty and measurement SMACEX. The distributionofofflux fluxtowers in the towers in theWalnut WalnutCreek Creek catchment catchment andand the land the land use classifications.LE theand H wereLEadjusted measured for closure and H were adjustederrors using the for closure approach errors suggested by Twine using the approach et al. balance whereis inherently energy balance assumed.6 29. [60] height Surface roughness (m) × × 0. 43 9 of 25 Table 1.229.3 5.3 1. 7) of the Landsat Albedo Vegetation Fractional Cover Estimated from NDVI images following Tasumi et al.6 98. During SMACEX.2 98. [60] LST NDVI Derived from with Calculated TIR.3.84 soybean for the sites cornandsites [64]. 9.Twelve All towers towerswithwere eddy covariance (EC) instrumented with systems sensorswere deployed to measure in the central turbulent fluxesarea andofsoil the heat experiment.84 In corn for the ordersitesto compare [64]. . Meteorological forcing and satellite-based and SEBS for SMACEX site. 2017. [66]. six of them are corn sites and six soybean sites.96 834 2.2 859 1. suggested whichetconserves by Twine the conserves al. In orderwithto modeled fluxmodeled compare with partitioning.6 29. 3. use classifications. 3. [55] LAI cover Land Empirical Obtainedrelationship given bySnow from the National Anderson and Iceet al. [57] Albedo was retrieved from the visible and near-infrared bands (1–5. All towers were instrumented with sensors to measure turbulent fluxes fluxes. 9. 0.35 Vapor Pressure Incoming (kPa) solar radiation (W/m2) 1.35 29. 7) of the Landsat Albedo Empirical relationship given by Crop height (m) × following Tasumi × images et al. 43 9 of 25 OSML observations required for OSMLSEBS Table 1. The land use types of study area and the distribution of flux towers are shown in Figure 3.

The left figure shows the kernel experimental area in MUSOEXE. Hiwater-MUSOEXE Experiment 3. HiWATER was a four-year four-year program which formally initialized in May 2012. The Experiment distribution of flux(MUSOEXE) on Evapotranspiration towers andover theZhangye land useoasis. Theover (MUSOEXE) Multi-Scale ZhangyeObservation oasis is partExperiment on Evapotranspiration of the HiWATER.5 ×large 5.5 km2 ). and intends to acquire acquire all components and principle parameters related to evapotranspiration processes through all components and principle parameters related to evapotranspiration processes through multi-scale multi-scale observations [68]. 21 eddy covariance system sets.2. [67] and Xu et al. The overall objective HiWATER is to improve the observability of hydrological and ecological processes. HiWATERstudies was a and water recourse management at the basin scale [67]. surfacewe select an additional complexities in the Heihe experimental River Basin region with contrasting climatic features and different surface complexities of northwestern China. residential 4). Description of HiWATER-MUSOEXE Campaign In order to prove the performance of proposed OSML.2.2.RemoteSens. area desert steppe and (site oneGobi #20).MUSOEXE involvedand Site Measurements twoRemotely flux observation matrixes Sensed Data in the for OSML middle Input over reach MUSOEXE of the Heihe River Basin: one large experimental area and one kernel experimental area (5.1. and surfaces of desert (site #19). and the subset figure in the lower right yellow rectangular shows in the left the location figure shows of MUSOEXE the kernel (marked experimental in red triangle) in area in MUSOEXE.5 km2 . The distribution of flux towers and the land use classifications in the Multi-Scale Observation Figure 4. 9. Hiwater-MUSOEXE Experiment 3. and intends to (MUSOEXE) over Zhangye oasis is part of the HiWATER. kernel experimental (site area (5.figure in the lower right shows the location of MUSOEXE (marked in red triangle) in the Heihe River Basin 3. we select an additional experimental In with region ordercontrasting to prove the performance climatic features of and proposed differentOSML. classifications The yellowin the Multi-Scale rectangular in the Observation Experiment on Evapotranspiration (MUSOEXE) over Zhangye oasis. and SiteinMeasurements China. HiWATER-MUSOEXE provides a areas. 43 43 10 of 10 of 25 25 3.1.2. 5. desert steppe (site #20). Figure 4.2. The large MUSOEXE experimental areainvolved containedtwo flux one observation superstation matrixes and in the stations four ordinary middle with reachunderlying of the Heihe River surfaces Basin: of one desert large (site experimental #19). Overall. et al. Gobi (site #18) and wetland (site #21). the Heihe and River Basin theinsubset and China. of River the framework of the objective The overall “Integratedof research on the eco-hydrological process of River Basin”. The Multi-Scale Observation Experiment program which formally initialized in May on Evapotranspiration 2012.2. residential studysandy area Gobi are primarily comprised desert (Figure by oasis cropland. 9. The underlying surfaces of study area are primarily comprised by observations [68]. The 5. cropland. 2017.5 experimental 2 area contained one superstation and four ordinary stations with underlying km kernel experimental area is located in the oasis. [68]. Remote Sens.oftoremote the applicability build asensing watershed in observing integrated system and to enhance eco-hydrological studiesthe andapplicability water recourseof remote sensingatinthe management integrated basin scaleeco-hydrological [67]. sandy Gobi desert (Figure 4). 2017. data source toA validate detailed and evaluate description ofthe ET models over heterogeneous HiWATER-MUSOEXE can be foundland in Lisurfaces.5 km 2 The). Heihe Watershed Allied Telemetry Experimental designed to be a comprehensive eco-hydrological experiment under the framework of the Research (HiWATER) is designed “Integratedto be a comprehensive research eco-hydrologicalprocess on the eco-hydrological experiment under Basin”. Heihe Watershed Allied Telemetry Experimental Research (HiWATER) in the Heihe River Basinis of northwestern China. The 5. to build a of HiWATER is to improve watershed theobserving observability of hydrological system and to enhanceand ecological processes. Description of HiWATER-MUSOEXE Campaign 3. HiWATER-MUSOEXE provides a valuable valuable data source to validate and evaluate the ET models over heterogeneous land surfaces.5 × 5. [68]. [67]Aanddetailed description Xu et al.5 × 5. The oasisunderlying surfaces ofareas.2.5 #18) and wetland (site× #21). of HiWATER-MUSOEXE can be found in Li et al. and Remotely Sensed Data for OSML Input over MUSOEXE 3.2.

Spatial distributions of meteorological variables. the visible. the OSML and SEBS models were evaluated under the same amount of available energy. The soil heat fluxes were measured using three heat flux plates located 6 cm below ground surface at each site. while the OSML. for fair comparison. Validation with Tower-Based Observations To validate against the observations from the SMACEX tower network. soil heat flux.org).1. near-infrared (NIR) bands of ASTER and albedo data set from HJ-1 were re-sampled to 90-m resolution. Meteorological observations including wind speed. Model bias in the estimated Rn is lower than 10 W/m2 and estimated G has a root mean square deviation (RMSD) of 23. 4. vapor pressure and atmospheric pressure were sampled at 10-min intervals. The albedo data have a spatial resolution of 30 m. [73] in this study.1. [70] from the thermal infrared bands of ASTER using temperature/emissivity separation (TES) algorithm proposed by Gillespie et al. The raw data were processed using Edire software and averaged over 30 min. In addition. [70]. [74]. combined with the water vapor scaling atmospheric correction method [72]. Results show that the estimated Rn and G agree well with corresponding ground observations. i.75].54.e. Net radiation was measured using a four-component radiometer. The LST and the land surface albedo are provided by “Heihe Plan Science Data Center.d respectively. driven by the same amount of Rn and G.3 W/m2 and mean absolute deviation (MAD) of 20. comparatively lower than those from SEBS .1.and SEBS-derived LE and H estimations are plotted against corresponding EC measurements in Figure 5c. crop height and fractional cover. The validation statistics are summarized in Table 2. Empirical relationships between ASTER-derived vegetation index and LAI.5 W/m2 . In this study.Remote Sens. The distribution of flux towers and the land use classifications in the MUSOEXE over Zhangye oasis are shown in Figure 4..b. 2 August (DOY 215) and 18 August (DOY 231) of 2012 were used in this study. 2017. Land surface albedo in the Heihe River Basin in 2012 was retrieved from charge-coupled device (CCD) camera on HJ-1 satellite by Sun et al. The RMSD for H and LE from OSML are 34. over the watershed were given by Yang et al. wind direction. please refer to Li et al. the estimated fluxes were averaged over the upwind source-area (1–2 pixels) for each flux tower [53. The LST data were retrieved by Li et al. air temperature. including air temperature. For consistency with the thermal infrared band. Three scenes from the Advance Spaceborne Thermal Emission and Reflection Radiometer (ASTER) on board Terra acquired on 10 July (DOY 192).heihedata. The visible and near-infrared bands of ASTER images have a spatial resolution of 15 m. 9. Validation of energy balance components of Rn and G are shown in Figure 5a. For details concerning the atmospheric correction and the retrieval of LST. Results 4. The soil heat flux was corrected using the temperature prediction-correction method proposed by Yang and Wang [69] with observed soil moisture profile and soil temperature profile. which is followed by pixel-based comparison of radiometric-convective resistance estimation from both models. [71]. and then the spatial patterns of LE and H generated from both models were inter-compared. and thermal infrared bands have a spatial resolution of 90 m. National Natural Science Foundation of China” (http://www. and Bowen ratio [65]. Validation at SMACEX Site The OSML and SEBS models were applied to the SMACEX region using input from satellite-derived and ground-based meteorological observations. 4. The estimation from both OSML and SEBS were firstly validated against tower-based observations. Sensible and latent heat fluxes were adjusted by forcing the energy balance closure using the measured net solar radiation. wind speed and relative humidity were interpolated using site observations based on the inverse distance weighting method. Turbulent fluxes were measured using EC system with 10 Hz.1 W/m2 . 43 11 of 25 21 automatic weather station sets are involved in the experiment. these empirical relationships were used to derive regional vegetation parameters for the HiWATER-MUSOEXE region.5 W/m2 and 46.

