Highlights An ANN based model has been suggested for values over EIA region stations during 1958–2008. The developed model predicts with a root mean square error of less than 1 MHz for the unseen testing data. The model may be used for practical purposes over Asian longitude sector, particularly for Pakistan.
Abstract The critical frequency of -layer in the Equatorial Ionization Anomaly (EIA) region is one of the most challenging ionospheric characteristics to model. We illustrate the efforts to model monthly hourly medians of using Artificial Neural Network (ANN) over five EIA crest stations between 23.1° N – 30.5° N of Eastern longitudes during 1958–2008 (as per data availability per station). The ANN based model was designed to obtain values as output by inputting seven parameters defining diurnal (hour), seasonal (month), geographical (latitude & longitude) and solar cycle (year, F10.7 & F10.7P indices) dependency of . With this input space, the model uses a hidden layer of 96 neurons to generate output , producing a mean squared error of less than . Additionally, the convergence of the model improved on introducing derived solar radio flux (F10.7P) by 6.0%. After training and validation, the model predicts values over the locations under study for unseen testing data of 1989, 1998 & 2008 as a representation of high, moderate and low solar activity, respectively. Our ANN based climatological model gave a Root Mean Square Error (RMSE) of 0.956 MHz and goodness of fit, =0.929 for the testing data. Data-model comparison with ANN model is better than the International Reference Ionosphere (IRI) estimations of with CCIR (IRIC) and URSI (IRIU) coefficients. IRIC and IRIU produced RMSE of 1.901 MHz () and 1.926 MHz (=0.716) for the same testing data set, respectively. The designed ANN based model successfully captures the climatological features of in the EIA region. After necessary modifications, the model may be applied for predicting daily hourly values, particularly in the Asian-longitudinal range over Pakistan, in future.
Modelling of using artificial neural network over Equatorial Ionization Anomaly (EIA) region stations
Advances in Space Research ; 72 , 12 ; 5539-5550
2022-09-15
12 pages
Article (Journal)
Electronic Resource
English
<math xmlns="http://www.w3.org/1998/Math/MathML"><mrow><msub><mrow><mi>f</mi></mrow> <mrow><mi>o</mi></mrow></msub> <msub><mrow><mi>F</mi></mrow> <mrow><mn>2</mn></mrow></msub></mrow></math> , Equatorial Ionization Anomaly (EIA) , Artificial Neural Network (ANN) , International Reference Ionosphere (IRI) , Solar radio flux , Karachi