Remote sensing parameterization of land surface heat fluxes over arid and semi-arid areas


Ma, Y.M.; Wang J.M.; Huang, R.H.; Wei G.A.; Menenti, M.; Su, Z.B.; Hu, Z.Y.; Gao, F.; Wen, J.

Advances in Atmospheric Sciences 20(4): 530-539

2003


Dealing with the regional land surfaces heat fluxes over inhomogeneous land surfaces in arid and semiarid areas is an important but not an easy issue. In this study, one parameterization method based on satellite remote sensing and field observations is proposed and tested for deriving the regional land surface heat fluxes over inhomogeneous landscapes. As a case study, the method is applied to the Dunhuang experimental area and the HEIFE (Heihe River Field Experiment, 1988-1994) area. The Dunhuang area is selected as a basic experimental area for the Chinese National Key Programme for Developing Basic Sciences: Research on the Formation Mechanism and Prediction Theory of Severe Climate Disaster in China (G1998040900, 1999-2003). The four scenes of Landsat TM data used in this study are 3 June 2000, 22 August 2000, and 29 January 2001 for the Dunhuang area and 9 July 1991 for the HEIFE area. The regional distributions of land surface variables, vegetation variables, and heat fluxes over inhomogeneous landscapes in arid and semi-arid areas are obtained in this study.

ADVANCES
IN
ATMOSPHERIC
SCIENCES,
VOL.
20,
NO.
4,
2003,
PP.
530-539
Remote
Sensing
Parameterization
of
Land
Surface
Heat
Fluxes
over
Arid
and
Semi
-arid
Areas
MA
Yaoming*"
(-'T
),
WANG
Jiemin
l
(I
*
-K),
HUANG
Ronghui
3
(A
VX),
WEI
Guoan
i
(RIM),
Massimo
MENENTI
4
,
SU
Zhongbo
4
(Vt'R),
HU
Zeyong
l
(
14g),
GAO
Feng
l
(4'
0)
,
and
WEN
Jun
4
(iC
y)
'Cold
and
Arid
Regions
Environmental
and
Engineering
Research
Institute,
Chinese
Academy
of
Sciences,
Lanzhou
730000
2
Institute
of
Tibetan
Plateau
Research,
Chinese
Academy
of
Sciences,
Beijing
100029
3
lnstitute
of
Atmospheric
Physics,
Chinese
Academy
of
Sciences,
Beijing
100029
4
Alterra
Green
World
Research,
Wageningen
UR,
6700
AA,
Wageningen,
The
Netherlands
(Received
July
28,
2002;
revised
March
17,
2003)
ABSTRACT
Dealing
with
the
regional
land
surfaces
heat
fl
uxes
over
inhomogeneous
land
surfaces
in
arid
and
semi-
arid
areas
is
an
important
but
not
an
easy
issue.
In
this
study,
one
parameterization
method
based
on
satellite
remote
sensing
and
fi
eld
observations
is
proposed
and
tested
for
deriving
the
regional
land
surface
heat
fl
uxes
over
inhomogeneous
landscapes.
As
a
case
study,
the
method
is
applied
to
the
Dunhuang
experimental
area
and
the
HEIFE
(Heihe
River
Field
Experiment,
1988-1994)
area.
The
Dunhuang
area
is
selected
as
a
basic
experimental
area
for
the
Chinese
National
Key
Programme
for
Developing
Basic
Sciences:
Research
on
the
Formation
Mechanism
and
Prediction
Theory
of
Severe
Climate
Disaster
in
China
(G1998040900,
1999-2003).
The
four
scenes
of
Landsat
TM
data
used
in
this
study
are 3
June
2000,
22
August
2000,
and
29
January
2001
for
the
Dunhuang
area
and
9
July
1991
for
the
HEIFE
area.
The
regional
distributions
of
land
surface
variables,
vegetation
variables,
and
heat
fl
uxes
over
inhomogeneous
landscapes
in
arid
and
semi
-arid
areas
are
obtained
in
this
study.
Key
words:
land
surface
heat
fl
ux,
arid
and
semi
-arid
area,
Landsat
TM,
fi
eld
observation
1.
Introduction
The
study
on
the
energy
exchanges
between
the
land
surface
and
atmosphere
is
of
paramount
impor-
tance
for
arid
and
semi
-arid
areas,
e.g.,
the
HEIFE
(Heihe
River
Field
Experiment)
area
and
Dunhuang
areas
in
northwestern
China.
Oasis,
Gobi,
sand
desert,
and
mountains
are
distributed
in
the
experimental
ar-
eas.
In
other
words,
the
experimental
areas
are
typi-
cal
inhomogeneous
land
surfaces.
Some
interesting
de-
tailed
studies
concerning
the
land
surface
heat
fl
uxes
over
the
two
areas
have
been
reported
(Tsukamoto
et
al.,
1992;
Tsukamoto
et
al.,
1995;
Mitsuta
et
al.,
1995;
Hu
et
al.,
1994;
Maitani
et
al.,
1995;
Zhang
et
al.,
2001
and
Hu
et
al.,
2002).
This
research
was,
however,
on
a
point
level
or
a
local
-patch
level.
Since
the
areal,
and
not
only
point
-wise,
information
of
land
-surface
atmosphere
interaction
is
required,
the
aggregation
of
the
individual
results
into
a
regional
scale
is
necessary.
Remote
sensing
from
satellites
offers
the
possibility
to
derive
regional
distribution
of
land
surface
heat
fl
uxes.
