Estimating Tea Plantation Area Based on Multi-source Satellite Data

Abstract

Tea is a characteristic cash crop native to China, mainly distributed in the south of the Yangtze River. Obtaining the planting area and spatial distribution of tea gardens is of great significance to improve the economic and ecological benefits of tea. In this paper, a method for extracting tea plantation area based on multi-source remote sensing satellite data is proposed. We collect the Landsat8 OLI′ Sentinel -2′ HJ-IA/B and GF-1 WFV data from 2017 to 2018, and then we do the pre-processing for all the remote sensing data, calculate the Normalized Difference Vegetation Index(NDVI) of the data, calculate the spectral characteristics of the data and obtain the Gabor textual characteristics after principal component analysis(PCA) of the data. In order to obtain the time-series data, all features of Sentinel-2′Y HJ-IA/B and GF-1 WFV data are relatively calibrated to Landsat8 OLI data, and finally the tea plantation area is extracted by support vector machine (SVM) classifier. We extract the area of tea garden of Huzhou City, Zhejiang Province, and the result is 235.68 km2 and the results were verified by precision. The results show that this method can obtain high precision for the extraction of tea garden area, which is of great significance for further production and application.