Page Header

Identification of citrus orchard under vegetation indexes using multi-temporal remote sensing

Chenxin Liang, Qiting Huang, Si Wang, Cong Wan, Qiangyi Yu, Wenbin Wu


Citrus has been widely planted in the south of China in recent years. However, the multiple crop types and frequent cloud cover locally have posed a great challenge to the direct identification of the large-scale citrus orchard using the spectral information. In this study, a systematic approach was developed to identify the citrus orchard using the phenological characteristics of citrus during the fruit expansion stage, where the threshold value was determined to the key time window.
Taking the Wuming District of Nanning Guangxi Zhuang Autonomous Region in southwest China as the study area, an empirical investigation was carried out using the Google Earth Engine platform. 1 751 ground samples were also collected in the field for validation. Meanwhile, the cloud coverage assessment was performed on the Sentinel-2 images over the whole year of 2018. According to the citrus phenology in the study area, the characteristics of flowering were not outstanding in the first half of 2018 (the flowering period of citrus), while the key identification features were found in the second half of 2018 (the peak stage of citrus fruit growth and expansion). As such, the second half of 2018 was determined as the study period, considering the data availability and citrus phenological stage. Then, a multi-temporal image dataset was obtained from August to December. The specific procedure was as follows. Firstly, some indices were calculated using the time series Sentinel-2 data in 2018, including the multiple vegetation indices (e.g. Normalized Difference Vegetation Index (NDVI), Green Normalized Difference Vegetation Index (GNDVI), Difference Vegetation Index (DVI)), and Sentinel-derived Red-Edge Spectral Indices (RESI). The vegetation information of multi-temporal remote sensing was then obtained for the entire study area. Secondly, the vegetation information of various planting types was compared in different growing periods, according to the measured ground data. The results showed that there was no outstanding spectral difference between the citrus orchards and other vegetation types (e.g. sugarcane, banana, maize, and rice) in the study area. However, the vegetation indices of multi-temporal remote sensing presented that the NDVI of 0.47 for the citrus orchards was distinctly lower than that for the other vegetation types in October. There was also a lower GNDVI of citrus orchards in October and November. But there was no difference in the DVI of citrus orchards from other vegetation. Furthermore, the fruit expansion stage of citrus was located from September to October, indicating weak vegetation information of citrus. Nevertheless, there were significant differences among different vegetation indices, where the NDVI presented the highest. In addition, a renormalization of vegetation indices was further constructed to identify the spatial distribution of citrus in the study area by a threshold, according to the NDVI in October. The overall accuracy of citrus orchards reached 82.75% using the renormalization of NDVI, where the Kappa coefficient was 0.66, indicating a better identification, compared with the rest of renormalized vegetation indices (GNDVI, DVI, and RESI). The NDVI presented more complete identification for the citrus orchards, whereas, the GNDVI performed more fragmented identification. Consequently, the total area of citrus planting was calculated as 3.42×104 hm2 using the phenological parameters of citrus orchards. This finding can provide strong theoretical and practical support to timely mapping the citrus orchards using remote sensing.


remote sensing; classification; phenology; citrus; vegetation index; Sentinel-2; Google Earth Engine

Full Text:




  • There are currently no refbacks.

Copyright (c) 2022 Chenxin Liang, Qiting Huang, Si Wang, Cong Wan, Qiangyi Yu, Wenbin Wu

Creative Commons License
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.

Creative Commons License
This site is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.