The content of this article comes from the "Surveying and Mapping Bulletin" No. 7 in 2024, drawing review number: GS Jing (2024) No. 1329
Cotton growth monitoring combined with coefficient of variation method and machine learning model
Yang Sijia1,2, Wang Renjun1,2, Zheng Jianghua1,2, Zhao Pengyu1,2, Han Wanqiang1,2, Mao Xurui1,2, Fan Hong1,2
1. College of Geography and Remote Sensing, Xinjiang University, Urumqi 830046, Xinjiang, China; 2. Oasis Key Laboratory of Xinjiang University, Urumqi 830046, Xinjiang, China
Funds: Remote Sensing Monitoring of Cotton Growth in the Seventh Division of Xinjiang Production and Construction Corps (202105140019)
Key words: cotton, cotton cartographic index, integrated growth monitoring, remote sensing
Citation format: Yang Sijia, Wang Renjun, Zheng Jianghua, et al. Monitoring of Cotton Growth Based on Coefficient of Variation Method and Machine Learning Model[J]. Bulletin of Surveying and Mapping, 2024(7): 111-116. DOI: 10.13474/j.cnki.11-2246.2024.0720
Abstract:In order to obtain the growth information of the key phenological period of cotton more accurately, this paper first extracts the cotton planting area through the cotton cartographic index. Then, the coefficient of variation method was used to construct a comprehensive growth index, namely cotton growth index, which was composed of five indexes reflecting cotton growth, including plant height, SPAD value, leaf wet weight, leaf dry weight and leaf area. Finally, the optimal characteristic variables were selected and combined with the random forest model to construct a cotton growth inversion model. The results showed that: (1) The overall classification accuracy of cotton reached 81.65%; (2) Compared with the five single growth indicators, the constructed FBCGI had a higher correlation with the vegetation index. (3) The R2 and RMSE of the cotton growth monitoring model based on the optimal characteristic variables and random forest model were 0.74, 0.07, 0.51 and 0.10 in the modeling set and the validation set, respectively. The results of this study can provide an important reference for cotton growth monitoring.
About authorAbout author:YANG Sijia (1999—), male, master's degree, mainly engaged in agricultural remote sensing research. E-mail:[email protected] Corresponding author: Zheng Jianghua. E-mail:[email protected]
First trial: Ji Yinxiao review: Song Qifan
Final Judge: Jin Jun
information