Input parameters selection for soil moisture retrieval using an artificial neural network
dc.contributor.author | Chai, Soo See | |
dc.contributor.author | Veenendaal, Bert | |
dc.contributor.author | West, Geoffrey | |
dc.contributor.author | Walker, J. | |
dc.contributor.editor | Ostendorf, B. | |
dc.contributor.editor | Baldock, P. | |
dc.contributor.editor | Bruce, D. | |
dc.contributor.editor | Burdett, M. | |
dc.contributor.editor | P. Corcoran | |
dc.date.accessioned | 2017-01-30T14:24:07Z | |
dc.date.available | 2017-01-30T14:24:07Z | |
dc.date.created | 2010-03-14T20:02:24Z | |
dc.date.issued | 2009 | |
dc.identifier.citation | Chai, Soo-See and Veenendaal, Bert and West, Geoff and Walker, Jeffrey P. 2009. Input parameters selection for soil moisture retrieval using an artificial neural network, in Ostendorf B., Baldock, P., Bruce, D., Burdett, M. and Corcoran, P. (ed), Surveying & Spatial Sciences Institute Biennial International Conference 2009, Sep 28 2009, pp. 1181-1193. Adelaide, Australia. | |
dc.identifier.uri | http://hdl.handle.net/20.500.11937/38643 | |
dc.description.abstract |
Factors other than soil moisture which influence the intensity of microwave emission from the soil include surface temperature, surface roughness, vegetation cover and soil texture which make this a non-linear and ill-posed problem. Artificial Neural Networks (ANNs) have been demonstrated to be good solutions to this type of problem. Since an ANN is a data driven model, proper input selection is a crucial step in its implementation as the presence of redundant or unnecessary inputs can severely impair the ability of the network to learn the target patterns. In this paper, the input parameters are chosen in combination with the brightness temperatures and are based on the use of incremental contributions of the variables towards soil moisture retrieval. Field experiment data obtained during the National Airborne Field Experiment 2005 (NAFE'05) are used. The retrieval accuracy with the input parameters selected is compared with the use of only brightness temperature as input and the use of brightness temperature in conjunction with a range of available parameters. Note that this research does not aim at selecting the best features for all ANN soil moisture retrieval problems using passive microwave. The paper shows that, depending on the problem and the nature of the data, some of the data available are redundant as the input of ANN for soil moisture retrieval. Importantly the results show that with the appropriate choice of inputs, the soil moisture retrieval accuracy of ANN can be significantly improved. | |
dc.publisher | Unknown | |
dc.title | Input parameters selection for soil moisture retrieval using an artificial neural network | |
dc.type | Conference Paper | |
dcterms.source.startPage | 1181 | |
dcterms.source.endPage | 1193 | |
dcterms.source.title | Surveying & Spatial Sciences Institute Biennial International Conference | |
dcterms.source.series | Surveying & Spatial Sciences Institute Biennial International Conference | |
dcterms.source.isbn | 9780958136686 | |
dcterms.source.conference | Surveying & Spatial Sciences Institute Biennial International Conference 2009 | |
dcterms.source.conference-start-date | Sep 28 2009 | |
dcterms.source.conferencelocation | Adelaide, Australia | |
dcterms.source.place | Unknown | |
curtin.note |
Published by Surveying & Spatial Sciences Institute (SSSI) | |
curtin.accessStatus | Open access | |
curtin.faculty | Department of Spatial Sciences | |
curtin.faculty | Faculty of Science and Engineering | |
curtin.faculty | WA School of Mines |