The
purpose
of
this
study
is
to
upscale
the
point
or
patch
scale
fi
eld
observations
of
land
surface
variables
and
land
surface
heat
fl
uxes
to
a
regional
distribution
of
them
by
using
Landsat
TM
data
and
fi
eld
observa-
tions.
2.
Landsat
TM
data
and
fi
eld
observation
data
The
Landsat-5
Thematic
Mapper
(TM)
and
Landsat-7
Thematic
Mapper
provide
a
spectral
radi-
*E-mail:
ymmaAns.lzb.ac.en
NO.
4
MA
YAOMING,
WANG
JIEMIN,
HUANG
RONGHUI
ET
AL.
531
ance
in
seven
narrow
bands,
with
a
spatial
resolution
of
about
30x30
m
2
for
three
visible
bands
(Band
1,
2,
3)
and
three
near
-infrared
bands
(Band
4,
5,
7),
and
120
x
120
m
2
for
the
thermal
infrared
band
6
of
Landsat-5
TM
and
60
x
60
m
2
for
the
thermal
infrared
band
6
of
Landsat-7
TM.
The
four
TM
images
used
in
this
paper
are
at
1000
LST
3
June
2000
(beginning
of
summer),
22
August
2000
(end
of
summer),
and
29
January
2001
(winter)
over
the
Dunhuang
area,
and
1000
LST
7
July
1991
over
the
HEIFE
area.
The
most
relevant
data,
collected
at
Dunhuang
and
the
HEIFE
surface
stations
to
support
the
pa-
rameterization
of
land
surface
heat
fl
uxes
and
analy-
sis
of
TM
images,
consist
of
surface
radiation
budget
components,
surface
radiation
temperature,
surface
re-
fl
ectance,
vertical
profiles
of
air
temperature,
humid-
ity,
wind
speed,
and
direction
measured
at
the
PBL
towers,
Sodar,
radiosonde,
tether
-sonde
and
turbulent
fl
uxes
measured
by
the
eddy
-correlation
technique,
soil
heat
fl
ux,
soil
temperature
profiles,
soil
moisture
pro-
fi
les,
and
the
vegetation
state.
3.
Theory
and
scheme
By
combining
satellite
remote
sensing
(e.g.,
Land
-
sat
TM
data)
with
fi
eld
observations,
the
land
surface
M
SAV
I
heat
fl
uxes
over
an
inhomogeneous
land
surface
can
be
derived.
The
general
concept
of
the
methodology
is
shown
in
Fig.
1.
3.1
Net
radiation
The
regional
net
radiation
fl
ux
can
be
derived
from:
R
o
(x,
y)
=
[1
r
o
(x,y)]1-
(x
,
y)
L
1
,(x,
y)
E0
(x,
WaTs
zi
fe(x,Y)
(
1
)
where
r
o
(x,
y)
is
surface
reflectance.
It
can
be
derived
from
Landsat
TM
data
using
a
four-
stream
radiative
transfer
assumption
for
atmospheric
correction
in
so-
lar
spectral
bands
(Vehoef,
1997;
Ma,
2001).
Surface
temperature
T
sk
(x,y)
in
Eq.(1)
can
be
derived
from
Landsat
TM
band
-6
(10.2-12.5
/../m)
spectral
radiance
(Ma,
2001).
The
incoming
short
wave
radiation
fl
ux
and
incoming
long
wave
radiation
fl
ux
K
L
(x,y)
and
L
L
(x,
y)
in
Eq.(1)
can
be derived
from
the
radiative
transfer
model
MODTRAN
(Ma,
2001).
Surface
emis-
sivity
E
0
(x,
y)
is
determined
by
Valor
and
Caselles's
method
(1997):
(x,
y)
Ev(x,
Y)Pv(x,
y)
E
g
(x,
y)
[1
P
v
(x,
y)]
+
4
<
e
>
[1
P
v
(x,
y)]P
v
(x,
y)
,
L
andsatTM
data
vegetation
cove/age
M
ODTRAN
R
n
rrface
and
aemlogicaldata
LA
I
blending
height
\
\,....appmach
Fig.
1.
Diagram
of
the
parameterization
procedure
by
combining
Landsat
TM
data
with
fi
eld
observations.
(2)
532
ADVANCES
IN
ATMOSPHERIC
SCIENCES
VOL.
20
and
vegetation
coverage
by
Carlson
and
Ripley
(1997):
2
P
v
(x,
=
[INDv(x
y)
-TNDVmin
(
3
)
-TNDVmax
-TNDVmin
where
/
N
p
v
is
the
Normalized
Difference
Vegetation
Index
(NDVI)
and
_T
NDvnan
and
_T
NDvmax
are
the
NDVI
values
for
bare
soil
and
full
vegetation
respectively.
