Regionalization of Surface Fluxes over Heterogeneous Landscape of the Tibetan Plateau by Using...

Journal of the Meteorological Society of Japan, Vol. 81, No. 2, pp. 277--293, 2003 277

Regionalization of Surface Fluxes over Heterogeneous Landscape of the

Tibetan Plateau by Using Satellite Remote Sensing Data

Yaoming MA

Cold and Arid Regions Environmental and Engineering Research Institute, Chinese Academy of Sciences,Lanzhou, China

Institute of Tibetan Plateau Research, Chinese Academy of Sciences, Beijing, China

Hirohiko ISHIKAWA

Disaster Prevention Research Institute, Kyoto University, Kyoto, Japan

Osamu TSUKAMOTO

Faculty of Sciences, Okayama University, Okayama, Japan

Massimo MENENTI, Zhongbo SU

Alterra Green World Research, Wageningen, The Netherlands

Tandong YAO

Cold and Arid Regions Environmental and Engineering Research Institute, Chinese Academy of Sciences,Lanzhou, China

Institute of Tibetan Plateau Research, Chinese Academy of Sciences, Beijing, China

Toshio KOIKE

Department of Civil Engineering, University of Tokyo, Tokyo, Japan

and

Tetsuzo YASUNARI

Institute of Geoscience, University of Tsukuba, Tsukuba, Japan

(Manuscript received 29 March 2002, in revised form 9 December 2002)

Corresponding author: Yaoming Ma, Departmentof Plateau Atmospheric Physics, Cold and AridRegions Environmental and Engineering ResearchInstitute (CAREERI), Chinese Academy of Sci-ences (CAS), 260 Dong-gang West Road, Lanzhou,Gansu 730000, China.E-mail: [email protected]( 2003, Meteorological Society of Japan

Abstract

In this study, a parameterization method based on NOAA-14/AVHRR data and field observations isdescribed and tested for deriving the regional land surface variables, vegetation variables and land sur-face heat fluxes over a heterogeneous landscape. As a case study, the method was applied to the TibetanPlateau area. The regional distribution maps of surface reflectance, MSAVI, vegetation coverage, surfacetemperature, net radiation, soil heat flux, sensible heat flux and latent heat flux were determined overthe Tibetan Plateau area. The derived results were validated by using the ‘‘ground truth’’. The resultsshow that the more reasonable regional distributions and their seasonal variations of land surface vari-ables (surface reflectance, surface temperature), vegetation variables (MSAVI and vegetation coverage),net radiation, soil heat flux and sensible heat flux can be obtained by using the method proposed in thisstudy. However, the approach of deriving regional latent heat flux, and their seasonal variation as theresidual of the energy budget, may not be a good method due to the unbalance of energy and the strongadvection over the study area. Further improvement of the method was also discussed.

1. Introduction

The Tibetan Plateau, with one million km2

area and the averaged altitude of about4000 m, plays a very important role in theAsian monsoon circulation, and the globalclimate change. The GEWEX Asian MonsoonExperiment over the Tibetan Plateau, GAME-Tibet, was carried out in the Tibetan Plateau,and the intensive observation period (IOP) wascontinued from May to September 1998. Theexperimental region, about 100� 200 km2, in-cludes a variety of land surfaces such as a largearea of grassy marshland, some arid areas,many small rivers and several lakes. At dif-ferent landscapes, two basic comprehensiveobservation stations (Anduo and NaquFx),two Flux-PAM (Portable Automated Meso-net)observation stations (MS3478-NPAM andMS3637-SPAM), three automatic weather sta-tions (D110-AWS110, Naqu and MS3608-AWS3608), one 3D Doppler Radar, radio sondesystem, soil moisture and soil temperature net,rain gauge net and barometer net had been op-erated continuously for almost five months. Alarge amount of surface observation data hasbeen collected (Fig. 1).

The study on the energy exchanges betweenthe land surface and atmosphere was of para-mount importance for GAME-Tibet. Some in-teresting detailed studies concerning the landsurface heat fluxes have been reported (Ishi-kawa et al. 1999; Tamagawa et al. 1999; Tsu-kamoto et al. 1999; Wang et al. 1999; Ma et al.1999a; Ma et al. 1999c; Yasunari 1999; Kuwa-gata et al. 1999; Ma et al. 2000; Koike 2000;Tanaka et al. 2001). These researches were,

however, on point-level or a local-patch-level.Since the aerial, but not only point-wise, infor-mation of land-surface atmosphere interactionis required, the aggregation of the individualresults into a regional scale is necessary. Re-mote sensing from satellites offers the possibil-ity to derive regional distribution of land sur-face heat fluxes.

The purpose of this study is to upscale thepoint or patch scale field observations of landsurface variables, and land surface heat fluxesto meso-scale distribution of them with the aidof NOAA-14/AVHRR data. First describe themethodology in section 2. The application ofthe methodology to the GAME/Tibet period ispresented in section 3, where the distributionof land surface variables, vegetation varia-bles, and land surface heat fluxes are estimatedfor three different phases, pre-monsoon, mid-monsoon, and post-monsoon. Discussions arealso given in the section.

