Kalman Filters in crop models: old experiences in new contexts
Autor(a) principal: | |
---|---|
Data de Publicação: | 2024 |
Outros Autores: | , , |
Tipo de documento: | preprint |
Idioma: | eng |
Título da fonte: | SciELO Preprints |
Texto Completo: | https://preprints.scielo.org/index.php/scielo/preprint/view/8033 |
Resumo: | Data assimilation has been widely used for improvement of crop models’ estimates, for example to incorporate the effects of external events or compensate calibration errors in large areas. The term describes multiple approaches for those who want to take advantage of satellite imagery to reduce uncertainty or improve accuracy of model estimates. Kalman Filters are among the most used methods for achieving these goals. But their use in new contexts, i.e., from open field to protected environments, requires untangling aspects of the pipeline that are often performed in many different ways without guidelines, such as which variables to assimilate or how to ascribe uncertainty to observations or model estimates. This study is then divided in two parts. In the first, we review details on how uncertainty is ascribed on crop model estimates and in observations for applications of the Kalman Filter and three variations of the method, i.e., the Extended, Unscented and Ensemble, as well as which state variables are often updated and the frequency with which assimilation may occur, as well as how these aspects are connected to each other. In the second part, we apply different approaches from the reviewed literature in a greenhouse tomato crop model. We use artificial data with controlled noise levels as well as artificial data generated by simulation using other tomato crop model. We assess the impacts of using different methods and different approaches for ascribing uncertainty in model estimates and in observations, by assimilating artificial observations of fruit and of mature fruit biomass. We note that covariances should not be fixed values, that there are trade-offs between ascribing model uncertainty to the state itself and to other elements of the process, that observation covariance may have been considered disproportionality higher when using some ensemble generation approaches in the EnKF, and that bias in model estimates may lead to worse outcomes even when observations are high-quality ones. While we discussed aspects that should be considered in a new environment, many of them are also important for field crops, and we concluded assimilation should follow an assessment of which variables could be useful for assimilation. |
id |
SCI-1_17103a8a1f4cdb87c477e47df9343e72 |
---|---|
oai_identifier_str |
oai:ops.preprints.scielo.org:preprint/8033 |
network_acronym_str |
SCI-1 |
network_name_str |
SciELO Preprints |
repository_id_str |
|
spelling |
Kalman Filters in crop models: old experiences in new contextscrop modeldata assimilationprotected environmentsuncertaintystate estimationData assimilation has been widely used for improvement of crop models’ estimates, for example to incorporate the effects of external events or compensate calibration errors in large areas. The term describes multiple approaches for those who want to take advantage of satellite imagery to reduce uncertainty or improve accuracy of model estimates. Kalman Filters are among the most used methods for achieving these goals. But their use in new contexts, i.e., from open field to protected environments, requires untangling aspects of the pipeline that are often performed in many different ways without guidelines, such as which variables to assimilate or how to ascribe uncertainty to observations or model estimates. This study is then divided in two parts. In the first, we review details on how uncertainty is ascribed on crop model estimates and in observations for applications of the Kalman Filter and three variations of the method, i.e., the Extended, Unscented and Ensemble, as well as which state variables are often updated and the frequency with which assimilation may occur, as well as how these aspects are connected to each other. In the second part, we apply different approaches from the reviewed literature in a greenhouse tomato crop model. We use artificial data with controlled noise levels as well as artificial data generated by simulation using other tomato crop model. We assess the impacts of using different methods and different approaches for ascribing uncertainty in model estimates and in observations, by assimilating artificial observations of fruit and of mature fruit biomass. We note that covariances should not be fixed values, that there are trade-offs between ascribing model uncertainty to the state itself and to other elements of the process, that observation covariance may have been considered disproportionality higher when using some ensemble generation approaches in the EnKF, and that bias in model estimates may lead to worse outcomes even when observations are high-quality ones. While we discussed aspects that should be considered in a new environment, many of them are also important for field crops, and we concluded assimilation should follow an assessment of which variables could be useful for assimilation.SciELO PreprintsSciELO PreprintsSciELO Preprints2024-02-06info:eu-repo/semantics/preprintinfo:eu-repo/semantics/publishedVersionapplication/pdfhttps://preprints.scielo.org/index.php/scielo/preprint/view/803310.1590/SciELOPreprints.8033enghttps://preprints.scielo.org/index.php/scielo/article/view/8033/14999Copyright (c) 2024 Monique Pires Gravina de Oliveira, Thais Queiroz Zorzeto-Cesar, Romis Ribeiro de Faissol Attux, Luiz Henrique Antunes Rodrigueshttps://creativecommons.org/licenses/by/4.0info:eu-repo/semantics/openAccessOliveira, Monique Pires Gravina deZorzeto-Cesar, Thais QueirozAttux, Romis Ribeiro de FaissolRodrigues, Luiz Henrique Antunesreponame:SciELO Preprintsinstname:Scientific Electronic Library Online (SCIELO)instacron:SCI2024-02-04T15:43:05Zoai:ops.preprints.scielo.org:preprint/8033Servidor de preprintshttps://preprints.scielo.org/index.php/scieloONGhttps://preprints.scielo.org/index.php/scielo/oaiscielo.submission@scielo.orgopendoar:2024-02-04T15:43:05SciELO Preprints - Scientific Electronic Library Online (SCIELO)false |