3.2
Soil
heat
flux
The
regional
soil
heat
fl
ux
G
o
(x
,
y)
is
determined
using
a
parameterization
based
on
Modified
Soil
Ad-
justed
Vegetation
Index
(MSAVI,
imsAv,
Qi
et
al.,
1994):
Go(x
,
y)
=
Rn(x
Y)[Tsfe(x
/ro(x
Y)]
x(a±
br
0
±
an)[1
+
dImsAv
(x
Y)
e
]
where
(4)
where
the
constants
a,
b,
c,
d,
and
e
are
determined
from
fi
eld
data
observed
at
the
HEIFE
and
Dunhuang
observation
stations;
T
o
is
a
daily
mean
reflectance
value,
i.e.
for
the
HEIFE
case:
G
o
(x
,
Tf
R
n
(x
,
y)
se
(X
y)
(0.00025
+
0.004367
0
ro(x,
y)
+0.0084570)
[1
0.979/msAv(x,
y)
i
]
(5)
For
the
Dunhuang
case:
G
o
(x,
y)
=
Rn(x,
y)Tsfe(x
Y)
(0.00028
+
0.004247
0
ro(x,
y)
+0.0087570)
[1
0.98
2
/msAv(x,
y)
4
]
(6)
2r
4
(x,
y)
+
1
02r4(x,
+
1]
2
8[r
4(x
,
y)
r3(x
Y)]
ImsAv(x
Y
)
=
(
7
)
2
where
r
3
and
r
4
are
the
band
reflectance
of
Landsat
TM
Band
-3
and
Band
-4
on
the
land
surface.
3.3
Sensible
heat
flux
The
regional
distribution
of
sensible
heat
fl
ux
can
be
estimated
from
[T
sfe
(x,
T
a
(x,
y)]
H
(x
,
pc
p
k
2
u(x
,
y)
(8)
[ln
zZo
n
,
d((°xY)
'
Y)
+
R
-1
(x
Y)
Y1(x
,
y)1
[1n
z
c)(x
Y)
'
1
1)m(x
Y)1
Z
ci
on,(x
Y)
To
simulate
sensible
heat
fl
ux
on
a
large
scale,
a
straightforward
method
is
to
scale
-up
or
aggregate
the
regional
sensible
fl
ux
by
a
weighted
average
of
the
con-
tributions
from
different
surface
elements,
based
on
the
principle
of
fl
ux
conservation.
A
method
of
"blend-
ing
height"
is
proposed
to
derive
the
regional
sensible
heat
fl
ux
in
this
study.
If
the
local
-scale
advection
is
comparatively
small
during
the
period
of
Landsat
TM
H
(x
,
pc
p
k
2
u
B
observation,
the
development
of
convection
boundary
layer
may
adjust
the
surface
-disorganized
variability
at
"blending
height"
where
the
atmospheric
characteris-
tics
become
proximately
independent
of
the
horizontal
position
(Mason,
1988).
Based
on
this
approach,
the
regional
sensible
heat
fl
ux
density
H
(x
,
y)
can
be
de-
scribed
as
[Tsfe(x,
y)
Ta(x,
y)]
,
z
B
y)
'Onz(x,
Y)1
,zs
do
(x,
y))
+
k13
-1
(x,
y)
7
/)h,(x,
y)1
[m
Zom(x, Zom(x,
Y)
where
ZB
is
the
blending
height
and,
u
B
is
the
wind
speed
at
the
blending
height.
ZB
and
u
B
can
be
de-
termined
by
fi
eld
measurements
or
numerical
mod-
els.
In
this
study,
they
are
determined
with
the
aid
of
fi
eld
measurements
of
radiosonde
(Dunhuang
area)
and
tethersonde
and
Sodar
(HEIFE
area).
T
a
(x,
y)
in
Eq.
(9)
is
the
regional
distribution
of
air
temperature
at
the
reference
height.
An
improved
interpolation
method
is
proposed
here
to
derive
the
regional
distri-
,
(
9
)
bution
of
air
temperature
over
the
oasis
-desert
system
of
HEIFE.
In
other
words,
the
regional
distribution
of
air
temperature
T
a
(x,
y)
over
the
HEIFE
area
can
be
derived
using
this
improved
numerical
interpolation
method
based
on
a
number
of
fi
eld
observations
of
air
temperature
and
regional
surface
temperature
as
(Ma
et
al.,
2002)
Ta(x,
T
sk
(x,
DT
a
(x
,
.
(10)
The
regional
distribution
of
air
temperature
T
a
(x,
y)
NO.
4
MA
YAOMING,
WANG
JIEMIN,
HUANG
RONGHUI
ET
AL.
533
in
the
Dunhuang
area
can
be
simply
derived
from
Ta-oasis
(X, y)
and
Ta_Gobi-desert
(XI
y)
due
to
only
two
kinds
of
surfaces
(oasis
and
Gobi
-desert)
existing
in
Dunhuang
area.
The
effective
aerodynamic
roughness
length
Z
orn
(x,
y)
in
Eq.(9)
over
the
HEIFE
area,
in-
cluding
the
effect
of
topography,
low
vegetation
(e.g.
grass),
and
taller
plants
(e.g.,
wheat
canopy,
trees
and
shrubs),
can
be
determined
by
Taylor's
model
(Tay-
lor
et
al.,
1989)
since
the
surface
conditions
in
the
HEIFE
area
are
the
same
as
in
Taylor's
model.
As
for
the
effective
aerodynamic
roughness
length
Z
orn
(x,
y)
in
the
Dunhuang
area,
it
can
be
simply
derived
from
Z0m-oasis
(x,
y)
and
Z
orn
_
Gobi
-desert
(x, y)
due
to
only
two
kinds
of
surfaces
(oasis
and
Gobi
-desert)
in
the
Dunhuang
area.
Raupach's
method
(Raupach,
1994)
Net
radiation
flux
density
(9
July
1991.