5 39. Spatial Distribution H of−Surface 18.2 and slightly overestimate H LE. observations. −18. Validation of: net radiation Rnn (a).2 74. 2017.2Fluxes 34. 43 Component Bias MAD RMSD Bias MAD RMSD 12 of 25 Rn 7. and sensible heat flux H (d) during Landsat overpass on Day of Year Year (DOY) (DOY) 174 174 and and 182 182 of of 2002.2 The spatial patterns of LE and H estimated by OSML and SEBS based on the Landsat-derived 4.3 18. and LE and H derived from the two models are very close at both regional and field SEBS-derived H and LE maps.5 −41. The statistics and thefor regional-scale contrast betweenLE and H estimations soybean areis and cornfields listed clearininTable both 3. this empirical relationship . A pixel-based intercomparison of LE and H estimated by OSML and SEBS was further conducted. 9.3 W/m2 and mean deviation (RMSD) OSML absolute deviation (MAD) ofSEBS 20.5 24.1. 43 12 of 25 Table 2. Model bias in the estimated Rn is lower than 10 W/m2 and estimated G has a root mean square Fluxof 23.4 30.and Landsat-derived SEBS-derived H and LE maps.5 51. (a). The spatial patterns derived from OSML and SEBSTheare spatial quite similar.6 46.e.2 4. latent heat flux LE (c).4 Heat 30.0 69. 9.5 24. soil soil heat heat flux flux G (b).0 7.3 LE 23. [60]. in cornfields and soybean 2 fields).5 51.0 7.5 39.2RW/mn 2 and 83.3 −6. The LE of the cornfields is higher than that in the soybean fields by scale (i.6 58.3 (RMSD are 69. about 100 W/m .2 Figure 5. The statistics for regional-scale LE and H estimations are listed in Table 3. respectively).2 74. and patterns of the LE contrast between by and H estimated soybean OSMLandandcornfields is clear SEBS based in both on the OSML. Spatial retrievals areDistribution of Surface shown in Figures Heat 6 and Fluxes 7 respectively.2 34.2.Remote Sens.1 23.5 Component BiasW/m and 2 MAD46.2 W/m2 .3 18. Summary of validation statistics for validating estimated surface fluxes (W/m2). 2002.2. The RMSD for H and LE from OSML are 34.9 20.5 24.6 46.2 W/m2.5 24. Summary of validationRnstatistics and G agree well with for validating corresponding estimated ground surface fluxes (W/m 2 ). The spatial patterns derived from OSML soybean fieldsandby about 100quite SEBS are W/m similar. 2 . The mean LE and H derived from the two models are very close at both regional and field scale (i.5 −41.3 18. respectively). The mean OSML. Because the employed empirical relationships between crop height and LAI are derived from in-situ observations in cornfields and soybean fields from Anderson et al. Flux OSML SEBS Remote Sens..9 20.9 20.1 23.3 18.2 83. Both models seem to slightly underestimate H 7..9 20. in cornfields and soybean fields). Results show Table that the estimated 2.6 58.0 and slightly overestimate G LE.0 G −6.1 23.2 83. comparatively Bias lower MAD thanRMSD those from SEBS (RMSD are 69. 2017. −6.1 W/m2. The LE of the cornfields is higher than that in the retrievals are shown in Figures 6 and 7 respectively.e.1.3 −6.5 W/m RMSD 2 .1 23. Both models seem to slightly underestimate H LE 23.2 W/m2 and 83.0 69.

2 TSEB.and METEIC-based H estimations in areas with partial vegetation cover (LAI less than 2) in the SMACEXA pixel-based site.0 areas.8 vegetation cover H (LAI less than 2)Regional in the SMACEX 95.6 resistance may be 74.8 derived from both models 68.1soybean fields450.2 84.2 Discrepancies in modeled LE are more Soybean 313. [53] reported 57.5 values derived from the two models are alsoOSML presentedCorn 78.Remote Sens.8 135.9 landscapes 75.8 139.7 H estimations 42.the cornfields and soybeanland surface fields.6 are discernible 101.8 indicating that HH estimations derived Regional 95.8obvious in pixels 61.5 are shown 62. Statistics for regional-scale land surface heat fluxes estimation (W/m2). indicating that Soybean LE values324.3 al.5 62. SEBS indicating Corn that H estimations derived 411.3these two models.2 103.2 in Figure 8.3 very similar.4 38.3 56. Theheat fluxes scatter estimation plots of the LE (W/m and ).7 38. In order to Soybean 313.7 from the two models 83.6 et al.6 discrepancies between SoybeanTSEB.1 450.6 83. on DOY 174 and 182.9 instance.2 in pixels 117.9with lower55.0with lower LE. The spatial distribution of LE estimated using OSML and SEBS from Landsat-based Figure 6. Regional 387.3 56. Thus.0 significant discrepancies between 53.2 103. and the difference SEBS Cornin the estimated 91.3 The performances models were344. The performances of different SEBS Corn models were 91. 9.3 Soybean the discrepancies of H estimation 135. The scatter 390.8 Discrepancies in modeled LE are more obvious Regional 93. Standard intercomparison Deviation Mean Standard was restricted Deviation to the pixels located in the cornfields and soybean Regional fields. intercomparison was restricted to Statistics for regional-scale 2 Table the pixels located in3. The discrepancies 84.9 in Figure 107.0 plots 386.9 107.0 discerned in large H values.8 highlighted in partially-vegetated Regional landscapes in existing literature. The majority of pixels Regional 390. For instance.9 in partially-vegetated 75.4 of pixels are closer to 1:1 line. 9.1are very similar. In this83.8 61. 43 13 of 25 Table 3. a similar Soybeanstudy by Choi et 47.0 OSML-derived LE SEBS shows goodCorn agreement 411. In thisintercomparison study.7 site.6 H values LE derived from the two models Regional are also presented 387.5 significant OSML Corn 78.8 8. DOY174 DOY182 may not be applicable to other landMean use types. The corresponding 303.8 of the LE and92. [60].9 175.6 highlighted 87.1 76. The corresponding H 57.0 53.and METEIC-based 146.3 344.1 39. The majority 470.4 are 38. The spatial distribution of LE estimated using OSML and SEBS from Landsat-based retrievals retrievals on DOY 174 and 182. 43 13 of 25 Remote Sens.7 LE.9 SEBS and OSML75.0 104. we use colorDOY182 shading to indicate pixel density.0 are one-source models. in cornfields and soybean fields from Anderson et al. For93.9 responsible for 75. OSML Corn 425.4 114. . DOY174 In order to avoid overlapping effect in the scatter plot.2study by Choi 117. OSML and andtheSEBS was further difference in the conducted.Soybean 324.2 8.1 in Figure 39. Mean Standard Deviation Mean Standard Deviation Generally.4discernible in partially-vegetated areas.8 413.4 are clearly 114.H derived from OSML against those derived from SEBS in the cornfields and soybean fields are shown in Figure 8. Generally.4 cornfields and59. both101.9 in 175.6 LE.6 83.3 H derived from OSML against those derived OSML Corn in the from SEBS 425. a similar 55.0 303. 2017. The discrepancies 413. both SEBSofand LEOSML and Hareestimated one-source bymodels. Because the employed empirical relationships between crop height estimated radiometric-convective resistance may be responsible for the discrepancies of H estimation and LAI are derived from these from in-situtwo observations models.8 92.7 42.3 in existing literature.7areas with partial 38.6 clearly discerned in large H values.2 Figure 6.6 in partially-vegetated 74. 2017.6study. Soybean 146. this empirical relationship may not be applicable to other land use types. Thus. from 47. OSML-derived LE shows good agreement with SEBS-derived LE.7 from the two models are 470. indicating that LE values derived from both models are 386. we use color shading to indicate pixel density.4 59. 76.0 with SEBS-derived 104.8 of different68. [53] reported139.6 radiometric-convective 87.6 avoid overlappingLE effect in the scatter plot.0 are closer to 1:1 line.

