Modelos de cambio de régimen de Markov aplicados a temperatura del agua de ríos en el macizo Cerro de la Muerte, Costa Rica
Date
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
El proyecto de investigación Global Observatory Network for freshwater Biodiversity in High Mountain Streams (GLOBIOS) inició en Costa Rica en el 2021, con el objetivo de determinar posibles cambios futuros en la biodiversidad de los ríos de alta montaña debido al cambio climático. Para ello se tomaron diversas variables, entre ellas la temperatura del agua del río medida mediante un sensor colocado en seis sitios, que registraba la temperatura cada 15 minutos durante un año. Sin embargo, debido al aumento del caudal en época lluviosa en algunos sitios, el sensor estuvo fuera del agua registrando la temperatura del aire en vez de la del agua. Se planteó el uso de modelos de cambio de régimen de Markov discretos, que permitieron detectar los cambios de régimen y así se logró identificar los períodos de tiempo en los que el sensor registró la temperatura del aire. Para construir los modelos, se definieron dos regímenes, Estuvo expuesto al aire y No estuvo expuesto al aire. Además, se transformó la temperatura del agua en la amplitud estandarizada de la temperatura, debido a que esta variable proporciona información más clara sobre los períodos en los que el sensor estuvo fuera del agua. También se utilizaron datos satelitales de la NASA (POWER Project) como covariables para alimentar los modelos de cada sitio. Para cada uno de los sitios, se ajustaron 3 modelos: modelos de cambio de régimen autorregresivos (AR(1)), modelos con covariables, y modelos con covariables y AR(1). Se concluyó que no existe un mejor modelo para todos los sitios. Se seleccionó el mejor modelo de acuerdo con el AIC y BIC, y se realizó imputación de los valores de amplitud de temperatura diaria de acuerdo con el mejor modelo en cada sitio.
The research project "Global Observatory Network for freshwater Biodiversity in High Mountain Streams (GLOBIOS)" started in Costa Rica in 2021, aiming to determine potential future changes in the biodiversity of high mountain rivers due to climate change. Various variables were measured, including the river water temperature, which was recorded by a sensor placed at six sites, taking readings every 15 minutes for one year. However, due to increased flow during the rainy season at some of the sites, the sensor emerged from the water and recorded air temperature instead of water temperature. The use of discrete Markov regime-switching models was proposed, which allowed for the detection of regime changes, identifying the periods when the sensor recorded air temperature. To build the models, two regimes were defined for the models: Exposed to air and Not exposed to air. Additionally, the water temperature was transformed into the standardized amplitude of the temperature, as this variable provided clearer information about the periods when the sensor was out of the water. NASA’s satellite data (POWER Project) was also used as covariates to feed the models for each site. For each site, three models were fitted: autoregressive regime-switching models (AR(1)), models with covariates, and models with covariates and AR(1). It was concluded that there is no single best model for all sites. The best model was selected according to the AIC and BIC, and imputation of the daily temperature amplitude values was performed according to the best model at each site.
The research project "Global Observatory Network for freshwater Biodiversity in High Mountain Streams (GLOBIOS)" started in Costa Rica in 2021, aiming to determine potential future changes in the biodiversity of high mountain rivers due to climate change. Various variables were measured, including the river water temperature, which was recorded by a sensor placed at six sites, taking readings every 15 minutes for one year. However, due to increased flow during the rainy season at some of the sites, the sensor emerged from the water and recorded air temperature instead of water temperature. The use of discrete Markov regime-switching models was proposed, which allowed for the detection of regime changes, identifying the periods when the sensor recorded air temperature. To build the models, two regimes were defined for the models: Exposed to air and Not exposed to air. Additionally, the water temperature was transformed into the standardized amplitude of the temperature, as this variable provided clearer information about the periods when the sensor was out of the water. NASA’s satellite data (POWER Project) was also used as covariates to feed the models for each site. For each site, three models were fitted: autoregressive regime-switching models (AR(1)), models with covariates, and models with covariates and AR(1). It was concluded that there is no single best model for all sites. The best model was selected according to the AIC and BIC, and imputation of the daily temperature amplitude values was performed according to the best model at each site.
Description
Keywords
temperatura del agua, ríos de alta montaña, modelos de Markov, montaña, GLOBIOS, water temperature, high mountain rivers, Markov models, mountain
Citation
item.page.doi
Collections
Endorsement
Review
Supplemented By
Referenced By
Creative Commons license
Except where otherwised noted, this item's license is described as Attribution-NonCommercial-NoDerivs 3.0 United States