HEIFE)
`
irn
<-290.0
Liza
1
21
-
ia
ngye
400.0
500.0
600.0
700.0
>=740.0
is
used
to
derive
the
zero
-plane
displacement
d
o
(x,
y)
in
Eq.(9)
over
the
HEIFE
and
Dunhuang
areas,
i.e.,
d
o
(x,
y)
1
-
exp(-
1
\Adi
/LA
(x,
y))
(
11
)
h(x,
y)
cd)ILA(x,
y)
where
/
LA
is
the
leaf
area
index
(LAI),
h(x,y)
is
the
height
of
vegetation
and
c
dl
is
a
free
parameter
(Raupach,
1994).
In
other
words,
the
zero
-plane
dis-
placement
d
o
(x,
y)
can
be
derived
when
LAI
and
the
vegetation
height
are
determined
over
the
two
areas.
k13
-1
(x,
y)
in
Eq.(9)
is
determined
by
using
the
rela-
tionship
between
k13'
(x,
y)
and
T
s
(x,y),
y)
and
//),,(x,
y)
in
Eq.(9)
are
the
integrated
stability
func-
tions.
They
can
be
determined
by
using
the
models
of
Paulson
(1970)
and
Webb
(1970).
Sell
heat
flux
density
(9
July
1991.
HEIFE)
,x
10.0
20.0
40.0
60.0
80.0
Linz
2hangye
4
1
100.0
>=165.0
W/m2
Sensible
heat
flux
density
(9
July
1991.
HEIFE)Wirn
2
1
0.0
Latent
heat
flux
density{
9
July
1991.
HEIFE)
W/m
2
-0.0
100.0
50.0
A
200.0
100.0
A
.
300.0
150.0
400.0
200.0
h.
.
5
*J.
500.0
2harta
250.0
ZtVrigre
600.0
>
,
-300.0
›-
700.0
15km
Fig.
2.
Maps
of
land
surface
heat
fl
uxes
for
the
HEIFE
area.
1000
LST,
July
9,
1991.
534
ADVANCES
IN
ATMOSPHERIC
SCIENCES
VOL.
20
Net
radiation
flux
t
J
June
zuuu
Net
radiation
fl
ux
(
22
August
2000)
Net
radiation
fl
ux
(
29
January
2001)
203
180
4s
-
1
264
2E0
56
4,-
358
411
344
432
801
Gobi
Ar
T
1
,
11
'
.""
'
1
1
h
A
SAtild
4e4
544
Gobi
AWS
PAM
NI
'Ed
216
560
Gobi+
El
504
272
L
L
652
7
638
1
-
324
728
-
1748
L
3E0
W.
nv`
2
W.
M"
2
VV.
Soil
heat
fl
ux
(
3
Jane
2300)
Cloud
Gobi
AWS
PAM
16.0
IN
31.0
47.0
62.0
78.0
93.0
109
AM
Sensible
heat
fl
ux
(
3
June
200))
o
70
133
208
ut
278
IN
346
I
-
414
-1
484
t
M
2
Cloud
4-‘1
Latent
neat
nUx
.1
June
mu
Cloud
AWS
PAM
GoI
;tsa-•.:-
Soil
heat
flux
(
22
august
2001T)
Soil
heat
fl
ux
4
29
January
2001)
70
0
4
1W''
:
Mc
PAM
.*
It'
8.0
27.0
•145.0
64.0
M83.0
1020
-1
120.0
-/1a1.0
W.m"
2
Sensible
he
flux(
22
August
2003)
aR
PA
Gobi
AVVS
0
IN
56
114
170
226
202
C
340
L
1
3%
W.
111'
2
Go
b
1
INS
PAM
-MO
-34.0
-31:1
17.0
43.0
HE
094.0
L
120,0
W.
m-
2
Sensible
heat
fl
ux(
29
January
2001)
Gob'
(-013
t
ICOI
fi
4=
0.0
-
:-.11
1
.4,
1
.
105.0
105.0
MI
205
0
210.0
310,0
410.0
Gobi
AWS:
.:
.
+
+,
fi
315.0
-
415.0
NI
AWS
PAM
Go
hi+
+
+
MI
5150
IN
533.0
n
0
815.0
ri
525.0
O
CI
720.1
17
730.0
VIAM
W.
Mal
10km
Fig.
3.
Maps
of
land
surface
heat
fl
uxes
for
the
Dunhuang
area.
1000
LST.
Ili
0
17
or
35
52
70
87
L
I
-
105
L
122
VU
m
1)
<=0.0
38.0
71.0
107.0
142.0
178.0
=
213.0
-
242.0
W.
tY1-2
NO.
4
MA
YAOMING,
WANG
JIEMIN,
HUANG
RONGHUI
ET
AL.
535
Table
1.
The
distribution
range
and
peaks
of
land
surface
variables,
vegetation
variables,
and
land
surface
heat
fl
uxes
over
the
Dunhuang
area
Range
Oasis
(peak)
Gobi
desert
(peak)
NDVI
MSAVI
Pv
To
0.00-0.46
0.00-0.40
0.00-0.95
0.10-0.30
0.30
0.26
-
0.28
r-
0.17
-
0.00
-
0.00
-
0.00
r‘
,
0.22
3
Jim
2000
LAI
0.00-8.80
rsJ
1.80
rsJ
0.00
T
s
f
e
(
°
C)
2.0-63.0
rs
,
26.0
rsJ
48.0
R.
