Predicting Plankton Production from Satellite Data. Oberholster, D., Oberholster, D., Curtis, R., Curtis, R., Fenn, R., & Fenn, R. Technical Report 2005. ![Predicting Plankton Production from Satellite Data [link]](https://bibbase.org/img/filetypes/link.svg "link")
Website abstract bibtex Estimates of plankton primary production are essential to understanding the functioning of the marine ecosystem and the possible impacts of climate change of the marine food web. Sub-surface chlorophyll is an excellent predictor plankton production, but collection of sub-surface chlorophyll data is slow. Surface data, however, can quickly be obtained via satellite. A method is therefore needed to predict sub-surface data using only surface information. Previous research in this field involved the use of self-organising maps (SOMS) to predict plankton-profiles. These SOMS are, however, hard to interpret and not very precise. The system proposed used Bayesian networks to predict sub-surface chlorophyll based on satellite data and other environmental factors. Bayesian Networks are comprised of two parts: a learning engine and inference engine. The learning engine finds patterns in historical data and the inference engine takes these patterns as input and predicts likely trends. An Investigation was undertaken to determine the use of topic maps for representing Bayesian network structure and beliefs. These topic maps needed to be visualised in an intuitive manner. A hyperbolic tree visualization was investigated as an alternative to static visualizations. The accuracy of predictions was limited by the use of Gaussian approximations to define the predicted profile, but the use of EM to create new profiles should give far better results in future. It was found that the topic maps provided a useful mechanism for passing the Bayesian network information between the inference engine and the interface. The hyperbolic visualisation of Bayesian networks was at least as easy to use as static representations.
@techreport{
title = {Predicting Plankton Production from Satellite Data},
type = {techreport},
year = {2005},
source = {System},
keywords = {Bayesian networks,hyperbolic trees,plankton prediction,topic maps},
pages = {1-11},
websites = {http://pubs.cs.uct.ac.za/archive/00000261/},
id = {c4c598d1-be4c-3da7-9be8-845a03ec424d},
created = {2015-04-22T22:28:22.000Z},
file_attached = {false},
profile_id = {95e10851-cdf3-31de-9f82-1ab629e601b0},
group_id = {b1fc9654-2f96-39de-a2ca-bc6069a08a77},
last_modified = {2017-03-14T13:43:36.285Z},
read = {false},
starred = {false},
authored = {false},
confirmed = {true},
hidden = {false},
private_publication = {false},
abstract = {Estimates of plankton primary production are essential to understanding the functioning of the marine ecosystem and the possible impacts of climate change of the marine food web. Sub-surface chlorophyll is an excellent predictor plankton production, but collection of sub-surface chlorophyll data is slow. Surface data, however, can quickly be obtained via satellite. A method is therefore needed to predict sub-surface data using only surface information. Previous research in this field involved the use of self-organising maps (SOMS) to predict plankton-profiles. These SOMS are, however, hard to interpret and not very precise. The system proposed used Bayesian networks to predict sub-surface chlorophyll based on satellite data and other environmental factors. Bayesian Networks are comprised of two parts: a learning engine and inference engine. The learning engine finds patterns in historical data and the inference engine takes these patterns as input and predicts likely trends. An Investigation was undertaken to determine the use of topic maps for representing Bayesian network structure and beliefs. These topic maps needed to be visualised in an intuitive manner. A hyperbolic tree visualization was investigated as an alternative to static visualizations. The accuracy of predictions was limited by the use of Gaussian approximations to define the predicted profile, but the use of EM to create new profiles should give far better results in future. It was found that the topic maps provided a useful mechanism for passing the Bayesian network information between the inference engine and the interface. The hyperbolic visualisation of Bayesian networks was at least as easy to use as static representations.},
bibtype = {techreport},
author = {Oberholster, D and Oberholster, D and Curtis, R.B. and Curtis, R.B. and Fenn, R.T. and Fenn, R.T.}
}
Downloads: 0
{"_id":"fRQmgcyd5PfkafZxJ","authorIDs":[],"author_short":["Oberholster, D.","Oberholster, D.","Curtis, R.","Curtis, R.","Fenn, R.","Fenn, R."],"bibbaseid":"oberholster-oberholster-curtis-curtis-fenn-fenn-predictingplanktonproductionfromsatellitedata-2005","bibdata":{"title":"Predicting Plankton Production from Satellite Data","type":"techreport","year":"2005","source":"System","keywords":"Bayesian networks,hyperbolic trees,plankton prediction,topic maps","pages":"1-11","websites":"http://pubs.cs.uct.ac.za/archive/00000261/","id":"c4c598d1-be4c-3da7-9be8-845a03ec424d","created":"2015-04-22T22:28:22.000Z","file_attached":false,"profile_id":"95e10851-cdf3-31de-9f82-1ab629e601b0","group_id":"b1fc9654-2f96-39de-a2ca-bc6069a08a77","last_modified":"2017-03-14T13:43:36.285Z","read":false,"starred":false,"authored":false,"confirmed":"true","hidden":false,"private_publication":false,"abstract":"Estimates