The 7. only. . Remote Sens. and comparison of OSML. 182.and SEBS-derived H: on DOY 174 (b). 43 43 14 of 14 of 25 25 Remote Sens. density).and SEBS-derived LE on: DOY 174 (a). 9.and SEBS-derived LE on: DOY Figure 8. soybean fields only. 9. and DOY 182 (d) (for cornfields and soybean soybean fields fields only. Comparison of OSML. and comparison of OSML. and DOY 182 (c). color shading indicates pixel density). 43 14 of 25 Figure Figure 7. Comparison of OSML. 9.The The spatial spatial spatial distribution distribution distribution of of Hof HH estimated estimated estimated using using OSMLOSML using OSML and and and SEBS SEBS SEBS from from fromLandsat-based Landsat-based Landsat-based retrievals retrievals retrievals on DOY 174 on on andDOY DOY 174 and 174 and 182. Figure 8. 2017.and SEBS-derived H: on DOY 174 (b). 2017. DOY 174 (a). and DOY 182 (c). 2017. and DOY 182 (d) (for cornfields and 8.7.Remote Sens. color color shading shading indicates indicatespixel pixeldensity). 182.

OSML in surface fluxes estimation.0relatively s/m and accurate estimate 27. 9.1. rae_OSMLresistance and rae_SEBS fromareboth OSML and comparable. DOY174DOY182 DOY174 DOY182 Mean Mean Mean Mean Corn Corn 51.and SEBS-derived rae.9 s/m.ofResultsrae_OSML show that both and rae_SEBS SEBS and for cornfields andOSML can provide soybean relatively fields against accurate observedestimate of radiometric-convective rae is shown in Figure 10. The mean values of rae_OSML and resistance was estimated in combination with the WDI derived from the conceptual VFC/LST rae_SEBS are very similar in both cornfields and soybean fields. thatstatistics the regional meanand for rae_OSML of LE andfor rae_SEBS H cornfields derived from the two fields and soybean models areare very listed in close Table at4. andas shownfields. In this study. and used radiometric-convective resistance is calculated from in-situ observations of H (denoted as rae_obs). rae_OSML and rae_SEBSareas partially-vegetated are comparable. benchmark shown Theinevaluation Figure 10. a variable resistance − 1 kB has approach great isimpact appliedon to estimate the performance of one-source radiometric-convective model. cornfields.and shown in Figure SEBS-derived rae .s/m and 27.460. 2017. as the rae_obs ).6 s/m.2 and Appendix A for Section 4.1. ae_OSML To andfurther compare rae_SEBS are the differences very similar in estimated in both cornfields radiometric-convective and soybean fields.6 39. Intercomparison between between rae_OSML rae_OSML . The statistics for rae_OSML a variable kB −1 approach is applied to estimate radiometric-convective resistance. Results show resistance.5 Corn Corn 47.6 43.8 43. radiometric-convective resistances derived from both SEBS and OSML 4.0 s/m. resistance.352. Figure 9 also illustrates the differences between OSML.8 for s/mrae_SEBS and are 9.d. .547. In OSML. 43 of the radiometric-convective resistance has great impact on the performance 15 of 25 of one-source model. soybean in Figure 9. which can be observed from the scatter plots of Figure 9.0 Figure Figure 9. respectively).This respectively).0 56. Table 4.8 s/m and 9.9 s/m.1.2. In SEBS. Intercomparison (denoted as rae_SEBS of and Radiometric-Convective rae_OSML respectively) Resistance Derived from OSML are inter-compared and SEBS to further investigate the differenceThein performance determination of these of thetworadiometric-convective models. Intercomparison 9. for rae_OSMLTheare RMSD and Biaslower noticeably (18. which resistance are infrom both with alignment OSML and in results SEBS. 2017. a pixel-based Section 4. and and as an important benchmark here. Intercomparison of Radiometric-Convective Resistance Derived from OSML and SEBS The Remotedetermination Sens. as the scatter plots of Figure 9. However.1.d.Remote Sens. rae_SEBSfor: .3 48. of radiometric-convective whereas the RMSD and Biasresistance. In this study. that both The SEBSRMSDand OSMLand Bias can provide for rae_SEBS are 39. illustrates the differences between OSML. that the comparison regional mean of rae_OSML of LE and and rae_SEBS from H derived in cornfields the and soybean two models fields are on very DOY close182 wascornfields at both conducted. rae_SEBS for:(a)(a) cornfields.5 56.2. radiometric-convective In order to evaluate resistance is calculated from radiometric-convective in-situ observations resistance derived from SEBS of H (denoted and OSML. and9. can abepixel-based observed from comparison of r ae_OSML and rae_SEBS in cornfields and soybean fields on DOY 182 was conducted.6 rae_OSML rae_OSML Soybean Soybean 52. the radiometric-convective and rae_SEBS for cornfields and soybean fields are listed in Table 4.2 and Appendix A for details).851. radiometric-convective resistances the radiometric-convective resistance derived from in was estimated both SEBS combination withandtheOSML WDI(denoted derived as fromrae_SEBS theand rae_OSML respectively) conceptual VFC/LST trapezoidare inter-compared space andtothe further electrical investigate the difference in performance of these two models. 43 15 of 25 4. In SEBS. 9. both cornfields The meanand values soybean of rfields.5 48. However. pixel density) .4 65. Table 4.3. Although different strategies this is consistent with thearediscrepancy adopted in SEBS in Hand overOSML. whereas the RMSD validation of raeand Bias further lends for rae_OSML are noticeably credibility to thelower proposed (18. analog formula of H (please see Section 2. Although To further comparedifferent strategies the differences in estimated are adopted in SEBS radiometric-convective and OSML. This validation of rae lends further credibility to the proposed OSML in surface fluxes estimation. and this is consistent with the In order to evaluate radiometric-convective resistance derived from SEBS and OSML. The evaluation of rae_OSML and r ae_SEBS for cornfields soybean usedfields as anagainst important observed rae ishere.0 rae_SEBS rae_SEBS Soybean Soybean 60. whichSEBS. which are in alignment with results in trapezoid space and the electrical analog formula of H (please see Section 2.0 65. andand (b) (b) soybean soybean fields fields on DOY on DOY 182 of 2002 (color shading indicates 182 of 2002 (color shading indicates pixel density). Statistics for regional-scale rae_OSML and rae_SEBS for cornfields and soybean fields (s/m). Figure 9 also observed in Figure 8b.The details).3. In OSML. Statistics for regional-scale rae_OSML and rae_SEBS for cornfields and soybean fields (s/m). the discrepancy in H over partially-vegetated areas observed in Figure 8b.