(W
m
-
)
250-550
rs
,
520
rs
,
330
Go
(W
m
-2
)
1-105
-
36
-
74
H
(W
m
-2
)
2-405
50
rs
,
245
AE
(W
m
-2
)
0-520
-
450
50
NDVI
0.00-0.74
0.64
0.06
MSAVI
0.00-0.65
rsJ
0.50 0.00
Pv
0.02-0.98
0.85
0.02
To
0.08-0.36
rs
,
0.14
0.23
22
Aug
2000
LAI
0.00-8.80
2.10
0.00
T
s
f
e
C)
5.0-57.0
rs
,
15.0
rsJ
49.0
R.
(W
m
-2
)
260-660
600
rsJ
330
Go
(W
m
-2
)
15-120
40
rsJ
103
H
(W
m
-2
)
0-300
30
230
AE
(W
m
-2
)
-20-650
550
30
Table
2.
The
distribution
range
and
peaks
of
land
surface
variables,
vegetation
variables,
and
land
surface
heat
fl
uxes
over
the
HEIFE
area
(9
July
1991)
Range
Oasis
(peak)
Gobi
desert
(peak)
NDVI
MSAVI
Pv
0.10-0.75
0.08-0.92
0.00-0.95
0.66
0.80
-
0.78
0.15
0.25
0.20
To
0.04-0.30
rs
,
0.12
0.26
LAI
0.00-5.80
rsJ
2.80
0.45
T
s
f
e
(
°
C)
5.0-55.0
rsJ
16.0
X44.0
R.
(W
m
2
)
290-750
650
rsJ
380
Go
(W
m
-2
)
30-105
rs
,
50
-
90
H
(W
m
-2
)
0-300
ti
90
ti
230
AE
(W
m
-2
)
0-700
500
100
3.4
Latent
heat
flux
The
regional
latent
heat
fl
ux
AE(x,
y)
is
derived
as
the
residual
of
the
energy
budget
theorem
for
the
land
surface,
i.e.,
AE(x,y)
=
R
n
(x,y)
-
H(x,y)
-
G
o
(x,
y)
.
(12)
4.
Case
study
and
validation
Figures
2
and
3
show
the
distribution
maps
of
land
surface
heat
fl
uxes
over
the
HEIFE
area
and
Duhuang
areas.
The
frequency
distributions
of
land
surface
fl
uxes
over
the
two
areas
are
shown
in
Fig.
4.
The
derived
land
surface
heat
fl
uxes
are
validated
by
fi
eld
measurements.
In
Fig.
5,
the
derived
results
are
plot-
ted
against
the
measured
values
in
the
fi
elds
of
Dun-
huang
and
the
HEIFE
for
four
terms
of
the
energy
balance.
The
1:1
line
is
also
plotted
in
the
graphs.
The
land
surface
heat
fl
uxes,
which
were
derived
from
SEBAL
(Wang
et
al.,
1995;
Ma
et
al.,
1999),
are
plot-
ted
in
Fig.
5
as
well.
Since
it
is
difficult
to
determine
where
the
exact
locations
of
the
experimental
sites
are,
the
values
of
a
5
x
5
pixel
rectangle,
surrounding
the
determined
Universal
Transverse
Mercator
(UTM)
co-
ordinate,
are
compared
with
the
fi
eld
measurements.
536
10
a
ADVANCES
35
30
"
25
IN
ATMOSPHERIC
20
5
_
,„
16
SCIENCES
10
8
ma
VOL.
20
obi-desert
G
obi-desert
Gobi-desert
n
Gobi-desert
cc
6
0
ass
20
12
"
6
15
8
4
(I
i
10
0
ass
0
ass
2
91
5
4
2
r,
0
20
40
60
80
100 120
0
50
100
150
200
250
300
0
100
200
300
400
5,,
600
700
800
200
300
400
500
600
700
800
SoRheatifl.
G
o
(W
m
2
)
Sensth heatflux
H
(W
m
2
)
Lantheatillx
I
e
(W
m
N
etxadiaton
R
(
W
m
)
6
6
(a)
10
3
June
2000
22
August2000
.
29
Januar/200f
4
G
obi-desert
C
bud
4
obi-desert
0
o
0
ass
a.)
2
0
E,
-1
2
44
1
N
2
EA
-rrifie
0
—A
-4L5555.5
0
300
400
500
600
700
800
400
500
600
700
800
200
300
400
200 200
300
100
Net/adiadm
flux
(W
on
2
)
Netradiatbn
flax
(W
on
2
)
N
etradiatbn
flax
(W
on
2
)
10
3
June
2000
10
30
22
August2000
-
Gobi-desert
25
-
29
Januaty
2001
-
6
20
-
ud
G
obi-desert
6
15
4
4
6
10
-
PI
2
0
ass
2
0
0
0
(
:)
.:11111111111111111110101
r4
5
0
20
40
60
80
100 120
Soaheatflax
on
2
)
0
20
40
60
80
100 120
140
-60 -40 -20
0
20
40
60
80
Soitheat
flux
(W
m
)
ScalheatfLx
(W
m
2
)
100
16
^
12
ow
1
-.