2. Data and methodology

2.1 DataThe NOAA-14 Advanced Very High Resolu-

tion Radiometer (AVHRR) provides spectral in-formation in 5 bands, with a spatial resolutionof about 1 km� 1 km. Three scenes of satellitedata used in this study were collected at14:43 h (Local Time at 92�E, LT) June 12,1998, 13:21 h (LT) 16 July, 1998 and 13:25 h(LT), August 21, 1998.

The most relevant field data, collected at theGAME/Tibet surface stations, consists of radio-sounding, vertical profiles of air temperature,wind speed and humidity, soil heat flux, surface

Journal of the Meteorological Society of Japan278 Vol. 81, No. 2

radiation budget components, turbulent fluxesmeasured by eddy-correlation technique andPBL tower, and the vegetation state.

2.2 MethodologyThe general concept of the methodology is

shown in a diagram (Fig. 2). The surface re-flectance for short-wave radiation (r0), and land

surface temperature (Tsfc), are retrieved fromNOAA-14/AVHRR data with the atmosphericcorrection by radiative transfer model MOD-TRAN (Berk et al. 1989) using aero-logical ob-servation data. The radiative transfer modelalso computes the downward short- and long-wave radiation at the surface. With these re-sults the surface net radiation (Rn) is deter-

Fig. 1. The geographic map and the sites layout during IOP of the GAME/Tibet.

Y. MA, et al. 279April 2002

mined. The soil heat flux (G0) is estimated fromRn, Tsfc, r0 and MSAVI (Modified Soil AdjustedVegetation Index, Qi et al. 1994) which is alsoderived from NOAA-14/AVHRR data. The sen-sible heat flux (H ) is estimated from Tsfc, sur-face and aero-logical data with the aid of socalled ‘blending height’ approach (Mason 1988).

a. Net radiationThe regional net radiation flux is expressed

as

Rnðx; yÞ ¼ ð1� r0ðx; yÞÞ � K#ðx; yÞ þ L#ðx; yÞ

� e0ðx; yÞsT4sfcðx; yÞ; ð1Þ

where surface reflectance r0ðx; yÞ is derivedfrom the channel-1 and -2 of NOAA-14/AVHRRdata with the models of Paltridge and Mith-chell (1990) and Valiente et al. (1995). The sur-face emissivity e0ðx; yÞ is a function of vegeta-tion coverage, and the vegetation coverage Pv isalso derived from channel 1 and 2 data with thealgorithm by Valor and Caselles (1996), i.e.,

Pvðx; yÞ

¼1�NDVIðx; yÞ

NDVIs

1�NDVIðx; yÞNDVIs

� �� K 1�NDVIðx; yÞ

NDVIv

� � ;

ð2Þ

where K ¼ ðr2v � r1vÞ/ðr2g � r1gÞ, which r2v, r1vand r2g, r1g are the minimum and maximumreflectance value of AVHRR channel-2 and

channel-1. NDVIs and NDVIv are the NDVIvalues for bare soil and full vegetation respec-tively.

The land surface temperature Tsfcðx; yÞ inEq. (1) is retrieved from the brightness tem-perature of channels four and five of NOAA-14/AVHRR according to Becker and Li (1990 and1995). We first assume that the Tsfc is ex-pressed as

Tsfc ¼ FðT4;T5; e4; e5;W; yÞ; ð3Þ

where T4 and T5 are the brightness temper-atures of channels 4 and 5 of AVHRR, e4 and e5are the spectral emissivities of channel 4 and 5respectively, W is water vapor content, whichcan be derived from MODTRAN (Kneizys et al.1996) model by using air temperature andhumidity profiles observed through the radiosonde, and y represents the view angle of satel-lite. Equation (3) was expressed by differentsplit window algorithms (Becker and Li 1995).The algorithm proposed by Sobrino and Rais-souni (2000) will be used in this study, i.e.,

Tsfcðx; yÞ ¼ T4ðx; yÞ þ 1:40½T4ðx; yÞ � T5ðx; yÞ�

þ 0:28½T4ðx; yÞ � T5ðx; yÞ�2

þ 0:83þ ð57� 5WÞð1� eÞ

� ð161� 30WÞDe; ð4Þ

where e ¼ ðe4 þ e5Þ/2, De ¼ e4 � e5 (Li andBecker 1993).

