Search Filter
- Publications
- Abstract of Theses and Dissertations
- Database
- Extraction of Paddy Rice Field by Applying Tasselled Cap, NDVI, RVI of...
Extraction of Paddy Rice Field by Applying Tasselled Cap, NDVI, RVI of Landsat TM7: A Case Study in Karawang, West Java, Indonesia
Thesis Abstract:
Early prediction of paddy field area under cultivation is very important for forecasting rice production, rice consumption, rice shortage, and in policy decision making and government budget planning, especially in countries where rice is as staple food. Many countries use the conventional technique of data collection for crop monitoring and yield estimation based on ground-based visits, farmers’ reports to village leaders, visual estimates by field officers (sub-district level), and number of seed per hectare. These methods are subjective, very costly, and time consuming.
The aim of this study was to extract the paddy field during rice season using the supervised classification: the classifying NDVI, RVI, and three tasselled cap (brightness, greenness, and wetness).
This research used single satellite imagery, Landsat TM7 to predict paddy field based on combining methods in rice irrigation area: supervised classification of original image, supervised classification of tasselled cap transformation, rule base based on NDVI value, rule base based on RVI value, and rule base based on brightness, greenness wetness of tasselled cap.
Pre-image processing consists of preparing image data from HDF format into ERS format, radiometric correction, and geometric correction. Study area was taken in Karawang district, Tempuran sub-district. Field survey was carried out from 15 to 21 July 2003. Data collection such as date transplanting and date harvesting was done by interviewing farmers.
The statistical analysis methods (ANOVA, LSD) were used for supporting distinguish digital number between classes of training area. The results of each method showed that the accuracies of defining paddy field are 99.74 percent of supervised classification in original image, 99.87 percent of supervised classification in tasselled cap transformation, 97.75 percent of classifying paddy field by NDVI, 98.84 percent of classifying paddy field by RVI, 0.25 percent of classifying paddy field by brightness, 98.71 percent of classifying paddy field by greenness, and 64.78 percent of classifying paddy field of wetness. The supervised classification of the original image and the supervised classification of the tasselled cap image were the good methods to identify paddy field. The transformation bands such as NDVI, RVI, and DN of greenness were the fastest and the best methods to identify paddy field area. Their accuracies were not less than 97 percent.