6 −4. Resultsshowshowthat thethe that RnRestimations n estimations from fromboth bothmodels models are are in good agreement with tower-observed observations.5 24.1 −23. 9. .5.6 20.2 20. Overall.8 11.3 W/m 2 for OSML.6 −4.3 38. 2017.4 .3 27. observations(W/m FluxFlux OSML OSML SEBSSEBS Component Bias MAD RMSD Bias MAD RMSD Component Bias MAD RMSD Bias MAD RMSD Rn −0.3 −23.Remote Sens.2.8 G −23. [59] and Bastiaanssen et etal. with the MAD theand MAD and RMSD RMSD of H49. which produced produced less credible less credible G estimation G estimation relative relative to to the estimated the estimated Rn . the the empirical empirical model model proposed proposed by by Allen Allen etet al. Evaluation Evaluation of of rae_OSMLand rae_OSML andrrae_SEBS ae_SEBS against againstrrae_obs.Statistics Table Statisticsof ofvalidating validatingOSML OSMLand andSEBS SEBSusing usingtower-based tower-basedobservations (W/m22). both OSML and and SEBS SEBS were were applied applied toto the the MUSOEXE MUSOEXE site using input from satellite meteorological observations. site.1 50.0 13. 9. Consequently.8 50. the performance of OSML was retrievals and regional meteorological evaluated using flux measurements from the observation observation matrix. (RnLE.8 67. overpass.10. Overall.5 51. . of satellite StatisticsStatistics overpass.8 Rn −0.8 51.2. which G.4 67. comparing OSML OSML comparing and SEBS output and SEBS and and output flux flux measurements measurements are summarized are summarized in Table in Table 5.H) andfrom H)OSML and SEBS from OSML andestimation at the time SEBS estimation of at the satellite time overpass. as the Rn.6 of W/m 2 38.4 57. .[76] [76]was was adopted adopted to estimate to estimate G.5 33.4 and 2 and 57.8 27.4 57.6 andW/mMAD 2 and of MAD33.0 67.4 of H being being W/m49.5 G −23.4 49. Validation 4.3 of W/m 2 33.5 50. 2017. withH.3 27.5 26.8 −1.8 27.andLE.42 . [74. flux.78].8 −1.5 26. EstimatedEstimated G has aGRMSD has a RMSD of 38. G. Validation with with Tower-based Tower-basedObservations Observations both OSML Similarly. pattern.6 20.77.5 LE LE 11.al.0 36. ae_obs 4.8 67. with the the absolute absolute mean mean bias bias in in estimated estimated RRnn less than 10 W/m 22 W/m . In termsIn terms of of surface surface heatSEBS heat flux.8 H H 13.2 42. AtAtHiWATER-MUSOEXE HiWATER-MUSOEXE site.Figure Figure11 11 shows shows thethe validation scatter validation plotplot scatter for eacheach for energy balance energy component balance component(Rn.77.2 42. 43 16 of 25 Figure Figure 10. at HiWATER-MUSOEXE Validation at HiWATER-MUSOEXE Site Site 4. for OSML. Table5. Results 5.536. G.3 38.2.tends SEBStotends to overestimate overestimate H.450.3 49. W/m W/m 2.5 33.scatter as the plot scatter plot between between model model derivedderived G and G and observations observations showsshows a morea more scattered scattered pattern. Flux measurements were linearly interpolated to to match match the the time timeofofsatellite satelliteoverpass.0 24.1.42 W/m57. the performance the performance of OSML of isOSML is comparable comparable to SEBS and to SEBS andpublished to other to other published studies studies [74.2 20. 43 16 of 25 Remote Sens.78].

LE derived from SEBS seems higher than OSML estimations. The contrast between sandy Gobi desert and Zhangye oasis are clear in OSML and SEBS-derived LE and H maps. However. 9. 9. DOY 215 and DOY 231 are shown in Figures 12 and 13.2.Remote Sens. On the which mainly comprises irrigated farmland. exhibits LE values lager than 350 W/m2 .2. The overpass time on DOY 192. The spatial pattern of estimated H and LE on DOY DOY 192 and DOY 215 are quite similar. which in Table 6. desert and Zhangye oasis are clear in OSML and SEBS-derived LE and H maps. LE derived from SEBS seems higher than OSML estimations. a pixel-based comparison of H and LE in the kernel experimental area (the yellow SEBS.2. respectively. 43 17 of 25 Remote Sens. contrary. LE increased in the surrounding desert steppe and LE map on DOY 231.Validation Validation of four of four energyenergy balancebalance components components from OSMLfrom OSML using using observations tower-based tower-based observations at at satellite satellite overpass overpass on DOY on DOY 192. On the contrary. 4. further indicate H and LE derived from two models are similar. . DOY 215 and DOY 231 are shown in Figures 12 and 13. has LE values the sandy Gobi desert. which further indicate H and LE derived from two models are similar. The statistics comparing OSML and SEBS are summarized SEBS is lower than OSML. However. Spatial Patterns of Estimated Regional Land Surface Heat Fluxes The spatial distributions of H and LE over MUSOEXE site based on OSML and SEBS at satellite The spatial distributions of H and LE over MUSOEXE site based on OSML and SEBS at satellite overpass time on DOY 192. Spatial Patterns of Estimated Regional Land Surface Heat Fluxes 4. In addition. Figure 11. spatial patterns produced by OSML and SEBS are very similar. due to the 2–4 mm of rainfall on DOY 231. In order to further compare the differences in estimated surface heat fluxes from both OSML In order to further compare the differences in estimated surface heat fluxes from both OSML and and SEBS. The statistics comparing OSML and SEBS are summarized in Table 6. while H derived from derived from SEBS is lower than OSML. 2017. LE increased in the surrounding desert steppe and Gobi desert and Gobi desert due to the 2–4 mm of rainfall on DOY 231. the difference between OSML. exhibits LE values lager than 350 W/m2. The fact that most in the Figure 4) of MUSOEXE was conducted and shown in Figure 14. DOY 215192. and DOY DOY 215 231 and DOY 231 of 2012. the difference between OSML. discerned.and SEBS-derived H and LE can also be In addition. The Zhangye Oasis. Specifically. The Zhangye Oasis. where precipitation is the only source for soil water. 43 17 of 25 Figure 11. has LE values less than less than 300 W/m2 and H values larger than 100 W/m2. where precipitation is the only source for soil water. 2017. respectively. a pixel-based comparison of H and LE in the kernel experimental area (the yellow rectangular rectangular in the Figure 4) of MUSOEXE was conducted and shown in Figure 14. Specifically. some spatial changes were found on the estimated H estimated H and LE map on DOY 231. The contrast between sandy Gobi The spatial patterns produced by OSML and SEBS are very similar. while H Specifically. comparable.2. The spatial pattern of estimated H and LE on 300 W/m2 and H values larger than 100 W/m2 . which mainly comprises irrigated farmland. of 2012. some spatial changes were found on the 192 and DOY 215 are quite similar. the sandy Gobi desert.and SEBS-derived H and LE can also be discerned. The fact that most points are points are concentrated along the 1:1 line indicates that heat fluxes derived from both models are concentrated along the 1:1 line indicates that heat fluxes derived from both models are comparable. Specifically.

8 12.0 16.13.4 LE H −0.5 47.3 12.1 MAD MAD RMSD 33. DOY 215 and DOY 231 of 2012. DOY 192 DOY 215 DOY 231 DOY 192 DOY 215 DOY 231 LE H DOY 192 LE DOY 215H LE DOY 231H BiasLE 24.5 19.1 H Bias MAD 24.9 21. 9.2012.3 38.6 16. 9. Statistics for OSML.6 19.1 21.4 47.6 Figure Figure12.1 36.0−13. 2017.and SEBS-derived H and LE for the kernel experimental area in MUSOEXE (W/m2). comparing 6.4 19.4 −12.1 13. DOY DOY215 215and andDOY DOY231 231ofof2012.8 49.8 38. Statisticsfor 6.1 22. Statistics forcomparing comparingOSML- OSML.3 21.1 43. 2017.4 21. 192.1 −21.1 21.6 47.0 36.8 −28. Figure Figure 13.0 49. 2012. DOY 215 and DOY 231 of 2012.8 22.0 21.4 28.6 43.1 LE H LE −21. 43 18 of 25 Remote Sens.4 RMSD RMSD 38. The The spatial spatial distribution distribution ofof H H over over MUSOEXE MUSOEXE site site based based on on OSML OSML and and SEBS SEBS from from satellite-based variables at satellite overpass time on DOY 192.5 0. 2017. . satellite-based variables at satellite overpass time on DOY 192. DOY 215 and DOY 231 of satellite-based variables at satellite overpass time on DOY 192.The spatial Thespatial distribution spatialdistribution distribution of of LE over over MUSOEXE MUSOEXEsite site basedononOSML OSML andSEBS SEBS from Figure 12.9 −0.0 47. and and SEBS-derived SEBS-derived H and H and LE the LE for for kernel the kernel experimental experimental area area in MUSOEXE in Table MUSOEXE (W/m 2 (W/m 2 ).5 49.8 Bias 24.9 33.0 H H28.0 47.4 −21. 9.0 16. 43 18 of 25 Remote Sens.12.5 43. The LE over MUSOEXE site based based on OSMLand and SEBSfrom from satellite-based variables satellite-based variablesatatsatellite satelliteoverpass overpass time time on on DOY 192. The spatial distribution of H over MUSOEXE site based on OSML and SEBS from satellite-based variables at satellite overpass time on DOY 192. 43 18 of 25 Table Table 6. Figure 13.1 22. DOY 215 and DOY 231 of 2012.647.1 36.4 33.5 H LELE −13.Remote Sens. ).