The incoming short-wave radiation fluxK#ðx; yÞ in Eq. (1) could be derived from radia-tive transfer model MODTRAN (Kneizys et al.1996), where atmospheric short-wave trans-mittance tsw is obtained by using the radiosonde data and the surface reflectance and sur-face temperature observed in the field. HenceK#ðx; yÞ can be obtained as

K#ðx; yÞ ¼ tswK#TOAðx; yÞ; ð5Þ

where the regional variation of radiationflux perpendicular to the top of atmosphereK#

TOAðx; yÞ is a spectrally integrated form of in-band radiation flux perpendicular to the top ofatmosphere K#

TOAðlÞ, and

K#TOAðx; yÞ ¼

K#exoðbÞ cos ysunðx; yÞ

d2s

; ð6Þ

where K#exoðbÞ is the averaged in-band solar exo-

atmospheric irradiance undisturbed by ysun be-

Fig. 2. Diagram of parameterization pro-cedure combining NOAA-14/AVHRRdata with field observations.


ing zero, b is abbreviation of in-band, ds is theearth-sun distance, ysun is sun zenith angle.The incoming long-wave radiation flux L#ðx; yÞin Eq. (1) could also be calculated from MOD-TRAN by using the radio sonde data and thesurface reflectance and surface temperatureobserved in the field.

b. Soil heat fluxThe regional soil heat flux G0ðx; yÞ is usually

determined by (Choudhury and Monteith 1988)

G0ðx; yÞ ¼ rsCs½ðTsfcðx; yÞ � Tsðx; yÞ�/rshðx; yÞ;ð7Þ

where rs is soil dry bulk density, Cs is soil spe-cific heat, Tsðx; yÞ stands for soil temperatureof a determined depth, rshðx; yÞ represents soilheat transportation resistance. However, theregional soil heat flux G0ðx; yÞ cannot directlybe mapped from satellite observations throughEquation (7) for the difficulty to determinethe soil heat transportation resistance rshðx; yÞ,and the soil temperature at a reference depthTsðx; yÞ. Many investigations have shown thatthe mid-day G0/Rn fraction is reasonably pre-dicted from special vegetation indices (Daugh-try et al. 1990). Some researchers have shownthat G0/Rn ¼ GðNDVIÞ (Clothier et al. 1986;Choudhury et al. 1987; Kustas and Daughtry1990). An improved fraction of G0/Rn ¼Gðr0;Tsfc;NDVIÞ was proposed (Menenti et al.1991; Bastiaanssen 1995). However, problemsexist in the NDVI definition equation becauseof the effects of external factors, such as soilbackground variations (Huete et al. 1985;Huete 1989). In order to reduce the soil back-ground effect in NDVI, a parameterizationbased on MSAVI is proposed over the Tibetanarea in this study as

G0ðx; yÞ ¼ Rnðx; yÞ � ðTsfcðx; yÞ/r0ðx; yÞÞ

� ðaþ br0 þ cr20Þ

� ½1þ dMSAVIðx; yÞe�; ð8Þ

where the constants a, b, c, d and e are de-termined by using the field data observed atsix observation stations (AWS110, Anduo,NPAM, Naqu, AWS3608 and SPAM) duringthe GAME/Tibet IOP; r0 is a daily mean re-flectance value obtained from field observa-tions. MSAVIðx; yÞ was derived from the bandreflectance of channel 1 and 2 of NOAA-14/

AVHRR as (Qi et al. 1994)

MSAVIðx; yÞ

¼2r2ðx; yÞ þ 1�

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi½2r2ðx; yÞ þ 1�2 � 8½r2ðx; yÞ � r1ðx; yÞ�

q2

:

ð9Þ

c. Sensible and latent heat fluxesThe sensible heat flux Hðx; yÞ can be derived

from

Hðx; yÞ ¼ rCPTsfcðx; yÞ � Taðx; yÞ

raðx; yÞ; ð10Þ

where aerodynamic resistance raðx; yÞ is

raðx; yÞ ¼1

ku�ðx; yÞ

� lnz� d0ðx; yÞZ0mðx; yÞ

� �þ kB�1ðx; yÞ � chðx; yÞ

� �;

ð11Þ

and

u�ðx; yÞ

¼ kuðx; yÞ lnz� d0ðx; yÞZ0mðx; yÞ

� �� cmðx; yÞ

� ��1

;

ð12Þ

where k is the Von-Karman constant, u� is thefriction velocity, z is reference height, d0 iszero-plane displacement height, Z0m is the ef-fective aerodynamic roughness, cm and ch arethe stability correction function, and kB�1 isthe excess resistance to heat transfer, and(Owen and Thomson 1963; Chamberlain 1968)

kB�1 ¼ lnz0mz0h

� �; ð13Þ

where z0m and z0h are aerodynamic roughnessand thermodynamic roughness, respectively.

Combining Eqs. (10), (11) and (12) yields

Hðx; yÞ ¼ rCPk2uðx; yÞ

� ½Tsfcðx; yÞ � Taðx; yÞ�

lnz� d0ðx; yÞZ0mðx; yÞ

þ kB�1ðx; yÞ � chðx; yÞ� �

� lnz� d0ðx; yÞZ0mðx; yÞ

� cmðx; yÞ� � :

