Scribd
Upload
95 views1 page

Mapping Paddy Rice Agriculture Using Multi-Temporal FORMOSAT-2 Images

This study used FORMOSAT-2 satellite images from 1996-2006 to map and monitor paddy rice agriculture in Yunlin County, Taiwan. Paddy rice fields were previously monitored using annual aerial photographs, but this method was time-consuming and covered small areas. Satellite images offer higher temporal resolution and wider area coverage. Images from the transplanting and tillering stages of rice growth were combined and classified using supervised classification. Object-based post-classification, where parcels were classified based on the pixel classification results, performed better than pixel-based or object-based pre-classification methods. Some errors occurred due to similar growth patterns between rice and other crops and changes in rice cultivation practices over time.

Uploaded by

Nurul Affizah Noor Azmi
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
95 views1 page

Mapping Paddy Rice Agriculture Using Multi-Temporal FORMOSAT-2 Images

This study used FORMOSAT-2 satellite images from 1996-2006 to map and monitor paddy rice agriculture in Yunlin County, Taiwan. Paddy rice fields were previously monitored using annual aerial photographs, but this method was time-consuming and covered small areas. Satellite images offer higher temporal resolution and wider area coverage. Images from the transplanting and tillering stages of rice growth were combined and classified using supervised classification. Object-based post-classification, where parcels were classified based on the pixel classification results, performed better than pixel-based or object-based pre-classification methods. Some errors occurred due to similar growth patterns between rice and other crops and changes in rice cultivation practices over time.

Uploaded by

Nurul Affizah Noor Azmi
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as DOCX, PDF, TXT or read online on Scribd
You are on page 1/ 1

Mapping Paddy Rice Agriculture Using Multi- Temporal FORMOSAT-2 Images Most of the Asian make rice as their

staple food. This causes paddy rice as the major crop in Asia. Monitoring and mapping paddy rice agriculture in a timely and efficient manner is therefore important for agricultural and environmental sustainability, food and water security, and greenhouse gas emissions. Cultivated paddy fields were monitored by the Agriculture and Food Agency in Taiwan using annually interpreting aerial photographs. However, this method has some disadvantages which are time and energy consuming, covering small areas and hinders the work of getting aerial photograph for the proper growing time. Due to the disadvantages, satellite images have become alternative to aerial photographs because it can cover wide areas and have higher temporal resolutions. This study was conducted in Yunlin County, Taiwan. Chinese Society of Photogrammetry and Remote Sensing produced data from 1996 to 2005 to produce training areas for the supervised classificasition, and the data of 2006 to validate the classification results. Since paddy rice has special growing characteristics occur in transplanting stage and tillering stage, so it is easier to recognize the texture of agricultural crops directly on FORMOSAT-2 multi-spectral satellite images. Images for both stages are combined into one image with 8-layer spectral information for the supervised classification. Normalized difference vegetation index (NDVI) values also included. Two assumptions have been made for the selection of training areas. It states that a field that had been paddy field for past ten years (1996-2005) will remain be the paddy field this year (2006) and vice versa. This study used three methods of image classification which are pixel-based classification, object-based pre-classification and objectbased post-classification. For pixel-based, the images that had been arranged from transplanting and tillering were masked using extracted rice and non-rice sites. For objectbased pre-classification, cultivation parcel was the classification unit. Firstly, we calculated the pixels mean values after found them in each parcel. The mean values represented the average of every original band and the average NVDI value during transplanting stage. For object-based post-classification, the classification unit also the parcel but we determined the class of parcel by using the pixel-based classification result. If half of the pixels covered in one parcel are non-rice, then the parcel is assumed the class of non-rice, and vice versa. As conclusion, we can see that the post-classification performed better than pixel-based and preclassification methods. Errors occurred from two conditions which are same growing pattern for rice and some non-rice crops and discrepancy of paddy rice cultivation habits.

You might also like