Vol. 100
Latest Volume
All Volumes
PIERB 100 [2023] PIERB 99 [2023] PIERB 98 [2023] PIERB 97 [2022] PIERB 96 [2022] PIERB 95 [2022] PIERB 94 [2021] PIERB 93 [2021] PIERB 92 [2021] PIERB 91 [2021] PIERB 90 [2021] PIERB 89 [2020] PIERB 88 [2020] PIERB 87 [2020] PIERB 86 [2020] PIERB 85 [2019] PIERB 84 [2019] PIERB 83 [2019] PIERB 82 [2018] PIERB 81 [2018] PIERB 80 [2018] PIERB 79 [2017] PIERB 78 [2017] PIERB 77 [2017] PIERB 76 [2017] PIERB 75 [2017] PIERB 74 [2017] PIERB 73 [2017] PIERB 72 [2017] PIERB 71 [2016] PIERB 70 [2016] PIERB 69 [2016] PIERB 68 [2016] PIERB 67 [2016] PIERB 66 [2016] PIERB 65 [2016] PIERB 64 [2015] PIERB 63 [2015] PIERB 62 [2015] PIERB 61 [2014] PIERB 60 [2014] PIERB 59 [2014] PIERB 58 [2014] PIERB 57 [2014] PIERB 56 [2013] PIERB 55 [2013] PIERB 54 [2013] PIERB 53 [2013] PIERB 52 [2013] PIERB 51 [2013] PIERB 50 [2013] PIERB 49 [2013] PIERB 48 [2013] PIERB 47 [2013] PIERB 46 [2013] PIERB 45 [2012] PIERB 44 [2012] PIERB 43 [2012] PIERB 42 [2012] PIERB 41 [2012] PIERB 40 [2012] PIERB 39 [2012] PIERB 38 [2012] PIERB 37 [2012] PIERB 36 [2012] PIERB 35 [2011] PIERB 34 [2011] PIERB 33 [2011] PIERB 32 [2011] PIERB 31 [2011] PIERB 30 [2011] PIERB 29 [2011] PIERB 28 [2011] PIERB 27 [2011] PIERB 26 [2010] PIERB 25 [2010] PIERB 24 [2010] PIERB 23 [2010] PIERB 22 [2010] PIERB 21 [2010] PIERB 20 [2010] PIERB 19 [2010] PIERB 18 [2009] PIERB 17 [2009] PIERB 16 [2009] PIERB 15 [2009] PIERB 14 [2009] PIERB 13 [2009] PIERB 12 [2009] PIERB 11 [2009] PIERB 10 [2008] PIERB 9 [2008] PIERB 8 [2008] PIERB 7 [2008] PIERB 6 [2008] PIERB 5 [2008] PIERB 4 [2008] PIERB 3 [2008] PIERB 2 [2008] PIERB 1 [2008]
2023-05-20
Estimation of Thickness and Dielectric Characteristics of Sea Ice from Near-Field EM Measurements Using Deep Learning for Large Scale Polar Ice Probing
By
Progress In Electromagnetics Research B, Vol. 100, 73-89, 2023
Abstract
The near and far field EM responses over layered media have long been exploited in diversified applications such as remote sensing, monitoring and communication. In this work, we utilize the near field dependence of the EM fields of a three layered structure resembling air-sea ice-sea water to estimate the thickness and dispersion characteristics of sea ice using deep learning technique. We explore two key methods of field measurement termed as the fixed and scaled sweep methods. In the fixed radial sweep method, the receiver distance and height from the source are kept constant, and in the scaled sweep method both the receiver distance and height are set as a scaled function of the operating wavelength. A synthetic training dataset has been generated (using analytical computation and FEM simulation) in the low MHz band, which is used to train a deep learning model. The model is tested on different test datasets with frequencies inside, below and above the training limits. Even though the fixed sweep method is simpler to implement, the scaled sweep appears to perform better across the wide range of test frequency, both in and outside the training range. When the test frequency is inside the training range, the percentage errors for thickness, dielectric constant, and loss tangent were found to be <2%, <10%, and <5%, respectively, for the fixed radial sweep, whereas for the scaled sweep the percentage error is < 1% for all three measurement parameters. When the test frequency deviates further from the training range, the percentage error gradually increases. Later, we investigate the problem of determining sea ice thickness assuming a priori knowledge of sea ice dielectric parameters, and results show that the model estimates the thickness of the sea ice bulk with error as low as 0.1%.
Citation
Mohammad Shifatul Islam Sadman Shafi Mohammad Ariful Haque , "Estimation of Thickness and Dielectric Characteristics of Sea Ice from Near-Field EM Measurements Using Deep Learning for Large Scale Polar Ice Probing," Progress In Electromagnetics Research B, Vol. 100, 73-89, 2023.
doi:10.2528/PIERB22122005
http://www.jpier.org/PIERB/pier.php?paper=22122005
References