ð14Þ

The straightforward way to model sensibleheat flux in a large area is to sum up the con-tribution from different surface elements. If thelocal scale advection is comparatively small, itis desired that the development of a convective


https://www.researchgate.net/publication/23976802_A_4-layer_model_for_the_heat_budget_of_homogenous_land_surfaces?el=1_x_8&enrichId=rgreq-0aced942-a55e-42af-ba13-1e03f578f1d9&enrichSource=Y292ZXJQYWdlOzI1MDEzOTk3MjtBUzoxMDQ2MDA0MDk0NzcxMjFAMTQwMTk1MDA4OTIxMw==

boundary layer may smooth the local heteroge-neity of surface disorganized variety at theso called ‘blending height’, where atmosphericcharacteristics become proximately indepen-dent of horizontal locations. The corresponding‘effective’ surface variables can be determinedaccordingly (Mason 1988). This approachhas been proved to be successful to calculateregional averaged surface fluxes recently(Lhomme et al. 1994; Bastiaassen 1995; Wanget al. 1995; Ma et al. 1999b; Ma et al. 2002b).Based on this approach, the regional sensibleheat flux Hðx; yÞ is expressed as

Hðx; yÞ ¼ rCPk2uB

� ½Tsfcðx; yÞ � Tair-B�

lnZB � d0ðx; yÞZ0mðx; yÞ

þ kB�1ðx; yÞ � chðx; yÞ� �

� lnZB � d0ðx; yÞZ0mðx; yÞ

� cmðx; yÞ� � ;

ð15Þ

where ZB is the blending height, uB and Tair-B

are wind speed and air temperature at theblending height respectively. ZB, uB and Tair-B

are determined by using field measurements ornumerical models. In this study, these varia-bles will be determined with the aid of fieldmeasurements of radio sonde. Z0mðx; yÞ is theeffective aerodynamic roughness length includ-ing the effect of topography and low vegetation(e.g., grass), and is determined by the Taylor’smodel (Taylor et al. 1989). The excess resis-tance to heat transfer, kB�1, is shown as afunction of surface temperature over the Tibe-tan Plateau area (Ma et al. 2002a). d0 is zero-plane displacement, which can be calculatedfrom Raupach’s model (Raupach 1994) overthis area. chðx; yÞ and cmðx; yÞ are the in-tegrated stability functions in equation (15).For unstable condition, the integrated stabilityfunctions chðx; yÞ and cmðx; yÞ are written as(Paulson 1970)

cmðx; yÞ ¼ 2 ln1þ X

2

� �þ ln

1þ X 2

2

� �

� 2 arctanðXÞ þ 0:5p;

chðx; yÞ ¼ 2 ln1þ X 2

2

� �;

8>>>>>><>>>>>>:

ð16Þ

where X ¼ f1� 16 � ½z� d0ðx; yÞ�/Lðx; yÞg0:25and Monin Obukhov stability length L ¼�kgH/ðTau

3�rCPÞ. For stable condition, the in-

tegrated stability function chðx; yÞ and cmðx; yÞbecome (Webb 1970)

cmðx; yÞ ¼ chðx; yÞ ¼ �5z� d0ðx; yÞ

Lðx; yÞ : ð17Þ

The stability function ½z� d0ðx; yÞ�/Lðx; yÞ willbe solved by using the Businger scheme (Bu-singer 1988), i.e.,

z� d0ðx; yÞLðx; yÞ ¼ Riðx; yÞ ðunstableÞ;

z� d0ðx; yÞLðx; yÞ ¼ Riðx; yÞ/½1� 5:2Riðx; yÞ�

ðstableÞ;

8>>>>>><>>>>>>:

ð18Þ

where Riðx; yÞ is the Richardson number.The regional latent heat flux lEðx; yÞ is the

residual of the energy budget theorem for landsurface, i.e.,

lEðx; yÞ ¼ Rnðx; yÞ �Hðx; yÞ �G0ðx; yÞ: ð19Þ

3. Results and discussions

It is better to select the satellite data in cleardays to study the distribution of land surfacevariables, vegetation variables and the energybudget components. Unfortunately, it was diffi-cult to select this kind of satellite data over theTibetan Plateau area for the strong convec-tive clouds when NOAA-14/AVHRR observa-tion took place. Only three scenes of the NOAA-14/AVHRR could be selected during the wholeIOP. The scene of June 12, 1998 was selected asa case of pre-monsoon and whole mesoscalearea. The scenes of July 16, 1998 and August21, 1998 were selected as the cases of mid-monsoon and the post-monsoon. The imagesaround Anduo station and NPAM stationwere selected as the comparable areas becausefluxes measurements using sonic anemometer-thermometer were undertook at these two sta-tions. It is also very clear around these twostations on the images of NOAA-14/AVHRR.

Figure 3 shows the distribution maps of sur-face reflectance, surface temperature, vegeta-tion coverage, MSAVI and surface heat fluxesof the mesoscale experimental area. Each pixelis 1� 1 km2 and 180 by 51 pixels are shown.Figure 4 shows their frequency distribution ofthese variables over the whole mesoscale ex-perimental area. The distribution of land sur-face variables and vegetation variables aroundAnduo station and NPAM station were com-pared for different phases of the monsoon inFig. 5. The distribution of land surface heat


fluxes around Anduo station and NPAM stationwere also compared in Fig. 6. Both Fig. 5 andFig. 6 are based on 45 by 40 pixels with a size of1� 1 km2.

