ON THE FEASIBILITY OF DYNAMIC MONITORING OF TEMPERATURE
OF VEGETATIVE CANOPY BY NOAA AVHRR DATA
Xu Xiru, Niu Zheng
Institute of Remote Sensing of Peking University
Beijing, China 100871
SUMMARY :
It is very significant to measure the temperature of vegetative canopy, because it is a key parameter to
know the interaction between atmosphere,soil and vegetative canopy which is very useful for drought moni-
toring, crop yield estimation and so on:
There are three basic difficulties to fulfill this task by infrared band data,the absorption of atmospher-
ic water vapour,the different thermo-emissivity of different ground targets and the mixed pixel problem.
Now, we offer a new method to realize this dynamic monitoring of canopy temperature by NOAA-AVHRR data.The
error of measurement of canopy temperature may be within 1°k.The basic idea of this method is to suppose
the vicinal pixels owning common physical conditions, such as vegetative canopy temperature,soil surface
temperature, thermo-emissivity of soil,profiles of atmospheric parameters and so on,to eliminate the comp-
lex influence of soil background using the difference of vegetative coverage between two vicinal pixels
which creats a favourable condition to cancel atmospheric effect by split window.
This paper also shows
that the 1$ accuracy of calculation of vegetative coverage is needed if we hope the
error of measurement of vegetative canopy temperature within l?k. An initial result is given,it says it is
feasible to measure canopy temperature by AVHRR data.
KEY WORDS:
l. INTRODUCTION
The measurement of land surface temperature by in-
frared band data is more complicated than the mea-
surement of sea surface temperature because of thr-
ee main obstacles. They are mixed pixel problem,the
complexity of land surface, namely the different
ground target owning different emissivity and tem-
perature,and the obvious difference between actual
atmospheric conditions and standard atmospheric
profile. It is very difficult to get parameter's
profile of atmosphere in situ, so it is both comp-
licated and almost imposible to correct atmospheric
effect by theoretical calculation. That is the rea-
son why after many years panstaking effort the pro-
blem of measurement of land surface temperatrue
Still wanders on the stage of feasibility study,but
it is a very significant parameter which describs
the interaction between earth surface and atmosph-
ere, for example, if we can say the monitoring of
drought which drives world attention today can be
quantitatively monitoring by evapotranspiration mo-
del, then it is a key problem to measure tempera-
ture of vegetative canopy with high accuracy by re-
mote sensing method. From the creation of this mo-
del almost two decades had passed, but till now it
does not be used in operation manner, because the
key problem have not been solved.
Dr. Wan had analized the possibility to measure the
temperature of snow surface, bare soil surface, ve-
getative canopy etc. by NOAA-AVHRR data. He also
pointed out the error possibly within l'k, but the-
re are two problems,we have to point out here, the
mixed pixel problem to be neglected by him and the
inverse calculation only based on U.S. standard
profile of atmosphere. Obviously, these two assum-
ptions are far from reality.They had seriously da-
maged its value of application.
The aim of this paper is to discuss the feasibility
of measurement of canopy temperatrue without these
two unreasonable preconditions. It is very diffi-
cult to generally discuss the measurement of land
surface temperature, but if our attention to be fo-
cused on canopy temperature, it will be feasible.
2. METHOD
2.1 Model
376
Feasibility, Monitoring, Vegetation, Canopy,Temperature, NOAA-AVHRR.
The thermo-emissivity of single leaf reaches 0.98.
Since the existence of 'cave effect', it seems
that it is reasonable to suppose the thermo-
emissivity of vegetative canopy equal to 'l'. If
we focus our attention to rural area, mixed pixel
are consisted of soil background and vegetative
canopy, therefore the radiance received by sensor
aboard satellite is
i
LeI[f ga Lp (Tg) (i-a E. 1 t+] L, (T(Z))dt
Ts
+(1-£) (1-a ) e Lou) Ge eese eee (1)
e UU
where L is the radiance arrived at sensor aboard
satellite, Lg represents Planck black body radia-
tive formular, T. and T represent physical tem-
perature of vegetative canopy and soil respective-
ly, a, means the vegetative coverage in single pi-
xel, O& a < T, £ is the thermo-emissivity of soil
: t,is the transmittance of whole atmosphere, t is
the atmospheric transmittance from Z to up bounda-
ry of atmosphere, T(Z) and e(Z) means profile of
atmospheric temperature and water vapour mass res-
pectively.
We state the physical meaning of every term appe-
ared on the right side of equation (1) as follows:
The first term represents the radiance emitted by
canopy and soil themself attenuated by atmosphere
at last reaching sensor aboard satellite. The se-
cond term means upward radiance of atmosphere. The
down ward radiance of atmosphere reflected by soil
back ground then experienced atmospheric attenu-
ation at last reaching sensor is calculated by
third term.
If we are interested in drought monitoring of lar-
ge area,namely the change with low spatial fre-
quency, therefore we can assume two vicinal pixel
owning the same values of T. Tor Es T(Z), e(Z)
and so on.
"l" and "2" are two vicinal pixel. |
v ü-al)- 20s ) |
V v
AL = = Ly C t
+ Ly (T(2))dt ee (2)
The
the
Let
Tha
the
Let
We
atu
The
the
ar.
acc
Let
The
inl;
pt:1
ent
ther
The
(3)
and
by 1
Whe1
cule
and
The
assu
cal
soil
COVE
the
absc
by!
to k
Fig.
chan
Wher