STRATEGIC ALTERNATIVES TO CALCULATE CROP AREA 
OF AN ADMINISTRATIVE DIVISION 
Hongliang Fang 
State Key Lab. of Resources and Environment Information System, 
Institute of Geography, Chinese Academy of Sciences, 100101, Beijing, China 
KEY WORDS: Agriculture, Estimation, Identification, Inventory 
ABSTRACT: 
In crop yield estimation using remotely sensed data, it usually needs to calculate the crop planting area in a particular 
administrative division. Most previous investigators do as follows: first, they cut down the target image of the study 
area with the administrative boundary, then conduct land cover/use classification and crop identification work, and last 
calculate the crop area. Other fewer researchers conduct the land cover/use classification work first and then cut down 
the study area with administrative boundary, and last calculate the crop area. We call these two methods strategy Acut 
and classify) and strategy B (classify and cut) respectively. In this paper, we applied these two strategies to rice planting 
area identification. Our results indicate that strategy B is obviously excellent than strategy A in the unsupervised 
classification-cluster process and the rice area accuracy it extracted is over 8496. 
1. INTRODUCTION 
Crop production forecasting using remote sensing 
technique involves the estimation of the cultivated 
area and yield per unit area. Estimating the cultivated 
area is the primary step for crop yield 
forecasting(Tennakoon et al., 1992). In estimating the 
cultivated area, it usually needs to calculate the crop 
planting area of a particular administrative division for 
the purpose of being in agreement with statistical 
channels. 
To calculate the crop planting area of a particular 
administrative division using remotely sensed data is 
computer and labor consuming. By using the map to 
determine the location of the administrative boundaries, 
we may be able to remove the inappropriate areas before 
beginning the bulk of image-processing computations, 
and thus decrease the costs of the analysis(Star and Estes, 
1990). In fact, not every researcher does as above. Some 
perform image analysis first, then mask out the 
uninterested area with the key boundaries, and thus get 
the particular target information(Hall-konyves, 1990; 
Makey et al., 1993). 
Thus, we have two strategies here: first, cut down the 
target image of the study area with the administrative 
boundary, and then conduct land cover/use classification 
and crop identification work, and last calculate the crop 
area. Second, conduct the land cover/use classification 
work firstly and then cut down the study area with 
administrative boundary, and last calculate the crop area 
of the study area. We call these two methods strategy A 
(cut and classify) and strategy B (classify and cut) 
respectively. Taking rice area identification as an 
240 
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B4. Vienna 1996