16
3
June
2000
22
August2000
Gobi-desert
_
29
January
2001
12
4
C
bud
Gobrdesert
0
asis
0
ass
0
efl
-
TarrrEITTA
11
1111
-
snrt.E.._
91
2
0
L,
4
60
0
100
200
300
400
500
50
100 150
200 250
300
80
100 120 140
Sensioe
heatfl,
(W
m
2
)
Sensiph
heatflix
W
m
2
)
Sensbh
heatflix
(W
m
2
)
16
16
10
3
June
2000
'
14
22
August2000
29
January
2001
12
ow
12
obi-desert
10
'‘
5
,
6
Gobi-desert
6
4
C
bud
_
4
0
ass
4.,
2
2
0
asis
Th
I
11771•1
Ill
I
Ll
II
n11
-
n-,
-200
0
200
400
600
800
0
100
200
300
400
500
600
700
Lantheatfux
(W
m
2
)
50
100
150
200
Lantheatflax
(W
m
2
)
Latentheatflax
(W
m
2
)
(b)
Fig.
4
Frequency
distribution
of
land
surface
heat
fl
uxes.
(a)
HEIFE;
(b)
Dunhuang.
NO.
4
MA
YAOMING,
WANG
JIEMIN,
HUANG
RONGHUI
ET
AL.
537
750
700
Cal2
nze
650
A
Call
Zhangye
5
600
550
500
I?,
a
e
450
G
obi
2
C.9
400
350
p
eseg.
. . . . . .
.
350
400 450
500 550
600 650
(TN
m
2
)
n-ra
easured
300
250
200
150
xD
100
50
50
100
150
200 250
400
350
5
300
250
200
150
700
750
Cal2
A
Call
Hmeaaurea
(w
2
)
300
. .
I
3
June 200*,
22
August2000)K
1a
5k
2r January
2001
e
150 200 250 300 350 400
(a)
110
100
90
80
70
60
50
40
30
30
40
50
60
70
80
90
Cal2
A
Call
-
Lhze
ngye
Gobi
1..,.
„sezt.
600
500
400
5
300
200
-61
1--1-'
4
100
E
250
(14
R
n-measumd
m
-2
)
3
June
2000
22
August2000
200
E
5
1l
150
x4:1
29
January 2002,....
)1(
'
100
100 150 200
H
measumd
(N
in
2
)
250
110
100
90
80
70
60
50
G
o,easu,,d
(W
m
2
)
100
110
. I . . . . . . .
Cal2
A
Call
_
Gob
0
0
Lhzei
Zhangye
Desert
. .
.
. . .
.
100
200
300
400
500
TT
;E„eau,ed
(W
m
2
)
600
22
August2000X
-
/
-
3
June 2000)1‹
.•••
January 2001
I
50
60
70
80
90
G
0-m
easumd
(147
m
2
)
100
80
60
40
20
0-
0
100 110
3
June
2000
N
29 January 2001
2
August2000
20
40
60 80
2
m
easumd
rn
)
100
(b)
Fig.
5
Validation
of
the
derived
results
against
the
fi
eld
measurements
for
land
surface
heat
fl
uxes,
together
with
the
1:1
line.
Ca1.1:
former
results
(Wang
et
al.,
1995;
Ma
et
al.,
1999);
Ca1.2:
this
research.
(a)
HEIFE;
(b)
Duhuang.
538
ADVANCES
IN
ATMOSPHERIC
SCIENCES
VOL.
20
The
mean
absolute
percent
difference
(MAPD,
DMAP)
can
quantitatively
measure
the
difference
between
the
derived
results
(H
derived(i)
)
and
measured
values
(Hmeasured(i)):
DMAP
100
n
i=1
(13)
(
Hderived(i)
Hmeasured(i)
Hmeasured(i)
The
results
show
that:
(1)
the
derived
land
surface
variables
(land
surface
reflectance
and
surface
tem-
perature),
vegetation
variables
(NDVI,
MSAVI,
veg-
etation
coverage
P
v
and
LAI,
and
land
surface
heat
fl
uxes
(net
radiation
fl
ux
R„,
soil
heat
fl
ux
G
o
,
sen-
sible
heat
fl
ux
H,
and
latent
heat
fl
ux
)E)
over
two
case
study
areas
are
in
good
accordance
with
the
land
surface
status.
These
parameters
in
summer
show
a
wide
range
due
to
the
strong
contrast
of
surface
fea-
tures
during
this
season,
and
there
are
two
peaks
in
the
fi
gures
of
all
distribution
maps
and
all
frequency
distri-
butions
histograms.
The
fi
rst
ones
correspond
to
oasis
and
the
other
peak
corresponds
to
the
Gobi
desert
(see
Table
1
and
Table
2).
Although
there
are
differ-
ences
between
oasis
and
Gobi
desert
for
land
surface
variables,
vegetation
variables,
and
land
surface
heat
fl
uxes
over
the
two
areas
in
winter,
they
are
not
clear;
(2)
the
derived
surface
reflectance
and
surface
temper-
ature
in
this
research
are
in
good
accordance
with
the
fi
eld
measurements,
and
they
are
better
than
the
re-
sults
derived
from
the
regression
relationship
(Wang
et
al.,
1995;
Ma
et
al.,
1999)
with
MAPD
(mean
absolute
percent
difference)
less
than
10%;
(3)
the
derived
net
radiation
fl
uxes
over
the
two
areas
are
very
close
to
the
fi
eld
measurements
with
MAPD
less
than
7%;
(4)
the
parameterization
method
based
on
MSAVI
for
soil
heat
fl
ux
is
suitable
for
the
inhomogeneous
land
sur-
face
of
arid
and
semi
-arid
areas.