. the aerodynamic surface characteristics are often determined empirically based on land are often use determined types or cropempirically height.(a). a One-Source Model for Land (OSML) is proposed to address this issue.64. Figure 14. The prominent merits of OSML are the ET models [4. is an wind speed indispensable is an model indispensable input.53. model input. DOY 215 (d).54.54. a One-Source Model for Land (OSML) is proposed to address this In this study.and SEBS-derived H on: DOY 192 (b). However.79–82]. (e). as follows: (1) In OSML. This manner VFC/LST trapezoid space and the electrical analog formula of H. which based on land use will inevitably causetypes or crop uncertainties. 2017. suitable for application at regional scale with heterogeneous surface.53. both of which have been widely data and HiWATER-MUSOEXE data.1. which has long abeen been a difficulty difficulty in landheat in land surface surface fluxesheat fluxes with estimation estimation remotely with sensedremotely LST.33]. and SEBS-derived H on: DOY 192 (b). variation of wind velocity at regional scale is difficult to variation of wind velocity at regional scale is difficult to reproduce using in-situ observations [17.32. This manner of physically deriving of physically deriving radiometric-convective resistance eliminates the need for an empirical extra radiometric-convective resistance eliminates the need for an empirical extra resistance. . 43 19 of 25 Remote Sens. color shading indicates pixel the kernel experimental area in MUSOEXE. both of which have been widely used to validate and evaluate used to validate and evaluate the ET models [4.33]. on Information (2) Information on wind velocity and aerodynamic surface characteristics is wind velocity and aerodynamic surface characteristics is not required in OSML.34. several approaches have been proposed to adjust aerodynamic resistance so that H can be estimated using LST directly. Discussion on Different Approaches to Mediate Difference between LST and Taero As stated in Section 1.1. and DOY and comparison of OSML.2. 2017. sensed LST.79–82]. The performance issue. 9. In mosttwo-source and one-source and two-source models such as SEBS models suchwind and TSEB. 43 19 of 25 Figure 14. information wind speed and Theaerodynamic independence fromcharacteristics surface information has of wind madespeed OSMLand more aerodynamic suitable for surface characteristics has made OSML more application at regional scale with heterogeneous surface.density). the aerodynamic surface characteristics In addition. and DOY 231 (f) (for 231 (f) (for the kernel experimental area in MUSOEXE.Remote Sens. The performance of OSML of OSML is is comparable comparable to to the the typical typical one-source one-source model model of of SEBS SEBS whenwhen validated validated with with thethe SMACEX SMACEX data and HiWATER-MUSOEXE data. The prominent merits of OSML are as follows: (1) In OSML. the radiometric-convective the radiometric-convective resistance resistanceis estimated is estimated byby combing combingthe theWDI WDIderived derivedfromfrom the conceptual the conceptual VFC/LST trapezoid space and the electrical analog formula of H. Discussion 5. Comparison ComparisonofofOSML- OSML-and andSEBS-derived SEBS-derivedLELE on: on: DOY DOY 192 192 (a). to account for the difference between LST and Taero. which has longresistance. Advantages of of OSML OSML over over Typical Typical One-Source One-SourceModels Models Due to Due to the the discrepancy discrepancy between between LST LST and andTTaeroaero.34. and comparison of OSML. as SEBS speed and TSEB. DOYDOY 215215 (c). whichThe will inevitably cause independence from considerableofuncertainties.(c). 5. Discussion 5.reproduce using in-situ observations [17. 5. Advantages 5. color shading indicates pixel density). However. the the determination determination of of radiometric-convective radiometric-convective resistance poses resistance poses aa major majorchallenge challengefor forregional regionalheat heatfluxfluxestimation estimation based based ononremotely remotely sensed LST. andand DOYDOY 231 231 (e). DOY 215 (d).32. In most one-sourcenot required in OSML.In this study.64. 9. In addition. (2)sensed LST. considerable height.

Remote Sens. and China Meteorological Administration (CMA) Land Data Assimilation System (CLDAS) data set [86]. the regional application of OSML employs Landsat images. the performance of OSML is very comparable to typical one-source models such as SEBS and other published studies. 5 km resolution) would also provide sufficient data source for routinely monitoring surface heat fluxes using OSML. As an alternative to the above-mentioned approaches. Modern Era Retrospective Analysis for Research and Applications (MERRA) data set [84]. 9. In this study.3. the Global Land Data Assimilation System data set [85]. is an indispensable variable for SEBS model. In OSML. Most of the works were based on the adjustments of the kB−1 parameter. the variable kB−1 approach used to determine radiometric-convective resistance in SEBS requires the parameterization of crop height. developing a reliable method for extrapolating daily values to longer time scales becomes imperative. To evaluate the performance of the proposed method. the SMACEX and HiWATER-MUSOEXE. surface roughness and displacement height. whereas the Moderate Resolution Imaging Spectroradiometer (MODIS.2.9. which is difficult to obtain at regional scale. Besides. radiometric-convective resistance is used in lieu of aerodynamic resistance for H estimation with radiometric surface temperature. 2017. A further intercomparison of radiometric-convective resistance derived from both OSML and SEBS indicates that . In this study. the OSML was applied over two regions with contrasting climatic features. The work by Lhomme and Montey [21] and Zhao et al. such as NASA Global Modeling and Assimilation Office’s (GMAO). the evaluation of rae_SEBS and rae_OSML against rae_obs suggests that OSML provides a novel and reliable way to parameterize radiometric-convective resistance (Figure 10). a variable kB−1 approach is adopted in SEBS model. Among them. would make it possible to routinely apply OSML at regional scale for ET monitor/prediction.5 W/m2 and 46. this approach is found to be unstable under sparse steppe [28] and sparsely vegetated area [77]. Chehbouni et al.58. For longer time applications. Operationality of OSML Satellite remote sensing is a promising technique for obtaining regional land surface heat fluxes and this information is important for precise quantification of the water balance for water resources planning and management. wind speed. namely. to account for the difference between LST and Taero .5 W/m2 at SMACEX site and 50.59]. the conceptual VFC/LST trapezoid model and the electrical analog formula of H are combined to analytically resolve the radiometric-convective resistance. OSML does not need to quantify aerodynamic related parameters. [83] indicate that a universal parameterization of kB−1 seems impossible. The related adjustment parameters are often estimated based on its empirical relationship with LAI. In contrast to SEBS model. Discussion on Different Approaches to Mediate Difference between LST and Taero As stated in Section 1. In comparison. 43 20 of 25 5. the RMSD of H and LE from OSML are 34. However. For instance. 6. identifying crop stress and drought impacts [3. and thus circumvent the difficulty of parameterizing the empirical “extra resistance” in conventional one-source energy balance models. optimizing crop production. Compared with tower observations.0 W/m2 at HiWATER-MUSOEXE. 5. Conclusions In this study. which will greatly promote the application of surface heat fluxes estimation at regional scale. 1 km resolution) or geostationary meteorological satellite images from Advanced Very High Resolution Radiometer (AVHRR. several approaches have been proposed to adjust aerodynamic resistance so that H can be estimated using LST directly. Overall. regional meteorological variables required for routinely application of OSML are also readily available from China Meteorological Administration. In addition.1 W/m2 and 67.8. a One-Source Model for Land (OSML) is proposed for estimation of regional-scale surface fluxes. [20] proposed a β function (βch ) and Matsushima [22] proposed a α function to interpret the relationship between LST and Taero . The real-time or near-real-time reanalysis products from a data assimilation system.