of plankton primary production are essential to understanding the functioning of the marine ecosystem and the possible impacts of climate change of the marine food web. Sub-surface chlorophyll is an excellent predictor plankton production, but collection of sub-surface chlorophyll data is slow. Surface data, however, can quickly be obtained via satellite. A method is therefore needed to predict sub-surface data using only surface information. Previous research in this field involved the use of self-organising maps (SOMS) to predict plankton-profiles. These SOMS are, however, hard to interpret and not very precise. The system proposed used Bayesian networks to predict sub-surface chlorophyll based on satellite data and other environmental factors. Bayesian Networks are comprised of two parts: a learning engine and inference engine. The learning engine finds patterns in historical data and the inference engine takes these patterns as input and predicts likely trends. An Investigation was undertaken to determine the use of topic maps for representing Bayesian network structure and beliefs. These topic maps needed to be visualised in an intuitive manner. A hyperbolic tree visualization was investigated as an alternative to static visualizations. The accuracy of predictions was limited by the use of Gaussian approximations to define the predicted profile, but the use of EM to create new profiles should give far better results in future. It was found that the topic maps provided a useful mechanism for passing the Bayesian network information between the inference engine and the interface. The hyperbolic visualisation of Bayesian networks was at least as easy to use as static representations.","bibtype":"techreport","author":"Oberholster, D and Oberholster, D and Curtis, R.B. and Curtis, R.B. and Fenn, R.T. and Fenn, R.T.","bibtex":"@techreport{\n title = {Predicting Plankton Production from Satellite Data},\n type = {techreport},\n year = {2005},\n source = {System},\n keywords = {Bayesian networks,hyperbolic trees,plankton prediction,topic maps},\n pages = {1-11},\n websites = {http://pubs.cs.uct.ac.za/archive/00000261/},\n id = {c4c598d1-be4c-3da7-9be8-845a03ec424d},\n created = {2015-04-22T22:28:22.000Z},\n file_attached = {false},\n profile_id = {95e10851-cdf3-31de-9f82-1ab629e601b0},\n group_id = {b1fc9654-2f96-39de-a2ca-bc6069a08a77},\n last_modified = {2017-03-14T13:43:36.285Z},\n read = {false},\n starred = {false},\n authored = {false},\n confirmed = {true},\n hidden = {false},\n private_publication = {false},\n abstract = {Estimates of plankton primary production are essential to understanding the functioning of the marine ecosystem and the possible impacts of climate change of the marine food web. Sub-surface chlorophyll is an excellent predictor plankton production, but collection of sub-surface chlorophyll data is slow. Surface data, however, can quickly be obtained via satellite. A method is therefore needed to predict sub-surface data using only surface information. Previous research in this field involved the use of self-organising maps (SOMS) to predict plankton-profiles. These SOMS are, however, hard to interpret and not very precise. The system proposed used Bayesian networks to predict sub-surface chlorophyll based on satellite data and other environmental factors. Bayesian Networks are comprised of two parts: a learning engine and inference engine. The learning engine finds patterns in historical data and the inference engine takes these patterns as input and predicts likely trends. An Investigation was undertaken to determine the use of topic maps for representing Bayesian network structure and beliefs. These topic maps needed to be visualised in an intuitive manner. A hyperbolic tree visualization was investigated as an alternative to static visualizations. The accuracy of predictions was limited by the use of Gaussian approximations to define the predicted profile, but the use of EM to create new profiles should give far better results in future. It was found that the topic maps provided a useful mechanism for passing the Bayesian network information between the inference engine and the interface. The hyperbolic visualisation of Bayesian networks was at least as easy to use as static representations.},\n bibtype = {techreport},\n author = {Oberholster, D and Oberholster, D and Curtis, R.B. and Curtis, R.B. and Fenn, R.T. and Fenn, R.T.}\n}","author_short":["Oberholster, D.","Oberholster, D.","Curtis, R.","Curtis, R.","Fenn, R.","Fenn, R."],"urls":{"Website":"http://pubs.cs.uct.ac.za/archive/00000261/"},"biburl":"https://bibbase.org/service/mendeley/95e10851-cdf3-31de-9f82-1ab629e601b0","bibbaseid":"oberholster-oberholster-curtis-curtis-fenn-fenn-predictingplanktonproductionfromsatellitedata-2005","role":"author","keyword":["Bayesian networks","hyperbolic trees","plankton prediction","topic maps"],"metadata":{"authorlinks":{}}},"bibtype":"techreport","biburl":"https://bibbase.org/service/mendeley/95e10851-cdf3-31de-9f82-1ab629e601b0","creationDate":"2015-04-23T01:31:49.501Z","downloads":0,"keywords":["bayesian networks","hyperbolic trees","plankton prediction","topic maps"],"search_terms":["predicting","plankton","production","satellite","data","oberholster","oberholster","curtis","curtis","fenn","fenn"],"title":"Predicting Plankton Production from Satellite Data","year":2005,"dataSources":["8hwkQbNZz66Dc7LzM","ya2CyA73rpZseyrZ8"]}