A phenological object-based approach for rice crop classification using time-series Sentinel-1 Synthetic Aperture Radar (SAR) data in Taiwan

Nguyen Thanh Son, Chi Farn Chen, Cheng Ru Chen, Piero Toscano, Youg Sing Cheng, Hong Yuh Guo, Chien Hui Syu

Research output: Contribution to journalArticlepeer-review

33 Scopus citations

Abstract

Spatial assessment of rice-cultivated areas is a crucial activity in Taiwan due to government initiatives to provide policymakers with reliable and timely information for devising successful rice grain import and export plans. The objective of this study is to develop a phenological object-based approach for collectively mapping patches of rice fields using the time-series Sentinel-1 Synthetic Aperture Radar (SAR) data. We processed the data for 2019 cropping seasons, following three main steps: (1) data pre-processing to construct the smooth twelve-day time-series composite vertical transmit and horizontal receive (VH) polarized data, (2) object-based image classification of rice-cultivated areas using phenological metrics, and (3) accuracy assessment of the mapping results. The classification maps compared with the government’s ground reference maps indicated satisfactory accuracies, with producer’s accuracies of 84.2% and 82.6% and user’s accuracies of 82.1% and 85.3% for the first and second crops, respectively. These results were reaffirmed by close agreement in area estimates between the satellite-derived rice area and government’s reference data at township level, with coefficient of determination (R 2) values of 0.96 and 0.94 for the first and second crops, respectively. In spite of some error sources, including mixed-pixel issues and edge effects, that lowered the mapping accuracy, the results of this study have demonstrated that our mapping approach using the time-series Sentinel-1 VH data and information of crop phenology could be potentially applied at a larger scale in Taiwan and transferable to other regions for updating rice crop maps on a timely and frequent basis.

Original languageEnglish
Pages (from-to)2722-2739
Number of pages18
JournalInternational Journal of Remote Sensing
Volume42
Issue number7
DOIs
StatePublished - 2021

Fingerprint

Dive into the research topics of 'A phenological object-based approach for rice crop classification using time-series Sentinel-1 Synthetic Aperture Radar (SAR) data in Taiwan'. Together they form a unique fingerprint.

Cite this