2. Tang. combine Equations (2) and (3) to eliminate H. C. Hydrology: An Introduction.O. China (No. pp. NY. Zhou.R. the Fundamental Research Funds for the Central Universities (No. P. P. K. Evapotranspiration: A process driving mass transport and energy exchange in the soil-plant-atmosphere-climate system. Vegetation.. NY. four roots can be derived. Advances in thermal infrared remote sensing for land surface modeling. Petropoulos. Author Contributions: Yongmin Yang and Jianxiu Qiu are the principal authors of this manuscript. and Applications. We greatly thank the National Snow and Ice Center for providing the SMACEX data set. 2005. Seo. S. Agric. 31. [CrossRef] [PubMed] .J.. [CrossRef] 6. For. M. Oren. 149. R.. 41501450. the National High Technology Research and Development Program (2012AA12A309) and IWHR Research & Development Support Program (No. 3801–3853. However. UK.. 3. Springer: New York. 5. 2012. 2016A030310154). Kustas. USA. [CrossRef] 9. 2007. K. and c0 are functions of variables. and all authors contributed in results discussion. c3 . Please contact first author or corresponding authors for the source code for OSML. A review of current methodologies for regional evapotranspiration estimation from remotely sensed data. Manzoni.L. Parlange.cn).P.H. Brutsaert. The HiWATER-MUSOEXE data set is provided by Cold and Arid Regions Sciences Data Center at Lanzhou (http://westdc. Humans and the Climate: A New Perspective on an Internactive System. W. Estimation of instantaneous and daily net radiation from MODIS data under clear sky conditions: A case study in East Asia. the study suggests that the OSML is promising and can be used to estimate regional surface heat fluxes over heterogeneous surface.. [CrossRef] 10. M.westgis. Acknowledgments: This work is supported by the National Science Foundation of China (Grant Nos.. Bi. 2009.. 7.B. Lee.. P. Appendix A.. Integrating remote sensing and ground methods to estimate evapotranspiration.Y.. J. and can be calculated without introducing new variables. Cambridge University Press: Cambridge. 139–168. only positive real solutions are meaningful. 1982. 16lgpy06). 2017.H. Institute of Geographic Sciences and Natural Resources Research. G. Hirschboeck. 2004. Zhang. W. Li. Sensors 2009. Y. without the requirement of information on wind speed and specification of surface characteristics.. Anderson. M. Glenn. 26.W...S. UK.ac. Key Laboratory of Water Cycle and Related Land Surface Processes. JZ0145B612016). Tang. UK. Choi. Estimation of rae in OSML First. Huete.M. The quartic equation can be solved with “FZ_Roots” function in IDL. 2071–2081. G. 9. B. Eagleson. Ecohydrology: Darwinian Expression of Vegetation Form and Function. 43 21 of 25 the proposed novel method can produce reliable radiometric-convective resistance and thus accurately estimate surface fluxes. Geophys. c2 .L. [CrossRef] 4. P. Hwang. 224–250. References 1. and then use “solve” function in MATLAB® to convert Equations (2)–(11) to a quartic form with one unknown variable (rae ): c4 r4ae + c3 r3ae + c2 r2ae + c1 rae + c0 = 0 (A1) where c4 . A. Water. 41571356 and 51420105014). Generally. Crit. 41501415. Meteorol.. Plant Sci. Springer: New York. The positive real solutions are then substituted into Equation (2) and estimated H is constrained to be positive and lower than available energy (Rn -G). 2015A006).Remote Sens. Cambridge University Press: Cambridge. the Natural Science Foundation of Guangdong Province. Higgins. All people involved in the field campaign are greatly acknowledged. 2002. G.K. C. Rev. Kabat. Wan. R. Z. History. Overall. 2012. IWHR Research & Development Support Program: “Study on regional drought monitoring and early warning based on satellite remote sensing and land surface hydrological model”. Katul. Rev. Conflicts of Interest: The authors declare no conflict of interest. 2013. USA.. Z. and LST. Brutsaert. X. S. W. 1173–1184.. Remote Sensing of Land Surface Turbulent Fluxes and Soil Moisture: State of the Art. Irrig.Y.. c1 . 9.. E. Sci. Evaporation into the Atmosphere: Theory. G. Nagler.G. 8.L.. Chinese Academy of Sciences (No. including Rn . Taylor & Francis: Abingdon. Yan. Brown.

. Hirschi. 261–274. Agric. Z. Liu. J. W. measurement and related problems.C. For. Z. [CrossRef] 26. Fisher..P.M. F. Dirmeyer. Gowda. B. Choudhury. Running.. 1995. A. Ludwig. Jimenez. 225–253. M. 9.. J. Meterol. E. Determination of sensible heat flux over sparse canopy using thermal infrared data.M. De Bruin.. Verhoef. Mu.J... 177–186.. 1996. Olioso.... P. M.W. O..D. [CrossRef] 12.. Earth Syst. M. 17.. O. Kustas. Agric.. G.. Chehbouni. K. Estimating land surface evaporation: A review of methods using remotely sensed surface temperature data. Balsamo... Meteorol. Mu. R. J.... Environ. C. Ciais. Sci. B.W. Fisher. 2005...L.. P. G. Howell.. 369–384. 58.. McCabe. E. B. 2000. J. Decadal trends in evaporation from global energy and water balances. Q. Timmermans. C. A... Dirmeyer. Peel.. J. Jung. Ceschia.. [CrossRef] 30.. 2005. Hydrol. J.. R. Anderson. 83. Li. Weaver. 116. Paul... 2011.. Mueller. T. M... Reginato. An empirical expression to relate aerodynamic and surface temperatures for use within single-source energy balance models. 29. Kustas. Lhomme. Olioso. Remote Sens. [CrossRef] 24. 2012. Evaluation of global observations-based evapotranspiration datasets and IPCC AR4 simulations. Integration of soil moisture in sebs for improving evapotranspiration estimation under water stress conditions.J. 43 22 of 25 11.. [CrossRef] 14. Sci. [CrossRef] 16. Chehbouni. Maignan. C. L. Geophys..F. Discuss. Irrig.. For. Vekerdy.H.. Environ. Mc-Cabe. 431–457. H. E.A. E. 379–391. R. 2011. T. 19. J. J. K. For. Betts. Nichols.V. A. 509. [CrossRef] 25. Moran. Rivalland. Shen. Gay. Corti. Monteny. Remote Sens.Q.. F. W. Remote Sens. Sensible heat flux-radiometric surface temperature relationship over sparse vegetation: Parameterizing b-1. S. Wang.M. M. Seguin.. 2011. 6. Meteorol. Investigating the influence of roughness length for heat transport (zoh ) on the performance of SEBAL in semi-arid irrigated and dryland agricultural systems. 3707–3720.S.P.. Relations between aerodynamic parameters of heat transfer and thermal-infrared thermometry in the bulk surface formulation. Meteorol. Marloie. [CrossRef] . Meteorol. W.. Humes. Environ. Van der Tol. Zhang. 2007.L. M.. [CrossRef] 20. Y. Syst. A. A.Remote Sens. Seen. Environ. G. Wood. C. A. J. Q. Monteny.. 1–27.M. 161. 263–293. [CrossRef] 17. 44. F. M. 38.R. 13. A.N. 121. M. Mueller.. 2007. Zhang. G.D. D. Jimenez. 1781–1800. W.S. Humes. Hydrol. B. M. Res.. 421–469.. Benchmark products for land evapotranspiration: Landflux-eval multi-data set synthesis. Review on estimation of evapotranspiration from remote sensing data: From empirical to numerical modeling approaches. Chiew.. Remote Sens. 519–536. W. P. Hydrometeorol. C. 2012.B. 97... Mueller.P. An intercomparison of the Surface Energy Balance Algorithm for Land (SEBAL) and the Two-source Energy Balance (TSEB) modeling schemes. [CrossRef] 21. L. T.. Kustas. Gokmen. W. Prasad. F.. Dolman.. Heinsch.. Chehbouni. Drain. Su. [CrossRef] 23. S. Bound. P. V. Sensible heat flux-radiometric surface temperature relationship for eight semiarid areas. 1989. P.. 77. 1995. Leuning. J. Development of a global evapotranspiration algorithm based on MODIS and global meteorology data. Remote Sens. F. [CrossRef] 19... [CrossRef] 22. B. 115. 2012. Seneviratne.W.L. [CrossRef] 28. Batelaan. S. C. F.S. Hydrol.A. Verhoef. McVicar. M.S.. H.J. Jpn. Source approach for estimating soil and vegetation energy fluxes in observations of directional radiometric surface-temperature. 108.J.. Res. Agric. Geophys. B. Dirmeyer.H. Z. [CrossRef] 29.. 231–244. S. P. Earth Syst.. 223–249. [CrossRef] 18. Njoku. 197–216.-P. M. Running. R. Ciais. Examination of the difference between radiative and aerodynamic surface temperatures over sparsely vegetated surfaces. Courault. 148–155.P. Zhao. Kustas. G. et al. 2013. Guo. Surv. Sun.. 1994. W.. B. French. 1–7. 85–99. Soc. P. B. Z.. Jackson. Neale. M. Chirouze. Moran.A. Geophys.. M. Balsamo. Remote Sens..J. 2017.. W. D. Lett.... 33. Seneviratne. A. J. 2002. Rossow. [CrossRef] 27. J.. Surface temperature and emissivity at various scales: Definition. Boulet. Norman. Appl.B. Jiménez. [CrossRef] 13. Hirschi.S.. Kalma..Z. 2008.... Zhao. Prigent. The Surface Energy Balance System(SEBS) for estimation of turbulent heat fluxes. Layer Meteorol.. Rev.S. 373–389. Stewart. 2014. 111. C. Coudert..B. Improvements to a MODIS global terrestrial evapotranspiration algorithm.L.. Matsushima. 1110–1117. 12.. Jung.D. Environ.C. [CrossRef] 15. Global intercomparison of 12 land surface heat flux estimates. D. Y.. Aiken. Becker..