The land surface variables and land surfaceheat fluxes derived from satellite data werecompared with the field measurements at An-duo and NPAM sites. They are shown in Fig. 7

Fig. 3. The distribution maps of land surface variables, vegetation variables and land surface heatfluxes in June 12, 1998 for the GAME/Tibet area.


and Table 1. The field observational data,which used for validation here, was just mea-sured at the time of satellite over passed thearea, or ten minutes averaging value around

that time. The mean absolute percent differ-ence (MAPD) was computed as a quantitativemeasure of the difference between the derivedresults on No. i point (HderivedðiÞ), and mea-

Fig. 4. The frequency distribution of land surface variables, vegetation variables, and land surfaceheat fluxes for the GAME/Tibet area. (June 12, 1998).


Fig. 5. The distribution maps of land variables and vegetation variables for the Tibetan Plateauarea.


Fig. 6. The distribution maps of land surface heat fluxes for the Tibetan Plateau area.


sured value on No. i point (HmeasuredðiÞ) of onescene as

MAPD ¼ 100

n

Xni¼1

jHderivedðiÞ �HmeasuredðiÞjHmeasuredðiÞ

� �:

ð20Þ

It is seen that: (1) the derived land surfacevariables, vegetation variables and heat fluxes

for the whole meaoscale area on 12 June 1998were in good accordance with the land surfacestatus. These parameters show a wide rangeof variations due to the strong contrast of sur-face features in the study area. MSAVI variesfrom 0.00 to 0.30. Vegetation coverage Pv variesfrom 0.00 to 1.00. Surface reflectance is from0.00 to 0.35 (some of the surface reflectancewith value large than 0.35 indicating cloud-

Fig. 7. Comparison of the derived results with the field measurements for the surface reflectance,surface temperature and land surface heat fluxes over the GAME/Tibet area, together with 1 :1line.


Table 1. Comparison of the derived results (Cal.) versus those measured values (Meas.) at theGAME/Tibet site with MAPD


covering). Surface temperature ranged from5�C to 48�C (surface temperature of 0.0�C in-dicating cloud-covering). Net radiation changedfrom 400 to 920 W/m2. Soil heat flux variedfrom 0 to 150 W/m2. Sensible heat flux is from0 to 320 W/m2, and latent heat flux varies from40 to 880 W/m2 (see Fig. 1, Fig. 3 and Fig. 4);(2) not only on June 12, but also on July 16 andAugust 21, the derived surface reflectance, sur-face temperature, net radiation flux, soil heatflux and sensible heat flux were close to thefield measurements. The difference betweenthe derived results and the field observationMAPD was less than 10% (see Table 2, Fig. 5,Fig. 6, Fig. 7 and MASVI case in the Table 1);(3) not only on June 12, but also on July 16 andAugust 2, the value of the vegetation coveragein this area was almost the same due to theland surface in this area is covered by the samegrassy marshland in these days. The only dif-ference was that the grass was dry on June 12(with small MSAVI). It became green and weton July 16 and August 21 (with higher MSAVI)for this area (see Fig. 5 and Fig. 6); (4) duringthe experimental periods, the derived net radi-ation flux was larger than that in the HEIFEarea (Ma et al. 1999) due to the high altitude(the higher value of downward short-wave ra-diation), and land surface coverage of grassymarshland (the lower value of the upward long-wave radiation) in this area (see Fig. 3, Fig. 5and Fig. 6). For example, the regional averagevalue of net radiation flux was 470 W/m2 overthe HEIFE area in 9 July, 1991 and that was750 W/m2 over the GAME/Tibet area in 16July, 1998; (5) the values of surface reflectance,surface temperature, soil heat flux and sensibleheat flux in June over this area were largerthan these values in July and August. Net ra-diation flux and latent heat flux in June werelower than their values in July and August.The reason is that June 12 was the day beforethe Asia Monsoon coming. The land surfacewas dry in that day. July 16 and August 21were within and after the Asia Monsoon. Theland surface was wet, and the grass was highand growing (see Fig. 3, Fig. 5 and Fig. 6); (6)problems existed in the NDVI definition equa-tion because of the external factor effect, suchas soil background variations (Huete et al.1985; Huete 1989). To reduce the soil back-ground effect, Qi et al. (1994) proposed using

