摘要
旱灾频繁发生给农业造成严重损失,将旱情监测视作异常信息识别过程,利用多分类器融合方法探讨集成多源遥感空间数据的高精度农业旱情等级评估模型。首先对各类旱情关联因子进行相关性分析,选择最优模型输入参数;然后利用3种单分类器(神经网络、支持向量机以及分类回归树)对多源遥感数据进行分析,构建阈值优化模糊投票法(threshold-optimized fuzzy majority voting,TFMV)对单分类器旱情等级评估结果进行决策级融合。结果表明:TFMV方法总体分类精度为72.55%,分别比神经网络、支持向量机、分类回归树高出约3.6、5.1和3.6个百分点;与经典投票法相比,TFMV方法分类精度也提高了约2.5个百分点。TFMV方法能有效提高单分类器旱情评估精度,多数情况下能较准确反映研究区干旱受灾区域以及旱情程度,具有应用于实际区域旱情等级评估业务的潜力。
Drought affects not only agriculture, but also triggers negative economic, social, and environmental impacts. Thisstudy proposes a multiple classifier fusion method, threshold-optimized fuzzy majority voting （TFMV）, for agriculturaldrought category evaluation. The standardized precipitation index （SPI） can be flexibly designed to measure drought severityfor a specific time period. The 3-month SPI （SPI-3） was computed based on the long-term monthly precipitation record.Considering that the relationship between different remote sensing drought indices and SPI varies over time, the correlationcoefficients were calculated between the remote sensing drought indices of each month from April to October, and SPI-3 from2003 to 2012 to select the input data of model. The results showed that the correlation coefficient values between thevegetation-related indices and SPI-3 varied over the different time periods. The VCI（vegetation condition index） showed thehighest correlation with the SPI-3 in August because of the vegetation phenological phase. The correlation coefficient betweenthe TCI（temperature condition index） and SPI-3 were statistically significant （P〈0.01） except in May. All the correlationcoefficient between soil moisture-related drought indices and SPI-3 were statistically significant （P〈0.01） and all thecorrelation coefficients were above 0.45. The precipitation-related indices with various timescales showed high correlationcoefficient with the in situ drought index, and these indices mostly have the highest correlation coefficient values with in situdrought index in the same timescales as that of the precipitation-related indices. Based on correlation analysis between remotesensing drought index and in situ drought index over the different time periods, VCI, TCI（temperature condition index）,SMCI（soil moisture condition index）） and PCI-3 were selected as the input data of model. Since the distribution of the trainingdata among drought classes is uneven, the synthetic minority over-sampling technique （SMOTE） method was used to balanceimbalanced training datasets. Three typical classifiers： Back-propagation neural network （BPNN）, support vector machines（SVM） and classification and regression trees （CART） were applied for assessment of regional drought category. The resultsshowed that the capability of each single classifier in drought grade classification varies along seasonal time and the overallprecision of these three classifiers for all samples from April to October were 69% （BPNN）, 67.49% （SVM） and 69% （CART）,respectively. Considering the limitation of single classifier, two classifier ensemble methods, majority voting （MV） andthreshold-optimized fuzzy majority voting （TFMV） were introduced to fuse the three single drought category results.Experimental results clearly demonstrated that： 1） Ensemble method could improve overall classification accuracy; 2） TFMVensemble method performed the highest overall accuracy in validation dataset, which was respectively 3.6, 5.1 and 3.6 percentpoint higher than that of BPNN, SVM and CART classification. Additionally, compared with majority voting method, TFMVachieved more accurate classification results in all different time periods. Additionally, the spatial drought conditions of theTFMV maps were compared with the actual drought intensity using the agro-meteorological disaster data recorded and thetemporal distribution of the precipitation and mean temperature data at the agro-meteorological sites. Results showed that theTFMV maps exhibited consistent variations with the in situ reference data. The practical application of TFMV demonstratedthat it can provide accurate and detailed drought condition and TFMV method can be effectively used for regional agriculturaldrought category evaluation.
作者
董婷
任东
孟令奎
张文
邵攀
Dong Ting;Ren Dong;Meng Lingkui;Zhang Wen;Shao Pan(College of Computer and Information Technology,Three Gorges University,Yichang 443002,China;School of Remote Sensing and Information Engineering,Wuhan University,Wuhan 430079,China;Department of Land Surveying and Geo-informatics,The Hong Kong Polytechnic University,Hongkong,China)
出处
《农业工程学报》
EI
CAS
CSCD
北大核心
2018年第12期137-145,共9页
Transactions of the Chinese Society of Agricultural Engineering
基金
地理国情监测国家测绘地理信息局重点实验室开放基金(2017NGCM04)
湖北省农田环境监测工程技术研究中心开放基金课题(201603
201610)
国家重点研发计划(2016YFD0800902)
关键词
干旱
遥感
算法
多分类器融合
标准化降水指数
阈值优化模糊投票法
drought
remote sensing
algorithms
multiple classifier fusion
standardized precipitation index
threshold-optimizedfuzzy majority voting