Environ. 1989. A. 1991. Zhang. Thomsen. 243–261. 1987..J. 112. [CrossRef] 32. Philos. G. 2014. Sandholt. 1665–1679. Xu. China Ser. Zhu. Tang. 54–65. [CrossRef] 44.. 49.. 540–551..B.. E. R. Ardö. R.. 33.. S.S. For. R. Res.. 592–604. Water Resour. S. J..... 26. 225–244. air temperature. J. Soegaard. 43. I. Manag. Norgaard.J.. Reginato. 2008.. 30. 2014. 114. Environ. Res. 329–343.P. Water Resour. Zhang. Planton. H. 1994.. 1245–1259. Tasumi.. Z. Henderson-Sellers. A. Mallick. Assoc. Choudhury.. Estimation of evaporative fraction from a combination of day and night land surface temperatures and NDVI: A new method to determine the priestley-taylor parameter. A review of Ts/Vi remote sensing based methods for the retrieval of land surface energy fluxes and soil surface moisture. Remote Sens.H. Sensitivity of landsat-scale energy balance to aerodynamic variability in mountains and complex terrain. Water Resour.. Appl. R. [CrossRef] 43. Idso.E.J. 2013. Moran. Trezza... Evapotranspiration models with canopy resistance for use in climate models.. Wigmosta.W. Z. Stisen. Phys. E. Sun. Sci.M. Rosenzweig. Inoue. J. K. 3267–3283. W. Drewry. Meteorol. 117. Fensholt. a review. 102. 39..S. B. 2017. 37. 2001. Remote Sens. G. Lorite. Fisher. Beringer. [CrossRef] 45. B. S.. [CrossRef] 34. Jarvis. Su. Sud. 50. Environ. Res. Weather Rev. Combining the triangle method with thermal inertia to estimate regional evapotranspiration-applied to MSG-SEVIRI data in the senegal river basin. Prog. Yang. 37.L. Boegh. An evaluation of two models for estimation of the roughness height for heat transfer between the land surface and the atmosphere.. 2009. R.. Remote Sens. Schmugge. J. P.. 273. The interpretation of the variations in leaf water potential and stomatal conductance found in canopies in the field.Remote Sens. R. Jensen. A Simple Biosphere Model (SIB) for use within general circulation models. J. Environ.H. X. R. 121. P. Environ. Water Resour. D. Remote Sens. Cribb.. Tang...G. Sci. Z. M. Mintz. A. 593–610. Analysis of an empirical model for soil heat flux under a growing wheat crop for estimating evaporation by an infrared-temperature based energy balance equation. Geogr.. 2015. Y... Tian. 54. Su.M. Trans. Wooster. An operational two-layer remote sensing model to estimate surface flux in regional scale: Physical background.. 373–388. Islam. Hook. 536–549. L. Meteorol. J. 141. Islam.. Wang. 505–531. Jarvis.L. J. W. Tasumi. Lu.. Remote Sens. 2012. H. Long... Petropoulos. Li. Noilhan. Meteorol. T. [CrossRef] . Dalcher. 2012.L. 1994.. 246–263. Sellers. H. [CrossRef] 46. 224–250. Z. A..P.. W. 43 23 of 25 31. Estimating crop water deficit using the relation between surface-air temperature and spectral vegetation index. Environ. M. 9. [CrossRef] 51. D. and surface net radiation measurements. 49. Sci. S. Lond. Lettenmaier. 2001. Clarke.. Aerodynamic parameterization of the satellite-based energy balance (METRIC) model for et estimation in rainfed olive orchards of Andalusia..A.. Li. K..C. R. R. [CrossRef] 38.P. Water Resour. 48. Atmos. V.E. 40. An Enhanced Two-source Evapotranspiration Model for Land (ETEML): Algorithm and evaluation. 329–340. 79. R.. A.. K. [CrossRef] 40. An application of the Ts/Vi triangle method with enhanced edges determination for evapotranspiration estimation from MODIS data in and and semi-arid regions: Implementation and validation. J. Estimation of surface evaporation map over southern great plains using remote sensing data.. 1986. Remote Sens. Remote Sens. A new parameterization scheme for estimating surface energy fluxes with continuous surface temperature. [CrossRef] 36.N. 283–297. Hulley. A. Tang. M. I. [CrossRef] 50.. Y. Boegh. L. 2006. Tu. J. S. [CrossRef] 49. 2010. Y.. J. Am. J.J. Singh. Li. 1976. T. [CrossRef] 33. Allen. For. Tian.. 168. L.L. T. [CrossRef] 47. [CrossRef] 35. Evaluating evapotranspiration rates and surface conditions using landsat tm to estimate atmospheric resistance and surface resistance. Kilic. Vidal. P. Li. Kustas. Dickinson. M. A simple parameterization of land surface processes for meteorological models. Remote Sens.. Massman. H. S. R.P...T. A Two-source Trapezoid Model for Evapotranspiration (TTME) from satellite imagery.. Y. Sellers. [CrossRef] 39. Wang. [CrossRef] 42.. [CrossRef] 41. D 2005. Tang. Environ. Santos. 293–305... C. Carlson. Soc..H. Allen. Environ. A. Li. A Surface Temperature Initiated Closure (STIC) for surface energy balance fluxes. Jiang. D. Mon. Z. M. A distributed hydrology-vegetation model for complex terrain.R. K.. 1242–1255. C. 2002. [CrossRef] 48. Vail. 370–388.Q. M. Agric. Agric. 1933–1951. B Biol.. Spain.