MSAVI. Therefore, the parameterizationmethod based on MSAVI for soil heat flux isbetter than that based on NDVI on heteroge-neous land surface of the Tibetan Plateau. Thederived regional soil heat fluxes, based onMSAVI were reasonable in different months inthis area with MAPD less than 10% (see Fig. 3,Fig. 6, Fig. 7 and Table 1), and it was betterthan the derived regional soil heat fluxes basedon NDVI (see Table 1); and, (7) all elementsof heat balance equation at NPAM site on June12 well corresponded to the satellite data. Onthe other hand, all but latent heat flux corre-sponded to the satellite data in other seasons,and other stations. The conclusions derivedfrom above facts were, (a) net radiation, sensi-ble heat flux, and soil heat flux could be derivedfrom satellite, (b) in the case of NPAM on June12, because the surface energy balanced in thesurface observation so that the latent heat fluxestimated by surface observation, well corre-sponded to that estimated by the residual ofsatellite data analysis. One dimensional energybudget did not balance due to large errors ofthe latent heat flux through the surface obser-vation and advection in this area. The largeerror of the measurement on latent heat fluxmay dpend on the accuracy of the turbulencemeasurement sensors (Ma et al. 1999a; Ishi-kawa et al. 1999; Wang et al. 1999; Ma et al.1999c; Ma et al. 2000; Tanaka et al. 2001). Thispoint and conclusion (a) clearly show the dis-agreement of latent heat flux between surfaceobservations and satellite one.

4. Concluding remarks

In this study, distributions of land surfacevariables (surface reflectance and surface tem-perature), vegetation variables (MSAVI andvegetation coverage Pv), land surface heatfluxes (net radiation, soil heat flux and sensi-ble heat flux) over the heterogeneous area ofGAME/Tibet were derived by using NOAA-14/AVHRR data and field observations. The re-sults were in good agreement with field obser-vations. The approach of deriving regional la-tent heat flux, and their seasonal variation asthe residual of the energy budget, may not be agood method due to one-dimensional energybudget unbalance and the strong advectionover the study area when the satellite passedover the area. Future improvements are to


Table 2. List of symbols in this paper


be made to derive more accurate regional la-tent heat flux over such areas. The advectionqSðx; yÞ/qt and the errors in the procedure ofdetermining regional net radiation, soil heatflux and sensible heat flux DRnðx; yÞ, DG0ðx; yÞand DHðx; yÞ will be considered in the equationof surface energy balance over this area, e.g.,

lEðx; yÞ ¼ Rnðx; yÞ þ DRnðx; yÞ

� ½Hðx; yÞ þ DHðx; yÞ�

� ½G0ðx; yÞ þ DG0ðx; yÞ�

� qSðx; yÞ/qt: ð21Þ

In other words, the regional latent heat fluxwill be correctly derived when the advectionqSðx; yÞ/qt and the errors DRnðx; yÞ, DG0ðx; yÞand DHðx; yÞ are determined by using the suit-able models and much better measurementsensors.

It is also worth trying SEBI (Surface En-ergy Balance Index, Menenti and Choudhury1993) method, which is based on the Penmen-Monteith equation (Monteith 1965).

Acknowledgement

The data used in this paper was obtained bythe GAME-Tibet project supported by the Min-istry of Education, Science, Sport and Cultureof Japan; the Science and Technology Agency ofJapan; the Chinese Academy of Science (CAS);the National Space Development Agency ofJapan, and the Frontier Research System forGlobal Change. This research is under the aus-pices of National Science Foundation of Chinaunder Project No. 40275003, the Chinese Na-tional Key Project (G1998040900), the In-novation Project of the Chinese Academy ofScience (KZCX2-301) and the Innovation proj-ect of Cold and Arid Regions Environmentaland Engineering Research Institute, CAS(CACX210072). The first author would also liketo acknowledge Dr. Zhaoliang Li for his verykindly help in the procedure of determiningsurface temperature, Prof. J. Wang, Dr. K.Ueno and Dr. J. Wen for their help in the pro-cedure of the paper. The authors thank all theparticipants from China and Japan in the fieldobservation. We are also grateful to the refer-ees for their very useful comments and helpfulsuggestions, which contributed to the final ver-sion of the manuscript.

References

Bastiaanssen, W.G.M., 1995: Regionalization of sur-face fluxes and moisture indicators in compos-ite terrain, PhD Thesis, Wageningen Agricul-tural University, 273pp.

Becker, F. and Zh. Li, 1990: Towards a local splitwindow method over land surfaces, Int. J. ofRemote Sensing, 11(3), 369–393.

——— and ———, 1995: Surface temperature andemissivity at various scales: definition, mea-surement and related problems, Remote Sens-ing Review, 12, 225–253.

Berk, A., L.S. Bernstein and D.C. Robertson, MOD-TRAN, 1989: A moderate resolution model forLOTRAN 7, GL-TR-89-0122.

Businger, J.A., 1988: A note on the Businger-Dyerprofiles, Bound.-Layer Meteor., 42, 145–151.

Chamberlain, A.C., 1968: Transport of gases to andfrom surfaces with bluff and wave-like rough-ness elements, Quart. J. Roy. Meteor. Soc., 94,318–332.

Choudhury, B.J. and J.L. Monteith, 1988: A four-layer model for the heat budget of homoge-neous land surfaces, Quart. J. Roy. Meteor.Soc., 114, 373–398.

———, S.B. Idso and R.J. Reginato, 1987: Analysisof an empirical model for soil heat flux under agrowing wheat crop for estimating evaporationby infrared-temperature based energy balanceequation, Agric. For. Meteor., 39, 283–297.

