###
DOI:
本文二维码信息
地面积雪天气现象人工智能观测识别研究
黄小玉
(国家气象中心)
Research on Artificial Intelligence Observation and Identification of Snow Cover Weather Phenomenon on Ground
HUANG Xiaoyu
(National Meteorological Centre)
摘要
相似文献
本文已被:浏览 83次   下载 12
投稿时间:2018-07-23    修订日期:2018-12-21
中文摘要: 本文应用湖南南岳高山气象观测站及怀化国家气候基准站外场地面积雪摄像图片,对地面积雪人工智能图片采集环境进行分析;基于卷积神经网络技术对南岳站地面积雪实验图片进行建模,并用南岳站与怀化站的测试数据进行验证,主要结论如下:南岳站识别的正确率为99.23%,空判率为0.49%,漏判率为0.28%,白天识别结果优于晚上;积雪形成期间出现概率增加,积雪期出现概率接近100%,且十分稳定,融雪时概率有所降低;地面积雪即将结束及形成初期的晚上图片不清晰时偶尔出现漏判,当有雨凇、雾凇及其他背景污染时会出现个别时刻空判现象。怀化站测试结果与南岳站相似,正确率为97.78%,空判率为1.92%,漏判率为0.3%。但怀化站概率曲线波动较大,一方面由于怀化站图片没有参与建模训练,另一方面可能与怀化站的摄像头固定不佳、对焦不准、拍摄不清晰有关。测试结果表明:积雪模型较好地提取了积雪特征,对于识别积雪有很好的效果,将为天气现象自动观测提供重要参考指导作用。
Abstract:he artificial intelligence identification model was developed based on Convolution Neural Network with the sample of ground snow in Nanyue mountain meteorological observation station in Hunan province, the identification model was tested using the images in Nanyue station and Huaihua and the environments of data acquisition were analyzed. Results show that: recognition accuracy in Nanyue station is 99.23%, false rate was 0.49%, non-response rates was 0.28%, the recognition result is better in daytime than it in night; The accuracy increased to approximately 100% with high stability during the formation stage of snow cover, and it decreased while snow melting. There were a little false cases when the snow cover on the ground is about to melt and in the early stage of snow cover formation as the images were not clear at night, there were occasional misjudgment because of background pollution of fog and rime. The test results of Huaihua station were similar to those of Nanyue, with an accuracy of 97.78%, a false rate of 1.92% and 0.3% missed,but less stable because , firstly the data of Huaihua were not used for model developing, and then the camera was not well fixed led to bad images. The test results show that the snow cover model can extract the characteristics of snow cover well, which has a good effect on the identification of snow cover and will provide an important reference for the automatic observation of weather phenomena.
文章编号:201807230320     中图分类号:    文献标志码:
基金项目:中国气象局小型业务项目
引用文本:
黄小玉,0.[en_title][J].Meteor Mon,():-.
HUANG Xiaoyu,0.Research on Artificial Intelligence Observation and Identification of Snow Cover Weather Phenomenon on Ground[J].Meteor Mon,():-.