Z.H. C.U. M. [CrossRef] 65. Su.M. Tasumi. [CrossRef] 56. 2007. Liu.. US) during smacex..U. H. X. Irrig. Meteorol.J. M. and preliminary findings. Wang..C.. Reflections on the surface energy imbalance problem. W. Environ. Parkin.P. Hydrometeorol. E. Li. Zhu. For. Xu. 9. Meteorol...P. M.G. Z... Phys. Jin.. 4395–4402. Jackson.L. S. Correcting eddy-covariance flux underestimates over a grassland.P. 2000.R. Remote Sens.. Trezza. C. A. S. MacPherson. Hipps.. Wang. Mateos..L. Prueger. J.. 149. C. Norman. H. 2005.. J.P. 395–406. Morse.. Jackson. M. Q. S. 2005. Satellite-based energy balance for Mapping Evapotranspiration with Internalized Calibration (METRIC) . Cook. W. Bastiaanssen.. Atmos. Bindlish. Gonzalez-Dugo.. J.N. Rokugawa. [CrossRef] 57. Deriving land surface temperature from Landsat 5 and 7 during SMEX02/SMACEX.M. C. [CrossRef] 71. Eng. C. Anderson. 36. Agric. 2014. Meteorol.F. [CrossRef] 69. Hydrometeorol. [CrossRef] 63. Liebethal. Mauder.. J.. F.. Q. H. At-surface reflectance and albedo from satellite for operational calculation of land surface energy balance. Drain. T.J. Geophys. R. J. J. Hydrometeorol. Neale. Chem. 2004. F. 2006. Tasumi. H. Wang. Sun. C. Anderson. [CrossRef] 53.... Evaluation of the VIIRS and Modis lst products in an arid area of Northwest China. Q. An intercomparison of three remote sensing-based surface energy balance algorithms over a corn and soybean production region (Iowa.. R. Shi. W.R. Choi.P..E. R. W. B. Prueger. 878–891. M.G. 65–74. Lorite. Liu.. The Soil Moisture-Atmosphere Coupling Experiment (SMACEX): Background.L.H.. A temperature prediction-correction method for estimating surface soil heat flux from soil temperature and moisture data. 2005. T. J.G.H. R.F. Res.. and carbon dioxide fluxes during smacex. 2004. Houser. Leuning. J.H. Ma... Eichinger. 43 24 of 25 52. Massman. J....C.L. For. M..P. 51. Twine. J. 2012. 94. Liu. A. W. 2008. Soc. D.. Prueger. [CrossRef] [PubMed] 62. C. Li. [CrossRef] 66. 6. 2082–2097.. J.J. Robison. Upscaling ground observations of vegetation water content. Modeling evapotranspiration during smacex: Comparing two approaches for local.C.applications. Meyers. Eng..G.. F. Tower and aircraft eddy covariance measurements of water vapor.. 6. [CrossRef] 59. Du. China Ser. Agric. Remote Sens. A temperature and emissivity separation algorithm for Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) images. McCabe. [CrossRef] 61. J. Z.. Liu.. hydrometeorological conditions.. Intercomparison of surface energy flux measurement systems used during the hiwater-musoexe. Thomas. Anderson. Kustas. energy. P.R. Meteorol. Xu. Hatfield.B. Y. 2017. van Gorsel. Neale. [CrossRef] 70. J. Xiao. Jayanthi. Kustas. Allen.. Remote Sens. 149.. M. Che. For. Allen... W.I...P. 521–534. 721–729. Cheng. 2013.and high-vegetation cover conditions.. 156. A comparison of operational remote sensing-based models for estimating crop evapotranspiration. Prueger.. Environ. F. A. T. 92.Remote Sens.. T. 103. 1351–1367.Q. Schmugge.H. Bull.. Cooper. Trezza.. Sci. 2013. Hydrol. Irrig.E. Allen. 2008. 92. 380–394. Wang. Matsunaga. 2005. J. A.U. 1843–1853.. Atmos. R.. Kustas.J. T.M. 2007. 2009.. 6.. Satellite-based energy balance for Mapping Evapotranspiration with Internalized Calibration (METRIC)—Model. [CrossRef] . E. 910–922. Li.H. W.. M. Utility of remote sensing-based two-source energy balance model under low. L. W. 142. Li. W.B. 447–464. Hook. Trezza. 791–804. Wright. Wood. [CrossRef] 60. T.. I... Liu. Kustas. J. Remote Sens. Hatfield. T.S. R.. Appl. J. French.. Su.. D.. Isaac. [CrossRef] 68. and leaf area index during SMEX02 using aircraft and landsat imagery. Qi. Norman. Cosh. 2008.. Some aspects of the energy balance closure problem.I... 279–300. J. W. Wang. T. K. T..H. 13. Kahle. M. F. 18. T...and regional-scale prediction.P. Chavez... 1998.J. Wang. Wesely... Li. Prueger... 954–960. P.. L. [CrossRef] 54.. Neale.. Li. R. Wimmer.P.... Li. R. J. W. 111–121. M. R.E.B.P. Starks.Q.M. Li. Allen. W.. Ma. J. R. Ecol. [CrossRef] 67. 1113–1126.... Yu. [CrossRef] 64.. D... Cao. M. Y.W. X... M. 133.. [CrossRef] 58. The energy balance closure problem: An overview. Neale. Yang. Cothern. F. J. D Earth Sci. Tasumi.M. Gillespie. 133. S. Drain Eng.H. J. Meteorol. Agric. J.. Am. Y. Kustas. Foken. G.M. Foken. 6. Kramber.. 51–63.. 6. Environ. [CrossRef] 55. P. Kustas. Hydrometeorol. Prueger. Heihe watershed allied telemetry experimental research (hiwater): Scientific objectives and experimental design. Agric. W. J. M. Geosci. Kustas. J.. M. Kustas. H. 1145–1160. Kjaersgaard. IEEE Trans.U.. 2009. Y. S. W.. J. canopy height. T. For.

Jiang. 471. McCabe. Hydrometeorol. C. 99. Switzerland. Surface energy fluxes with the Advanced Spaceborne Thermal Emission and Reflection Radiometer (ASTER) at the Iowa 2002 smacex site (USA)..M.F. Comparing aircraft-based remotely sensed energy balance fluxes with eddy covariance tower data using heat flux source area functions. S. 27 April–2 May 2014. Tonooka..T. IEEE Trans. Mitchell. Environ. Wood... X.E. Land Resour.R. Hydrol. Wen. W. Huang. Holtslag. Prueger. 2005. 58–63. 923–940. [CrossRef] 78.. M.. [CrossRef] 84. J. W. M... Bacmeister. Lu. Todling. Z. © 2017 by the authors. Zhang.A. C. 158.. B. E.. Rodell. R.. Zhang.E. Neale.M. Long. Z.. 212. S.. 2015. Shi... 43 25 of 25 72. A. Y. Ma. 16795–16814. Su.. Y. J. R.. M..org/licenses/by/4. 2016. J. Influence of leaf area index on the radiometric resistance to heat transfer. Hipps. M.A. Liang.M. Res. licensee MDPI. M. Xu. 25.H. Houser. Anderson. [CrossRef] 81..Remote Sens. C. J.. K.L.E.P. 198–212. 2006. J. L. Status and plans of CMA Land Data Assimilation System (CLDAS) project.F.. In Proceedings of the EGU General Assembly 2014.. L. 892–909. Bosilovich. 85. C. 2015... 2013.. Gelaro. 1. [CrossRef] 77. 1998. Effects of vegetation clumping on two-source model estimates of surface energy fluxes from an agricultural landscape during smacex. Remote Sens.. 2017.. Norman. Menenti.. Y. Austria.N.. Y. J. 2005.C. A. M. Jia. X. 24.. O. E.. W.. G. [CrossRef] 76. S.. Bastiaanssen. Res. et al. Geosci..Q. Li. Han.. A. J.-K. Prueger. Wen. W. W. Li. J..J. J. Basel.. Merra: Nasa’s modern-era retrospective analysis for research and applications.. McCabe. Arsenault. C. 2015.. Cosgrove. X. M. . B.H. Kustas.... 2011.. F.. 105. Hydrometeorol. 2005.. Suarez.D.M. Chavez. Lee. Geophys. Gottschalck.. Zhou. 271–285. Rienecker. Shang. Zhao.. Environ.. J. D. Comparison of three dual-source remote sensing evapotranspiration models during the musoexe-12 campaign: Revisit of model physics. 6. 79. Song. A. IEEE Trans. A remote sensing Surface Energy Balance Algorithm for Land (SEBAL). L. Meteorol. Gu. 118. L. Li. Takacs. Estimations of regional surface energy fluxes over heterogeneous oasis–desert surfaces in the middle reaches of the Heihe River during hiwater-musoexe. Yang.. 6.. 2284–2300.U.C. P. Bull. S.M. 3624–3648. 43. [CrossRef] 73. Scale influences on the remote estimation of evapotranspiration using multiple satellite sensors. Remote Sens. Remote Sens. Geosci.F. Atmos. H... Bosilovich. R. J. [CrossRef] 80.. T. J.. Suyker. [CrossRef] 75. Batelaan. Jacob.G. Remote Sens. 74... Guan. Remote Sens. 2013. F. Meng.. 671–675. Radakovich. A hybrid dual-source scheme and trapezoid framework-based evapotranspiration model (HTEM) using satellite images: Algorithm and model test. An algorithm for retrieving land surface albedo from HJ-1 CCD data.. Kustas. Li. H.0/).P. 9. Am.. Simmons. Schubert. Gieske. [CrossRef] 86. Sun. [CrossRef] 85. Liu. Bound. F. Remote Sens. C. M. [CrossRef] 83. French. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC-BY) license (http://creativecommons. 2005. Water Resour. Su. Soc. Timmermans. Clim.P. Accurate atmospheric correction of ASTER thermal infrared imagery using the WVS method. Liu. J. Jambor. H. 105–123.. W... Kim. [CrossRef] 82. K. Mecikalski. Anderson. Liu. Formulation.A. L. Feddes. Q. J. F.. 381–394. M. Improving estimation of evapotranspiration under water-limited conditions based on sebs and MODIS data in arid regions. 12. Layer Meteorol. M. The global land data assimilation system. 7.. U. Vienna.G. Yang. 2004. L. Kustas. 2778–2792.H. J. J..