Clothier, B.E., K.L. Clawson, P.J. Pinter, M.S.Moran, R.J. Reginato and R.D. Jackson, 1986:Estimating of soil heat flux from net radiationduring the growth of alfalfa, Agric. For. Me-teor., 37, 319–329.

Daughtry, C.S.T., W.P. Kustas, M.S. Moran, P.J.Pinter, R.D. Jackson, P.W. Brown, W.D. Nich-ols and L.W. Gay, 1990: Spectral estimates ofnet radiation and soil heat flux, Remote Sens.Environ., 32, 111–124.

Huete, A.R., R.D. Jackson and D.F. Post, 1985:Spectral response of a plant canopy with dif-ferent soil backgrounds, Remote Sens. Envi-ron., 17, 37–53.

———, 1989: Soil influences in remotely sensed veg-etation-canopy spectra. Theory and Applica-tions of Optical Remote Sensing (G. Asrar, Ed.),107–141.

Ishikawa, H., O. Tsukamoto, T. Hayashi, I. Ta-magawa, S. Miyazaki, J. Asanuma, Y. Qi, H. Fu-deyasu and K. Tanaka, 1999: Summary of theboundary layer observation and the prelimi-nary analysis, Third International ScientificConference on the Global Energy and WaterCycle, Beijing, China, 16–19 June 1999, 45–459.


Kenizys, F.X., L.W. Abreu, G.P. Anderson, J.H.Chetwynd, E.P. Shettle, A. Berk, L.S. Bern-stein, D.C. Robertson, P. Acharya, L.S. Roth-man, J.E.A. Selby, W.O. Gallery and S.A.Clough, 1996: The MODTRAN3/2 report andLODTRAN 7 Model. (L.W. Abreu and G.P.Andson, Eds), prepared by Ontar Corp., NorthAndover, MA, for Phillips Laboratory, Geo-physical Directorate, Hanscom AFB, MA.,Contract No. F19628-91-C-0132.

Koike, T., 2000: The overview of GAME/Tibet, TheSecond Session of International Workshop onTIPEX-GAME/TIBET, Kunming, China, July20–22.

Kustas, W.P. and C.S.T. Daughtry, 1990: Estimationof the soil heat flux/net radiation ratio fromspectral data, Agric. For. Meteor., 39, 205–223.

Kuwagata, T., H. Kanno, J. Asanuma, Y. Ma andX. Ma, 1999: Preliminary estimation of thedaytime heating rate of the atmosphere overthe Tibetan Plateau during ’98 IOP usingthe diurnal variation of the surface pressure,Third International Scientific Conference on theGlobal Energy and Water Cycle, Beijing, China,16–19 June 1999, 417.

Lhomme, J.-P., A. Chehbouni and B. Monteny, 1994:Effective parameters of surface energy balancein heterogeneous landscape, Bound.-Layer Me-teor., 71(3), 297–310.

Li, Z.-L. and F. Becker, 1993: Feasibility of land sur-face temperature and emissivity determinationfrom AVHRR data, Remote Sens. Environ., 43,67–85.

Ma, Y., O. Tsukamoto, J. Wang, Z. Hu, H. Ishikawa,Z. Hu, I. Tamagawa and H. Gao, 1999a:The characteristics of micrometeorology in thenorthern Tibetan Plateau area, Proceedingsof the 1st International Workshop on GAME-Tibet, Xi’an, China, 11–13 January 1999, 99–102.

———, J. Wang, M. Menenti and W. Bastiaanssen,1999b: Estimation of fluxes over the heteroge-neous land surface with the aid of satelliteremote sensing and field observation, ACTAMeteor. Sinica, 57, 180–189 (in Chinese withEnglish abstract).

———, O. Tsukamoto, J. Wang, Z. Hu, H. Ishikawa,I. Tamagawa and H. Gao, 1999c: Transfer andmicrometeorological characteristics in the sur-face layer of the atmosphere above TibetanPlateau area, Third International ScientificConference on the Global Energy and WaterCycle, Beijing, China, 16–19 June 1999, 55–56.

———, O. Tsukamoto, X. Wu, I. Tamagawa, J.Wang, H. Ishikawa, Z. Hu and H. Gao, 2000:

Characteristics of energy transfer and micro-meteorology in the surface layer of the atmo-sphere above marshland of the Tibetan Pla-teau area, Chinese Journal of AtmosphericSciences, 24(5), 715–722 (in Chinese with En-glish abstract).

———, O. Tsukamoto, J. Wang, H. Ishikawa andI. Tamagawa, 2002a: Analysis of aerodynamicand thermodynamic parameters on the grassymarshland surface of Tibetan Plateau, Prog-ress in Natural Science, 12(1), 36–40.

———, O. Tsukamoto, H. Ishikawa, Zh. Su, M. Me-nenti, J. Wang and J. Wen, 2002b: Determina-tion of Regional land surface heat flux den-sities over heterogeneous landscape of HEIFEIntegrating satellite remote sensing with fieldobservations, J. Meteor. Soc. Japan, 80(3),485–501.