Although
the
derived
regional
soil
heat
fl
ux
is
slightly
higher
than
the
mea-
sured
value,
the
MAPD
gets
smaller
than
the
former
derived
value
based
on
the
NDVI
(Wang
et
al.,
1995;
Ma
et
al.,
1999);
(5)
the
derived
regional
sensible
heat
fl
uxes
with
MAPD
of
around
6%
at
fi
ve
validation
sites
in
Dunhuang
and
HEIFE
are
in
good
agreement
with
the
fi
eld
measurements;
(6)
the
derived
regional
latent
heat
fl
ux,
which
is
based
on
the
energy
balance
equa-
tion,
is
acceptable
for
the
whole
HEIFE
area
and
Dun-
huang
area;
and
(7)
net
radiation
fl
ux
and
latent
heat
fl
ux
in
summer
(June
and
August)
are
much
higher
than
in
winter
over
the
oasis
area
of
Dunhuang,
but
the
sensible
heat
fl
ux
is
much
lower
than
in
winter.
Net
radiation
fl
ux,
sensible
heat
fl
ux,
and
soil
heat
fl
ux
are
much
higher
in
summer
than
in
winter
over
Gobi
desert
area
of
Dunhuang,
and
latent
heat
fl
ux
over
the
the
Gobi
desert
is
very
low
in
summer
(-40
W
m
-2
).
This
is
caused
by
the
seasonal
difference
and
land
sur-
face
status
difference
between
summer
and
winter.
5.
Concluding
remarks
In
this
study,
the
regional
distributions
of
land
sur-
face
variables
(surface
reflectance
and
surface
temper-
ature),
vegetation
variables
(NDVI,
MSAVI,
vegeta-
tion
coverage,
and
LAI)
and
land
surface
heat
fl
uxes
(net
radiation,
soil
heat
fl
ux,
and
sensible
and
latent
heat
fl
ux)
over
the
inhomogeneous
areas
of
Dunhuang
and
HEIFE
are
derived
with
the
aid
of
Landsat
TM
data
and
fi
eld
observations.
Compared
with
previous
studies
(Wang
et
al.,
1995;
Ma
et
al.,
1999),
the
new
method
has
been
proved
to
be
a
better
approach
for
getting
related
air
-land
parameters
over
heterogeneous
landscape
due
to
the
improvements
in
old
parameter-
izations
(Wang
et
al.,
1995;
Ma
et
al.,
1999).
This
study
forms
a
sound
basis
to
study
land
surface
vari-
ables,
vegetation
variables,
and
land
surface
fl
uxes
over
inhomogeneous
landscapes.
The
vegetation
variables
cannot
be
validated
in
this
research
due
to
the
lack
of
such
measurements
during
the
HEIFE
and
Dunhuang
experiments.
In
fu-
ture
experiments,
more
attention
should
be
paid
to
the
measurements
of
vegetation
variables,
such
as
NDVI,
LAI,
and
vegetation
coverage.
Acknowledgments.
This
work
was
under
the
aus-
pices
of
the
Innovation
Project
of
the
Chinese
Academy
of
Sciences
(KZCX3-SW-329),
the
Chinese
National
Key
Programme
for
Developing
Basic
Sciences
(G1998040900),
and
the
National
Natural
Science
Foundation
of
China
(40275003).
Some
parts
of
this
study
were
done
as
coop-
erative
research
works
in
the
Disaster
Prevention
Research
Institute,
Kyoto
University,
and
the
Alterra
Green
World
Research,
Wageningen
UR,
the
Netherlands.
The
authors
wish
to
acknowledge
Profs.
H.
Ishikawa,
0.
Tsukamoto,
M.
Maitani,
and
E.
Ohtaki
for
their
kind
help
and
useful
discussions.
REFERENCES
Carlson,
T.
N.,
and
D.
A.
Ripley,
1997:
On
the
relation
between
NDVI,
fractional
vegetation
cover,
and
leaf
area
index.
Remote
Sens.
Environ.,
62,
241-252.
Hu
Yinqiao,
Gao
Youxi,
Wang
Jiemin,
Ji
Guoliang,
Shen
Zhibao,
Cheng
Linsheng,
Chen
Jiayi,
and
Li
Shouqian,
1994:
Some
achievements
in
scientific
re-
search
during
HEIFE.
Plateau
Meteorology,
13(3),
225-236.
(in
Chinese)
Hu
Zeyong,
Huang
Ronghui,
Wei
Guoan,
and
Gao
Huang-
cun,
2002:
Variations
of
surface
atmospheric
variables
and
energy
budget
during
a
sandstorm
passing
Dun-
huang
on
June
6
of
2000.
Chinese
Journal
of
Atmo-
spheric
Sciences,
26(1),
1-8.
(in
Chinese)
NO.
4
MA
YAOMING,
WANG
JIEMIN,
HUANG
RONGHUI
ET
AL.
539
Ma
Yaoming,
Wang
Jiemin,
M.
Menenti,
and
W.
G.
M.
Bastiaanssen,
1999:
Estimation
of
fl
uxes
over
the
heterogeneous
land
surface
with
the
aid
of
satellite
remote
sensing
and
fi
eld
observation.
Acta
Meteor.
Sinica,
57(2),
180-189.
(in
Chinese)
Ma,
Y.,
2001:
Parameterization
of
land
surface
heat
fl
ux
densities
over
inhomogeneous
landscape
by
combining
satellite
remote
sensing
with
fi
eld
observations.
Ph.
D
dissertation,
Okayama
University,
Japan,
195pp.
Ma,
Y.,
0.
Tsukamoto,
H.
Ishikawa,
Z.
Su,
M.
