Improved calibrations of TDR measurements using artificial neural networks
(2001) p.128-139- Abstract
- Time domain reflectometry (TDR) has been proven to take accurate readings of the apparent dielectric
constant Ka and bulk electrical conductivity σa. These measurements have been used for estimations of
water content (θ) and concentration of different chemicals. The concentration of an ionic pollutant or
tracer can be estimated from the soil solution electrical conductivity σw. Several different calibration
equations have been used to describe the Ka-θ and σa-σw-θ relationships. In the present study, artificial
neural networks (ANNs) were used for calculations of θ and σw from TDR measurements in pure sand.
The results showed that the ANN performed consistently better than several other models, suggesting
the... (More) - Time domain reflectometry (TDR) has been proven to take accurate readings of the apparent dielectric
constant Ka and bulk electrical conductivity σa. These measurements have been used for estimations of
water content (θ) and concentration of different chemicals. The concentration of an ionic pollutant or
tracer can be estimated from the soil solution electrical conductivity σw. Several different calibration
equations have been used to describe the Ka-θ and σa-σw-θ relationships. In the present study, artificial
neural networks (ANNs) were used for calculations of θ and σw from TDR measurements in pure sand.
The results showed that the ANN performed consistently better than several other models, suggesting
the suitability of ANNs for accurate TDR calibrations. The ANN was also used for predicting the Ka-θ
relationship using soil physical parameters for ten different soil types. The Ka-θ relationship for each
soil type was predicted using the other nine for calibration. It was shown that the ANN predictions were
comparable to a soil specific calibration. Thus, by using ANN, highly accurate data can be obtained
without the need for elaborate soil specific calibration experiment (Less)
Please use this url to cite or link to this publication:
https://lup.lub.lu.se/record/ba899aeb-3c45-4224-9ceb-67bc630a74a0
- author
- Persson, Magnus LU
- organization
- publishing date
- 2001-08-01
- type
- Chapter in Book/Report/Conference proceeding
- publication status
- published
- subject
- host publication
- Second International Symposium and Workshop on Time Domain Reflectometry for Innovative Geotechnical Applications
- editor
- Dowding, Charles
- pages
- 12 pages
- publisher
- Infrastructure Technology Institute
- ISBN
- 0-9712631-0-8
- language
- English
- LU publication?
- yes
- id
- ba899aeb-3c45-4224-9ceb-67bc630a74a0
- alternative location
- http://www.civil.northwestern.edu/people/dowding/tdr/publications/TDR-CP-2001.pdf
- date added to LUP
- 2023-10-05 13:41:26
- date last changed
- 2023-10-16 11:23:54
@inproceedings{ba899aeb-3c45-4224-9ceb-67bc630a74a0, abstract = {{Time domain reflectometry (TDR) has been proven to take accurate readings of the apparent dielectric<br/>constant Ka and bulk electrical conductivity σa. These measurements have been used for estimations of<br/>water content (θ) and concentration of different chemicals. The concentration of an ionic pollutant or<br/>tracer can be estimated from the soil solution electrical conductivity σw. Several different calibration<br/>equations have been used to describe the Ka-θ and σa-σw-θ relationships. In the present study, artificial<br/>neural networks (ANNs) were used for calculations of θ and σw from TDR measurements in pure sand.<br/>The results showed that the ANN performed consistently better than several other models, suggesting<br/>the suitability of ANNs for accurate TDR calibrations. The ANN was also used for predicting the Ka-θ<br/>relationship using soil physical parameters for ten different soil types. The Ka-θ relationship for each<br/>soil type was predicted using the other nine for calibration. It was shown that the ANN predictions were<br/>comparable to a soil specific calibration. Thus, by using ANN, highly accurate data can be obtained<br/>without the need for elaborate soil specific calibration experiment}}, author = {{Persson, Magnus}}, booktitle = {{Second International Symposium and Workshop on Time Domain Reflectometry for Innovative Geotechnical Applications}}, editor = {{Dowding, Charles}}, isbn = {{0-9712631-0-8}}, language = {{eng}}, month = {{08}}, pages = {{128--139}}, publisher = {{Infrastructure Technology Institute}}, title = {{Improved calibrations of TDR measurements using artificial neural networks}}, url = {{http://www.civil.northwestern.edu/people/dowding/tdr/publications/TDR-CP-2001.pdf}}, year = {{2001}}, }