Mason, P., 1988: The formation of areally averagedroughness lenghths, Quart. J. Roy. Meteor.Soc., 114, 399–420.

Menenti, M., W.G.M. Bastiaanssen, K. Hefny andM.H. Abd EI Karim, 1991: Mapping of groundwater losses by evaporation in the WesternDesert of Egypt, DLO Winand Staring Centre,Report no. 43, Wageningen, The Netherlands,116pp.

——— and B.J. Choudhury, 1993: Parameteraiza-tion of land surface evaporation by means oflocation dependent potential evaporation andsurface temperature range, in (eds.) Bolle,Feddes and Kalma, Exchange processes at theland surface for a range of space and timescales, IAHS Publ. No. 212, 561–568.

Monteith, J.L., 1965: Evaporation and environ-ment, XIXth Symposium, Soc. For Exp. Biol.,Swansea, Cambridge University Press 1, 205–234.

Owen, P.R. and W.R. Thomson, 1963: Heat transferacross rough surface, J. Fluid Mech., 15, 321–334.

Paltridge, W. and Mitchell, M. Ross, 1990: Atmo-spheric and viewing angle correction of vege-tation indices and grassland fuel moisturecontent derived from NOAA/AVHRR, RemoteSens. Environ., 31, 121–135.

Paulson, C.A., 1970: The mathematic representationof wind speed and temperature profiles in theunstable atmospheric surface layer, J. Meteor.Soc. Japan, 9, 856–861.

Qi, J., A. Chehbouni, A.R. Huete, Y.H. Kerr and S.Sorooshian, 1994: A Modified Soil AdjustedVegetation Index, Remote Sens. Environ., 48,119–126.

Raupach, M.R., 1994: Simplified experissions forvegetation roughness length and zero-planedisplacements as functions of canopy height


and area index, Bound.-Layer Meteor., 71,211–216.

Sobrino, J.A. and N. Raissouni, 2000: Toward remotesensing methods for land cover dynamic mon-itoring, application to Morocco, Int. J. of Re-mote Sensing, 21(2), 353–366.

Tamagawa, I., H. Ishikawa, O. Tsukamoto, T. Ha-yashi, S. Miyazaki, J. Asanuma, Y. Qi, H.Fudeyasu, K. Tanaka, Y. Ma, H. Gao andJ. Wang, 1999: Preliminary analysis on theturbulent characteristics at Amdo PBL site onTibetan Plateau, Third International ScientificConference on the Global Energy and WaterCycle, Beijing, China, 16–19 June 1999, 413–414.

Tanaka, K., H. Ishikawa, T. Hayashi, I. Tamagawaand Y. Ma, 2001: Surface Energy Budget atAmdo on Tibetan Plateau using GAME/TibetIOP’98 Data, J. Meteor. Soc. Japan, 79(1B)GAME Special Issue, 505–517.

Taylor, P.A., R.I. Sykes and P.J. Mason, 1989: On theparameterization of drag over small scale to-pography in neutrally stratified Boundary flow,Bound.-Layer Meteor., 48, 409–422.

Tsukamoto, O., H. Fudeyasu, S. Miyazaki, K. Ueno,Y. Qi and Y. Ma, 1999: Turbulent surface fluxmeasurements over Tibetan Plateau with flux-PAMsystem,Third International ScientificCon-ference on the Global Energy and Water Cycle,Beijing, China, 16–19 June 1999, 411–412.

Valiente, J.A., M. Nunez, E. Lopez-Baeza and J.F.Moreno, 1995: Narrow-band to broad-bandconversion for Meteosat-visible channel andbroad-band reflectance using both AVHRR-1and -2 channels, Int. J. Remote Sensing, 16(6),1147–1166.

Valor, E. and V. Caselles, 1996: Mapping land sur-face emissivity from NDVI: application to Eu-ropean, African and South American areas,Remote Sens. Environ., 57, 167–184.

Wang, J., Y. Ma, M. Menenti and W. Bastiaanssen,1995: The scaling-up of processes in the heter-ogeneous landscape of HEIFE with the aid ofsatellite remote sensing, J. Meteor. Soc. Japan,73(6), 1235–1244.

———, J. Kim, Y. Liou, Zh. Gao, Y. Yan, T. Choi andH. Lee, 1999: Energy balance analysis and one-dimensional simulation of land surface processin a short-grass site of Central Tibetan Pla-teau, Third International Scientific Conferenceon the Global Energy and Water Cycle, Beijing,China, 16–19 June 1999, 424–425.

Webb, E.K., 1970: Profile relationships: the log-linerrange and extension to strong stability, Quart.J. Roy. Meteor. Soc., 96, 67–90.

Yasunari, T., 1999: Overview and highlights of theGAME-IOP, 1998, Third International Scien-tific Conference on the Global Energy andWater Cycle, Beijing, China, 16–19 June 1999,402–403.


Regionalization of Surface Fluxes over Heterogeneous Landscape of the Tibetan Plateau by Using...

Documents

Transcript of Regionalization of Surface Fluxes over Heterogeneous Landscape of the Tibetan Plateau by Using...