7. Istanbul 2004
ment practices,
of individual
owth period (i.e,
e complex crop
, pesticide, and
r irrigation and
gram: manually
applied in the
inual mode are
1). The land use
the Australian
alysis Network,
/heat - Summer
1 China
WHEAT
ation Volume
ation 40mm
lizer 72kg/ha'
lizer SSkg/ha"
ting
ation 100mm
lizer 48kg/ha"
ation 100mm
vest
iemical fertilizer.
1 chemical fertilizer.
sed on three-
developed for
1 (Myneni et al.
vegetation is
asses and cereal
leaf forests and
s among these
iogeneous VS.
multi- story)
brightness and
used to define
fixed parameter
. LUTS are then
del for various
gorithm ingests
ctance factors,
/ geometries. It
parable values
JTs and derives
his method fails
ed on relations
index (NDVI)
s retrieved from
and on an 8-day
xtensive quality
data processing
est-quality LAI
omposited over
e monthly data
[S LAI product
loaded from the
| research group
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004
in the Department of Geography at Boston University (available
fip://ersa.bu.edu/pub/rmynen i/myneniproducts/datasets/MODIS
/MODIS BU/C4/).
5. RESULTS AND DISCUSSION
5.1 Validation of the Spatial EPIC Model
The average yield of winter wheat and summer corn in North
China for 1980s, simulated by the spatial EPIC model, can be
seen from the Figure 2 and 3. The simulated yields are just
compared with the statistical yields from the China Statistical
Yearbook from 1982-1991, due to the lack of the actual yield
data. The Table 2 shows the comparison results. The differences
in percentage between simulated and statistical yield are mostly
under 1096, except the situation in Beijing and Shandong. It is
evident that crop yield of the area is underestimated by the
spatial EPIC model, especially for Beijing. The reason is that
Beijing and Shandong is the developed region in North China.
The cropland in these regions are applied a very good field
management with a better irrigation condition, fertilizer
condition and so on. But only the simple and ordinary field
operation parameters are inputted into the spatial EPIC model,
which result in the underestimating situation. If the EPIC crop
parameters established by USDA can be adjusted to be suitable
for the application in North China, and the detailed field
management information, such as the cropping system,
irrigation schedule, fertilizer schedule and tillage schedule, can
be obtained and be inputted into the spatial EPIC model. It
should be possible to improve the simulation accuracy.
Table 2. Comparison between the simulation yield from spatial
EPIC model and the statistical yield (Ton/hectare)
SUMMER CORN WINTER WHEAT
Region
Simulated Statistical Error Simulated Statistical Error
BeiJing 2.979 4.820 38.2% 2.598 3.959 34.4%
TianJin 3.686 3.864 4.6% 2.812 2.829 0.6%
HeBei 3.647 3.623 0.7% 2.654 2.965 10.5%
ShanDong 3.639 4.356 16.5% 2.669 3.388 21.2%
HeNan 3.706 3323 11.596 3.002 3.322 9.6%
ShanXi 3.796 3.993 4.9% 2.528 2.576 1.9%
Note: “Simulated” means yield simulated by model; "Statistical" means the
average statistical yield from the China Statistical Yearbook from 1982-1991.
(À
_
3 '
“Beiding V
d TlanJin "A RI
5 ENT
HeBei gs,
[n reta
: ; y te
S } A "d 2* Ach -
ShanXi x j A erm e
y = Bi A5
/ " ShanDong e Ae
DEA E
i : 3
7 va
; c5 A +
; if
HeNan.
EN 36-41
[77] Other Landuse
Figure 2. The simulated yield per hectare of winter wheat by
spatial EPIC model in North China
153
A
“>
Dm > > = YY
7 4 ~~ BeiJing,” -
"ai i { AN. ,
? erat "x a
; „AT landint- As
2 x £. f
À HeBei P eR
x X ric. DF EA mas 4 # ? ^
^Y % + ? 97 x
ç ShanXi / = d ;
/ / et » (7
f d f ot
{ e P ShanDong TE
í eg
| [Oe
» y
£ . ~ ^
i - mi
bh cad HeNan
* 1 10.0-1.0 T/HA
5 14-20
+ 121-320
ui (731-40
E Sn [141.60
= EE 61-70
we 1 74.7.5
80 0 80 — 160 Kilometers
Figure 3. The simulated yield per hectare of summer corn by
spatial EPIC model in North China
5.2 Validation of Combining the Spatial EPIC Model and
MODIS LAI Product
Winter wheat — summer corn rotation is the dominant cropping
system in North China. According to ground observation data,
the key crop phenological stages are emergence (October),
recovering (February), heading (May), maturity (June) of winter
wheat and emergence (June or July), tasseling (August),
maturity (October) of summer maize. The maturity of winter
wheat and the sowing of summer maize usually occur within 20
days. The leaf area index should reach maximum values during
the heading (winter wheat) and tasseling (maize) stages. The
colour-coded images of monthly MODIS LAI product for East
Asia from year 2002 (September) to year 2003 (August) are
shown in figure 4. From the consecutive images of monthly LAI
in one year, it is evident to see the change profile of LAI value.
But temporal resolution seems to be impossible to retire the
model parameters to calibrate the spatial EPIC model. The
higher resolution MODIS LAI product in 8-days or daily in
some key stage should be obtained for the integration.
Therefore, the validation of combining the spatial EPIC model
and MODIS LAI product is not conducted yet.
6. CONCLUSIONS
The operational methodology of crop yield assessment in
regional level was introduced in this study by integrating EPIC
model with NASA MODIS LAI product, ground-based
ancillary data, and GIS. The spatial EPIC model was developed
and validated in North China firstly. The result indicated that
the spatial EPIC model could simulate crop yield efficiently at
regional level. But crop management information required by
model, such as planting time, irrigation schedule and fertilizer
schedule et al., is crucial for simulation accuracy and is not
available by field measurement. Satellite remotely sensed data
cans provide a real-time assessment of the magnitude and
variation of crop condition parameters. Therefore the
methodology of combining MODIS LAI product with Spatial
EPIC model to improve yield simulation accuracy was built
secondly, but it was not conducted and validated due to be lack
of the necessary input data set yet.