Menenti,
J.
Wang,
and
J.
Wen,
2002:
Determination
of
re-
gional
land
surface
heat
fl
ux
densities
over
heteroge-
neous
landscape
of
HEIFE
Integrating
satellite
remote
sensing
with
fi
eld
observations.
J.
Meteor.
Soc.
Japan,
80(3)
485-501.
Maitani,
T.,
K.
Sahashi,
E.
Ohtaki,
0.
Tsukamoto,
and
Y.
Mitsuta,
1995:
Measurements
of
turbulent
fl
uxes
and
model
simulation
of
micrometeorology
in
a
wheat
fi
eld
at
Zhangye
oasis.
J.
Meteor.
Soc.
Japan,
73(5):
959-965.
Mason,
P.,
1988:
The
formation
of
areally
averaged
rough-
ness
lengths.
Quart.
J.
Roy.
Meteor.
Soc.,
114,
399-
420.
Mitsuta,
Y.,
I.
Tamagawa,
K.
Sahashi,
and
J.
Wang,
1995:
Estimation
of
annual
evaporation
from
the
Linze
desert
during
HEIFE.
J.
Meteor.
Soc.
Japan,
73(5),
967-974.
Paulson,
C.
A.,
1970:
The
mathematical
representation
of
wind
speed
and
temperature
profiles
in
the
unstable
atmospheric
surface
layer.
J.
Appl.
Meteor.,
9,
857-
861.
Qi,
J.,
A.
Chehbouni,
A.
R.
Huete,
Y.
H.
Kerr,
and
S.
Sorooshian,
1994:
A
modified
soil
adjusted
vegetation
index.
Remote
Sens.
Environ.,
48,
119-126.
Raupach,
M.
R.,
1994:
Simplified
expressions
for
vegeta
tion
roughness
length
and
zero
-plane
displacements
as
functions
of
canopy
height
and
area
index.
Bound.
-
Layer
Meteor.,
71,
211-216.
Taylor,
P.
A.,
R.
I.
Sykes,
and
P.
J.
Mason,
1989:
On
the
parameterization
of
drag
over
small
scale
topography
in
neutrally
stratified
boundary
fl
ow.
Bound.
-Layer
Meteor.,
48,
409-422.
Tsukamoto,
0.,
J.
Wang,
and
Y.
Mitsuta,
1992:
A
sig-
nificant
evening
peak
of
vapour
pressure
at
an
oasis
in
the
semi
-arid
region.
J.
Meteor.
Soc.
Japan,
70(6),
1155-1159.
Tsukamoto,
0.,
K.
Sahashi,
and
J.
Wang,
1995:
Heat
bud-
get
and
evapotranspiration
at
an
oasis
surface
sur-
rounded
by
desert.
J.
Meteor.
Soc.
Japan,
73(5),
925-
935.
Valor,
E.,
and
V.
Caselles,
1997:
Mapping
land
sur-
face
emissivity
from
NDVI:
Application
to
European,
African,
and
South
American
areas.
Remote
Sens.
En-
viron.,
57,
167-184.
Verhoef,
W.,
1997:
Theory
of
radiative
transfer
models
ap-
plied
in
optical
remote
sensing
of
vegetation
canopies.
Ph.
D.
dissertation,
Remote
Sensing
Department
of
National
Aerospace
Laboratory,
The
Netherlands.
Wang,
J.,
Y.
Ma,
M.
Menenti,
W.
G.
M.
Bastiaanssen,
and
Y.
Mitsuta,
1995:
The
scaling
-up
of
processes
in
the
heterogeneous
landscape
of
HEIFE
with
the
aid
of
satellite
remote
sensing.
J.
Meteor.
Soc.
Japan.,
73(6),
1235-1244.
Webb,
E.
K.,
1970:
Profile
relationships:
The
log
-linear
range
and
extension
to
strong
stability.
Quart.
J.
Roy.
Meteor.
Soc.,
96,
67-90.
Zhang
Qiang,
Wei
Guoan,
and
Huang
Ronghui,
2001:
Bulk
transfer
coefficients
of
momentum
and
sensible
heat
over
the
Gobi
desert
surface
in
north-western
arid
area.
Science
in
China
(Series
D),
31(9),
783-792.
(in
Chinese)
T.
1
4
1
.1x#T.
1
4
1
.ithEt1
a'
ika
.
11
4
).T.Eas.
*.
Rit
RIM
Massimo
Menenti
j'h
-
ottR
414A
'A
ic
AitgliairrAT9VA
t
-
E5
-
)
-
IIERA-1
Hth:r6JE.
11
FM4MT
t*TTESSfrittffialgn_Mlidn
-
A,
INAJLitE<
"A
INIVAMFIX'A
Xbitfra*1
r9-i,"(111.*IkAMMagNAIJA
G1998040900,
1999-2003)11VRti,40"
Eq11"ffl'irT1,440"
(HEIFE,
1989-
1994)E<
3
T0114
1
N1("g(
i,414A":
200046)13H
-Ink.
200048/122H-k*fr120:
129E1-1-X;
"
'irTiAgeE<:
1991
7)19H
-I)
1713-MVANTELandsat-5
TM
filLandsat-7
T
dr1V-
-, -7
t
-
EdAtt,VVf_Mik.
3%1
13.1i4ik;T__MtZ
4-n
-
Anit)1116111#1ffl&A
.
-T-
Ezt-f-
i•-th,E<,
Landsat
TM,
itffiggsq