GF-1卫星多时相组合近红外数据水稻识别能力

发布时间：2018-02-01 21:04

本文关键词： 遥感作物面积识别多时相近红外波段 GF-/WFV 水稻　出处：《农业工程学报》2017年23期 　论文类型：期刊论文

【摘要】：针对近红外波段水稻识别能力的问题,选择银川市所属的5个县区为研究区域,采用2016年5月18日、6月16日、7月30日、9月13日4个时相GF-1/WFV影像的近红外波段(0.76~2.526μm)数据,基于决策树分类方法,获取了4个单时相、3个多时相条件下的水稻识别结果,并与全波段数据分类结果进行了比较。单时相5、6、7和9月份近红外波段水稻识别精度分别为83.63%、57.40%、75.82%和62.61%,除5月份精度高于全波段5.75个百分点外,其他时相都低于全波段识别精度,6月份相差最高为30.23个百分点。多时相5/6、5/7、5/6/7/9月份组合,近红外水稻识别精度分别为83.76%、93.93%和94.03%,分别比全波段低5.47,高8.58和0.73个百分点。结果表明,水稻生长早期的5月份、中期的7月份,近红外波段可以作为单时相遥感识别数据源,包括生长早期和中期2个时相在内的多时相近红外波段组合都可以作为遥感识别的数据源,研究结果可以作为GF-1数据水稻遥感识别的依据。
[Abstract]:In order to solve the problem of rice recognition ability in near infrared band, five counties of Yinchuan city were selected as study areas, May 18th 2016, June 16th and July 30th were adopted. In September 13th, four single phase GF-1/WFV images were obtained based on the decision tree classification method based on the near infrared band (0.76 ~ 2.526 渭 m) of four temporal GF-1/WFV images. The results of rice identification under three multi-phase conditions were compared with the results of full-band data classification. The accuracy of rice recognition in single-phase 5Y6H7 and NIR region in September was 83.63%, respectively. The accuracy of 57.40% and 62.61% is lower than that of full band recognition except that in May the accuracy is higher than that in the whole band by 5.75%. In June, the maximum difference was 30.23 percentage points. In the combination of 5 / 6 / 7 / 5 / 7 / 7 / 7 / 9, the accuracy of near-infrared rice recognition was 83.76%, respectively. The values of 93.93% and 94.03 were 5.47 lower and 8.58 and 0.73 percentage points higher than those of the whole band, respectively. The results showed that the rice growth period was May in the early stage and July in the middle period. The near infrared band can be used as the data source of single phase remote sensing recognition, and the combination of multi-time and similar infrared bands, including the early and middle phases of growth, can be used as the data source of remote sensing recognition. The results can be used as the basis of rice remote sensing recognition based on GF-1 data.
【作者单位】：中国农业科学院农业资源与农业区划研究所;
【基金】：国家重点研发计划“粮食作物生长监测诊断与精确栽培技术”课题“作物生长与生产力卫星遥感监测预测”(2016YFD0300603)
【分类号】：S511;TP79
【正文快照】： 0引言水稻是中国三大粮食作物之一,准确掌握水稻种植面积是中国农作物种植结构调整的依据。遥感技术具有客观性强,覆盖范围广的特点,是区域农作物面积信息准确获取的主要技术[1]。近红外谱段(0.76~2.526μm)介于可见光与中红外光之间,水稻在近红外波段有较高的反射率,而水体则

【相似文献】