文章基本信息

标题：Crop Phenology-Based, Object-Oriented Classification Approach Using SENTINEL-2A and NDVI Time Series: Sunflower Crops in Kırklareli TURKEY
本地全文：下载
作者：Armağan ALOE KARABULUT ; Nihal CEYLANW ; Erdem BAHAR 等
期刊名称：International Journal of Environment and Geoinformatics
电子版ISSN：2148-9173
出版年度：2021
卷号：8
期号：3
页码：316-327
语种：English
出版社：IJEGEO
摘要：Achieving food security on a global scale depends on regular spatio-temporal monitoring and management of national and local agricultural production. The aim of this study is to develop a methodology for determining sunflower cultivated areas using high resolution time series of the SENTINEL-2A satellite images that represent phenological stages of crop growth cycle. A time series of spectral signatures of crop phenological periods and normalized difference vegetation index (NDVI) was produced from satellite images for year 2018. A stepwise object-oriented classification approach was developed. In this approach, object-oriented segmentation and classification decision tree algorithms were produced by using time series of spectral signatures and NDVI values as well as object shape criteria and other auxiliary thematic maps. The multiresolution method of "Canny edge” algorithm was used in order to determine boundary of agricultural parcels. The best performance in segmentation to determine the agricultural parcels was achieved by increasing weight coefficient of the "Canny edge” layer. Object-oriented classification was carried out based on these segmented parcels. First, summer crops, winter crops, fallow and permanent vegetated areas were determined through classification decision tree algorithms. Later, the summer and winter crops were classified using the parcel spectral signatures of samples collected with field work. The crops whose class definition could not be determined were passed through a second elimination in "unclassified" group and later assigned to their classes. Finally, the parcels whose class definition could not be determined were grouped as "other" class. According to results of confusion matrix and accuracy analysis, sunflower, which was determined in two classes as early and late sowing, was classified at 98% and 92% accuracy, respectively.
关键词：Object-oriented Classification;NDVI Time Series;Sunflower;Crop Phenology;SENTINEL-2A;Decision Tree