dc.title.none.fl_str_mv |
Kalman Filters in crop models: old experiences in new contexts |
title |
Kalman Filters in crop models: old experiences in new contexts |
spellingShingle |
Kalman Filters in crop models: old experiences in new contexts Oliveira, Monique Pires Gravina de crop model data assimilation protected environments uncertainty state estimation |
title_short |
Kalman Filters in crop models: old experiences in new contexts |
title_full |
Kalman Filters in crop models: old experiences in new contexts |
title_fullStr |
Kalman Filters in crop models: old experiences in new contexts |
title_full_unstemmed |
Kalman Filters in crop models: old experiences in new contexts |
title_sort |
Kalman Filters in crop models: old experiences in new contexts |
author |
Oliveira, Monique Pires Gravina de |
author_facet |
Oliveira, Monique Pires Gravina de Zorzeto-Cesar, Thais Queiroz Attux, Romis Ribeiro de Faissol Rodrigues, Luiz Henrique Antunes |
author_role |
author |
author2 |
Zorzeto-Cesar, Thais Queiroz Attux, Romis Ribeiro de Faissol Rodrigues, Luiz Henrique Antunes |
author2_role |
author author author |
dc.contributor.author.fl_str_mv |
Oliveira, Monique Pires Gravina de Zorzeto-Cesar, Thais Queiroz Attux, Romis Ribeiro de Faissol Rodrigues, Luiz Henrique Antunes |
dc.subject.por.fl_str_mv |
crop model data assimilation protected environments uncertainty state estimation |
topic |
crop model data assimilation protected environments uncertainty state estimation |
description |
Data assimilation has been widely used for improvement of crop models’ estimates, for example to incorporate the effects of external events or compensate calibration errors in large areas. The term describes multiple approaches for those who want to take advantage of satellite imagery to reduce uncertainty or improve accuracy of model estimates. Kalman Filters are among the most used methods for achieving these goals. But their use in new contexts, i.e., from open field to protected environments, requires untangling aspects of the pipeline that are often performed in many different ways without guidelines, such as which variables to assimilate or how to ascribe uncertainty to observations or model estimates. This study is then divided in two parts. In the first, we review details on how uncertainty is ascribed on crop model estimates and in observations for applications of the Kalman Filter and three variations of the method, i.e., the Extended, Unscented and Ensemble, as well as which state variables are often updated and the frequency with which assimilation may occur, as well as how these aspects are connected to each other. In the second part, we apply different approaches from the reviewed literature in a greenhouse tomato crop model. We use artificial data with controlled noise levels as well as artificial data generated by simulation using other tomato crop model. We assess the impacts of using different methods and different approaches for ascribing uncertainty in model estimates and in observations, by assimilating artificial observations of fruit and of mature fruit biomass. We note that covariances should not be fixed values, that there are trade-offs between ascribing model uncertainty to the state itself and to other elements of the process, that observation covariance may have been considered disproportionality higher when using some ensemble generation approaches in the EnKF, and that bias in model estimates may lead to worse outcomes even when observations are high-quality ones. While we discussed aspects that should be considered in a new environment, many of them are also important for field crops, and we concluded assimilation should follow an assessment of which variables could be useful for assimilation. |
publishDate |
2024 |
dc.date.none.fl_str_mv |
2024-02-06 |
dc.type.driver.fl_str_mv |
info:eu-repo/semantics/preprint info:eu-repo/semantics/publishedVersion |
format |
preprint |
status_str |
publishedVersion |
dc.identifier.uri.fl_str_mv |
https://preprints.scielo.org/index.php/scielo/preprint/view/8033 10.1590/SciELOPreprints.8033 |
url |
https://preprints.scielo.org/index.php/scielo/preprint/view/8033 |
identifier_str_mv |
10.1590/SciELOPreprints.8033 |
dc.language.iso.fl_str_mv |
eng |
language |
eng |
dc.relation.none.fl_str_mv |
https://preprints.scielo.org/index.php/scielo/article/view/8033/14999 |
dc.rights.driver.fl_str_mv |
https://creativecommons.org/licenses/by/4.0 info:eu-repo/semantics/openAccess |
rights_invalid_str_mv |
https://creativecommons.org/licenses/by/4.0 |
eu_rights_str_mv |
openAccess |
dc.format.none.fl_str_mv |
application/pdf |
dc.publisher.none.fl_str_mv |
SciELO Preprints SciELO Preprints SciELO Preprints |
publisher.none.fl_str_mv |
SciELO Preprints SciELO Preprints SciELO Preprints |
dc.source.none.fl_str_mv |
reponame:SciELO Preprints instname:Scientific Electronic Library Online (SCIELO) instacron:SCI |
instname_str |
Scientific Electronic Library Online (SCIELO) |
instacron_str |
SCI |
institution |
SCI |
reponame_str |
SciELO Preprints |
collection |
SciELO Preprints |
repository.name.fl_str_mv |
SciELO Preprints - Scientific Electronic Library Online (SCIELO) |
repository.mail.fl_str_mv |
scielo.submission@scielo.org |
_version_ |
1797047814858997760 |