! 2004
——— —
ilt of
y.
stinct
mer,
four
a. 4),
ate a
Hc in
f the
first
vn in
re in
n, in
r the
1) is
crops
with
ouble
also
ig.5),
NDVI
crop
€ but
ice in
nonic
. The
time
1 our
crop
goes
nd is
7e Or
with
vheat
ak is
im.
ar
69
79
85
87
. four
crop
Huai-
International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, Vol XXXV, Part B7. Istanbul 2004
10 T T T T T T T
[^ i —- 1st Harmonic
FEN ---- 2nd Harmonic
/ \ € :
/ X 3rd Harmonic
Wd MN | — 1st+2nd+3rd+4th
Er Ur % q
1;
y. 7
s TES /
= Ut * 7
2 7 :
/ à ^
5r / \ N 4
=f - X
/ X
=] 0 ae | 1 1 1 1 L
5 10 15 20 25 30 35 40
Period
Figure 4. The curve constructed from the first four harmonics
for single crop rotations in Loess Plateau Region
and Yellow-Huai-Hai Rivers’ Region.
m
A —- 1st Harmonic
{ A ---- 2nd Harmonic
10 j \ 3rd Harmonic
er f£ | À | Tst+2nd+3rd+4th
"i \ / \
/ \
/ \ um | \
L / = tI zl
25 | v ;
5 / \ ! N
E j \ / L^ \
gw l' \ J U 7
Jj boni b
y d Arte
2 d MN -
E e ed v Pur
-10 1 1 1 1 L 1 i
0 5 10 15 20 25 30 35 40
Period
Figure 5. The curve constructed from first four harmonics for
double crop rotations in Loess Plateau Region and
Yellow-Huai-Hai Rivers’ Region.
4.2 Result Evaluation
The results of the crop rotations classification were evaluated
in two ways. First, we compared the Fourier based
classification with. the Chinese cultivated system
regionalization map. Second, they were compared with the
ground based agricultural monitoring data derived from local
stations in year 2000.
The crop rotations map (Fig.3) described above was evaluated
first by comparison with the Chinese cultivated system
regionalization map, which was georeferenced to the same map
coordinate system. The result of the crop rotations
classification keeps comparative consistency with the reference
map. The high accuracy derived from this comparison is partly
due to the coarse description of the Chinese cultivated system
regionalization map. But anyway the comparison shows that
there are no notable mistakes in the result s of the crop
rotations classification.
In addition to the map comparisons, we also evaluated the
results with the ground-based observations, which include the
information about crop phonological calendar, and main crop
types (winter wheat, spring wheat, spring maize, summer
maize, rice, rapeseed, cotton, soybean, sorghum and peanut)
237
planted in the local area. There are 96 local stations
observations (43 for Loess Plateau Region, 53 for Yellow-
Huai-Hai River' Region) in year 2000 in the study areas.
In Loess Plateau Region, double crop rotations were observed
in 14 of 43 stations and winter wheat/summer maize (8 stations)
and winter wheat/cotton (4 stations) were the main rotations
types. From others local stations observations, winter wheat,
spring wheat, spring maize and sorghum were known as the
main crops planted in single rotations. In Yellow-Huai-Hai
River' Region, double crop rotations were observed in 46 of 53
stations, and four of them for double cropped winter wheat/
rice, others for double cropped nor rice crops. Winter
wheat/summer maize was the main double crop rotations type
in this region since it was observed in thirty local stations. The
number of the local stations observations for single or double
crop rotations showed the general agreement with the result of
the classification.
The accuracy of the crop rotations map evaluated with
observations from 96 local stations was shown in Table. 2. The
overall agreement between our crop rotations map and ground
observations was 85 94.
Single Double User's
cropped cropped accuracy (94)
Single cropped 28 6 82.4
Double cropped 8 54 87.1
Table 2. The accuracy of the crop rotations map evaluated with
observations from 96 local stations.
5. CONCLUSIONS
In this study we detected the situation of crop rotations in
Yellow-Huai-Hai Rivers’ Region and Loess Plateau Region,
China using discrete Fourier analysis and unsupervised
classification approach with AVHRR time series NDVI. The
result of this study shows that the methodology used in this
study is, in general, feasible for detecting the distributions of
single cropped and double cropped rotations in China.
Detection of triple crop rotations using discrete Fourier
transform will be considered in the further study.
6. REFERENCES
Andres, L., Salas, W.A., Skole, D., 1994. Fourier analysis of
multitemporal AVHRR data applied to a land cover
classification. International Journal of Remote Sensing, 15(5),
pp.1115- 1121.
Azzali,S., Menenti, M. 2000. Mapping vegetation-soil-climate
complexes in southern Africa using temporal Fourier analysis
of NOAAAVHRR data. /nternational Journal of Remote
Sensing, 21(5), pp.973-996.
Chinese agricultural regionalization committee, 1991. Chinese
agricultural natural resource and agricultural regionalization.
Press of agriculture, Beijing. pp.162— 165. (In Chinese ).
Jakubauskas, M.E., Legates, D.R., Kastens, J., 2001. Harmonic
analysis of time-series AVHRR NDVI data. Photogrammetric
Engineering and Remote Sensing 67 (4), pp.461-/470.
