Rainfall Rate Estimation Using Terahertz Channel Performance and Deep Neural Networks

Wanzhu Chang; Yuheng Song; Wenbo Liu; Mingxia Zhang; Jiabiao Zhao; Peian Li; Jianjun Ma; Houjun Sun

doi:10.1109/UCMMT67044.2025.11287089

Rainfall Rate Estimation Using Terahertz Channel Performance and Deep Neural Networks

Wanzhu Chang
, Yuheng Song
, Wenbo Liu
, Mingxia Zhang
, Jiabiao Zhao
, Peian Li
, Jianjun Ma^*
, Houjun Sun

^*Corresponding author for this work

Beijing Institute of Technology
Tangshan Research Institute

Research output: Contribution to journal › Conference article › peer-review

Abstract

This work investigates rainfall intensity estimation using terahertz (THz) channel performance and deep neural networks. A THz time-domain spectroscopy system captured channel measurement data under controlled rain at multiple frequencies. Four architectures-CNN, RNN, LSTM, and BiGRU - were trained and evaluated. Results show that model performance varies with frequency, CNN performs well across most bands, while BiGRU excels at the mid-high frequency of 220 GHz.

Original language	English
Journal	UK, Europe, China Millimetre Waves and THZ Technology Workshop, UCMMT
Issue number	2025
DOIs	https://doi.org/10.1109/UCMMT67044.2025.11287089
Publication status	Published - 2025
Externally published	Yes
Event	18th IEEE United Conference on Millimeter Waves and Terahertz Technologies, UCMMT 2025 - Nanjing, China Duration: 25 Aug 2025 → 28 Aug 2025

Access to Document

10.1109/UCMMT67044.2025.11287089

Cite this

@article{a3c7c9869b2e42439172014ac1d39df4,

title = "Rainfall Rate Estimation Using Terahertz Channel Performance and Deep Neural Networks",

abstract = "This work investigates rainfall intensity estimation using terahertz (THz) channel performance and deep neural networks. A THz time-domain spectroscopy system captured channel measurement data under controlled rain at multiple frequencies. Four architectures-CNN, RNN, LSTM, and BiGRU - were trained and evaluated. Results show that model performance varies with frequency, CNN performs well across most bands, while BiGRU excels at the mid-high frequency of 220 GHz.",

author = "Wanzhu Chang and Yuheng Song and Wenbo Liu and Mingxia Zhang and Jiabiao Zhao and Peian Li and Jianjun Ma and Houjun Sun",

note = "Publisher Copyright: {\textcopyright} 2025 IEEE.; 18th IEEE United Conference on Millimeter Waves and Terahertz Technologies, UCMMT 2025 ; Conference date: 25-08-2025 Through 28-08-2025",

year = "2025",

doi = "10.1109/UCMMT67044.2025.11287089",

language = "English",

journal = "UK, Europe, China Millimetre Waves and THZ Technology Workshop, UCMMT",

issn = "2639-4537",

publisher = "Institute of Electrical and Electronics Engineers Inc.",

number = "2025",

}

TY - JOUR

T1 - Rainfall Rate Estimation Using Terahertz Channel Performance and Deep Neural Networks

AU - Chang, Wanzhu

AU - Song, Yuheng

AU - Liu, Wenbo

AU - Zhang, Mingxia

AU - Zhao, Jiabiao

AU - Li, Peian

AU - Ma, Jianjun

AU - Sun, Houjun

PY - 2025

Y1 - 2025

N2 - This work investigates rainfall intensity estimation using terahertz (THz) channel performance and deep neural networks. A THz time-domain spectroscopy system captured channel measurement data under controlled rain at multiple frequencies. Four architectures-CNN, RNN, LSTM, and BiGRU - were trained and evaluated. Results show that model performance varies with frequency, CNN performs well across most bands, while BiGRU excels at the mid-high frequency of 220 GHz.

AB - This work investigates rainfall intensity estimation using terahertz (THz) channel performance and deep neural networks. A THz time-domain spectroscopy system captured channel measurement data under controlled rain at multiple frequencies. Four architectures-CNN, RNN, LSTM, and BiGRU - were trained and evaluated. Results show that model performance varies with frequency, CNN performs well across most bands, while BiGRU excels at the mid-high frequency of 220 GHz.

UR - https://www.scopus.com/pages/publications/105033155935

U2 - 10.1109/UCMMT67044.2025.11287089

DO - 10.1109/UCMMT67044.2025.11287089

M3 - Conference article

AN - SCOPUS:105033155935

SN - 2639-4537

JO - UK, Europe, China Millimetre Waves and THZ Technology Workshop, UCMMT

JF - UK, Europe, China Millimetre Waves and THZ Technology Workshop, UCMMT

IS - 2025

T2 - 18th IEEE United Conference on Millimeter Waves and Terahertz Technologies, UCMMT 2025

Y2 - 25 August 2025 through 28 August 2025

ER -

Rainfall Rate Estimation Using Terahertz Channel Performance and Deep Neural Networks

Abstract

Access to Document

Other files and links

Fingerprint

Cite this