Preliminary investigation of real‐time mapping of foF2 in northern China based on oblique ionosonde data
Preliminary investigation of real‐time mapping of foF2 in northern China based on oblique ionosonde data
复制标题
DOI：
10.1002/jgra.50262
复制
发表时间：
2013-05
期刊：
影响因子：
--
通讯作者：
Chen Zhou;Ruopeng Wang;Wenyu Lou;Jing Liu;B. Ni;Zhongxin Deng;Zhengyu Zhao
中科院分区：
文献类型：
--
作者：
Chen Zhou;Ruopeng Wang;Wenyu Lou;Jing Liu;B. Ni;Zhongxin Deng;Zhengyu Zhao
利用神经网络（NN）建立了中国北方 foF2 的实时测绘模型。为了避免与传统神经网络相关的局部最小值问题，使用太阳活动、地磁活动、中性风、季节信息和地理坐标的输入参数，开发了一种新改进的基于遗传算法的神经网络（GA-NN）。 foF2数据是通过对2009年8月至2011年12月期间中国地基地震电离层监测网每30 min倾斜电离图进行反演而提取的。以北京、长春、青岛、新乡、苏州5个发射站和滨州1个接收站的数据作为实时foF2测图模型的输入参数，将数据大连和金阳发射站的数据被用来验证结果。利用济宁发射站数据对模型的性能进行了测试。计算均方根误差和百分比偏差来估计模型的性能。相关系数用于评估观测值和预测值的相关性。此外，将北京、长春、青岛、苏州站垂直电离探空仪观测到的foF2与模型预测的foF2进行了比较。结果表明，基于遗传算法的神经网络开发的实时 foF2 映射模型对于电离层研究非常有前景。
A real‐time mapping model of foF2 in northern China was established using neural networks (NNs). To avoid the local minimum problem associated with traditional NNs, a newly improved genetic algorithm‐based NN (GA‐NN) was developed using the input parameters of solar activities, geomagnetic activities, neutral winds, seasonal information, and geographical coordinates. The foF2 data were extracted by inversing the oblique ionograms obtained from the oblique ionosondes of the China Ground‐based Seismo‐ionospheric Monitoring Network every 30 min for the period from August 2009 to December 2011. The data associated with five transmitter stations (Beijing, Changchun, Qingdao, Xinxiang, and Suzhou) and one receiver station in Binzhou were considered the input parameters for the real‐time foF2 mapping model, and the data from the Dalian and Jinyang transmitter stations were used to verify the results. The Jining transmitter station data were used to test the capability of the model. The root‐mean‐square error and percent deviation were calculated to estimate the performance of the model. The correlation coefficient was used to evaluate the correlation of observed and predicted values. In addition, observations of foF2 from the vertical ionosondes at Beijing, Changchun, Qingdao, and Suzhou stations are compared with the model prediction of foF2. The results indicate that the developed real‐time foF2 mapping model based upon genetic algorithm‐based NN is very promising for ionospheric studies.