基于多源数据的新疆棉田螨害大范围监测研究

doi:10.3964/j.issn.1000-0593(2021)12-3949-08

摘要
参考文献
相关文章 (15)

全文: PDF (5152 KB)
输出: BibTeX | EndNote (RIS)

摘要：针对新疆棉田传统螨害监测方法耗时低效的问题，提出了一种基于冠层高光谱、近地多光谱、环境数据与地面调查相结合的多源数据棉田螨害大范围监测方法。首先，分别采集地面尺度的棉花冠层350~2 500 nm高光谱遥感数据和不同时期低空尺度的棉田无人机多光谱遥感影像数据，通过分析高光谱的原始光谱和一阶微分光谱特征，提取出了4个螨害敏感波段，分别为：绿光波段553 nm附近、红光波段680 nm附近、红边波段680~750 nm、近红外波段760~1 350 nm，这几个波段同时包含在无人机搭载的多光谱传感器波段范围内，验证了低空尺度下无人机遥感螨害监测的可行性。其次，初选23种植被指数和13种田间环境数据，结合地面调查的螨害发生情况做相关性分析。其中，SAVI、OSAVI、TVI、NDGI、平均湿度、温湿系数和10 cm土壤平均温度均与螨害发生达到极显著相关水平（sig≤0.01）；RDVI、RVI、MSR、最高温度、平均温度、积温、10 cm土壤最高温度和10 cm土壤平均湿度均与螨害发生达到显著相关水平（sig≤0.05）。选取sig值在0.05以下的15种特征值，分别建立基于单一环境数据、单一植被指数、环境数据与植被指数相结合的3种支持向量机（SVM）棉田螨害发生监测模型。最后，根据评价效果最优的监测模型，绘制不同时期的螨害遥感监测空间分布图，通过统计分布图中螨害和健康像元数计算出螨害面积占比，将螨害占比与同时期田间环境数据进行相关性分析，筛选出显著特征值，再通过多元逐步回归分析法确定出与螨害面积值关系最密切的环境因子，建立棉田螨害面积预测模型。结果表明：基于单一环境数据的棉田螨害发生监测模型准确率为62.22%，基于单一植被指数的棉田螨害发生监测模型准确率为75.56%，基于环境数据与植被指数相结合的棉田螨害发生监测模型效果最优，准确率为80%。螨害面积预测模型的决定系数R²=0.848，模型拟合度较好。本研究基于多源数据建立的棉田螨害发生监测模型和螨害面积预测模型，可以为新疆地区棉田螨害的大范围监测和趋势预警提供参考。

关键词：无人机遥感；棉叶螨；环境数据；监测；预测

Abstract：Xinjiang’s traditional cotton field spider mite monitoring method is time-consuming and inefficient. The paper proposes combining ground hyperspectral, UAV multi-spectrum, environmental data and field survey for dynamic monitoring of large-scale spider mite damage. Firstly, cotton canopy hyperspectral data and low-altitude-scale UAV multi-spectral data in different cotton periods are collected separately by analyzing the original hyperspectral spectrum and first-order differential spectral characteristics, four sensitive bands of spider mite damage are extracted as below: green light band near 553 nm, red light band near 680 nm, red side band of 680~750 nm, and near infrared band of 760~1 350 nm, which are also included in the multi-spectrum carried by UAV. Secondly, the correlation analysis among 23 vegetation indices, 13 field environmental data, and the occurrence of spider mites surveyed on the ground is done. SAVI, OSAVI, TVI, NDGI, average humidity, temperature-humidity coefficient and average soil temperature of 10 cm are all significantly correlated with spider mite occurrence (sig≤0.01); RDVI, RVI, MSR, maximum temperature, average temperature, accumulated temperature, the highest temperature of 10 cm soil and the average humidity of 10 cm soil all reach a significant correlation level with the occurrence of spider mite damage (sig≤0.05). 15 characteristic values with sig values below 0.05 were selected; cotton field mite monitoring models based on single environmental data, single vegetation indices, and a combination of environmental data and vegetation indices are established respectively using support vector machine (SVM). Finally, from the optimum model, we can draw the spatial distribution map of spider mite damage in different periods and calculate the proportion of spider mite damage are based on the number of spider mite damage and healthy pixels in the statistical distribution map. Then the field environmental data is analyzed for correlation, the environmental factors most closely related to the spider mite area value are determined by multiple stepwise regression analysis, and the cotton field spider mite area prediction model is established. The results show that the accuracy rate of the cotton field spider mite monitoring model based on a single environmental data is 62.22%, while the accuracy rate of the cotton field mite monitoring model based on a single vegetation index is 75.56%. Moreover, the most effective model is based on the combination of environmental data and vegetation indices with an accuracy rate of 80%. The coefficient of determination of the spider mite area prediction model is R²=0.848. In this study, based on multi-source data, the cotton field spider mite occurrence monitoring model and spider mite area prediction model can provide a reference for the large-scale monitoring and trend warning of cotton field mite damage in Xinjiang.

Key words：Unmanned aerial vehicle remote sensing; Cotton spider mite; Environmental data; Monitoring; Prediction

收稿日期: 2020-11-03 修订日期: 2021-02-19

中图分类号:

TP751

基金资助: 国家重点研发计划项目（2016YFB0501805）资助

通讯作者: 吴才聪 E-mail: wucc@cau.edu.cn

作者简介: 杨丽丽，女，1974年生，中国农业大学信息与电气工程学院副教授 e-mail：llyang@cau.edu.cn

引用本文:

杨丽丽，王振鹏，吴才聪. 基于多源数据的新疆棉田螨害大范围监测研究[J]. 光谱学与光谱分析, 2021, 41(12): 3949-3956.
YANG Li-li, WANG Zhen-peng, WU Cai-cong. Research on Large-Scale Monitoring of Spider Mite Infestation in Xinjiang Cotton Field Based on Multi-Source Data. SPECTROSCOPY AND SPECTRAL ANALYSIS, 2021, 41(12): 3949-3956.

链接本文:

https://www.gpxygpfx.com/CN/10.3964/j.issn.1000-0593(2021)12-3949-08 或 https://www.gpxygpfx.com/CN/Y2021/V41/I12/3949

[1] YAN Jian-feng(闫建峰). Village Science and Technology(乡村科技), 2020, 11(22): 47.
[2] YANG Li-li, ZHANG Da-wei, LUO Jun, et al(杨丽丽，张大卫，罗君，等). Transactions of the Chinese Society for Agricultural Machinery(农业机械学报), 2019, 50(2): 14.
[3] LAN Yu-bin, ZHU Zi-hao, DENG Xiao-ling, et al(兰玉彬，朱梓豪，邓小玲，等). Transactions of the Chinese Society of Agricultural Engineering(农业工程学报), 2019, 35(3): 92.
[4] HUANG Huan-hua, MA Xiao-hang, HUANG Hua-yi, et al(黄焕华，马晓航，黄华毅，等). Journal of Environmental Entomology(环境昆虫学报) , 2018, 40(2): 306.
[5] TIAN Ye, ZHANG Qing, LI Xi-can, et al(田野, 张清，李希灿，等). Arid Zone Research(干旱区研究)，2017，34(2)：423.
[6] WANG Shou-hui, DAI Jian-guo, LI Shuan-ming, et al(王守会，戴建国，李栓明，等). Jiangsu Agricultural Sciences(江苏农业科学), 2018, 46(13): 96.
[7] CUI Mei-na, DAI Jian-guo, WANG Shou-hui, et al(崔美娜，戴建国，王守会，等). Xinjiang Agricultural Sciences(新疆农业科学) , 2018, 55(8): 1457.
[8] ZHAO Liang, CHEN Bing, XIAO Chun-hua, et al(赵亮，陈兵，肖春华，等). Journal of Gansu Agricultural University(甘肃农业大学学报), 2015, 50(5): 94.
[9] LIU Hong-xia(刘红霞). China Cotton(中国棉花), 2016, 43(1): 45.
[10] MA Hui-qin, HUANG Wen-jiang, JING Yuan-shu, et al(马慧琴，黄文江，景元书，等). Transactions of the Chinese Society of Agricultural Engineering(农业工程学报), 2017, 33(5): 162.
[11] QU Sha, LI Zhen-hai, QIU Chun-xia, et al(屈莎，李振海，邱春霞，等). Transactions of the Chinese Society of Agricultural Engineering(农业工程学报), 2017, 33(12): 186.
[12] Kanke Y, Tubaña B, Dalen M, et al. Precision Agriculture, 2016, 17(5): 507.
[13] Cao Qiang, Miao Yuxin, Shen Jianning, et al. Precision Agriculture, 2016, 17(2): 136.