Rainfall nowcasting is a challenging task due to the time-dependencies of the variables and the stochastic behavior of the process. The difficulty increases when the geographic area of interest is characterized by a large spatio-temporal variability of its meteorological variables, causing large variations of rainfall even within a small area in places such as the Tropical Andes. To address this problem, we propose a methodology for building a group of models based on Long Short-Term Memory (LSTM) neural networks using Bayesian optimization. We optimize the model hyperparameters using accumulated experience to reduce the hyperparameter search space over successive iterations. The result is a large reduction in modeling time that allows the building of specialized LSTM models for each zone and nowcasting time. We evaluated the method by nowcasting rain events in the urban area of Cuenca City in Ecuador, a city with large spatio-temporal variability. The results show that our proposed model offers better performance over the trivial forecaster for up to 9 hours of future forecasts with an accuracy of up to 84.4%
The leachate discharges generated in sanitary landfills contain many pollutants that are harmful to the environment; treatments are scarce and should be carried out better. The use of coagulation-flocculation processes has been one of the most widely used, but that; Due to the complexity of the characterization of the leachate, the dosing strategy of coagulants and biopolymers needs to be clarified. Therefore, the present study is carried out to determine the doses of coagulants and biopolymers suitable for coagulation-flocculation processes in the treatment of leachates by using computational models of machine learning techniques such as Artificial Neural Networks (ANN); these allow to decrease the operations of the tests of jars in the laboratory; optimizing resources. Through laboratory experimentation, there are real results of the effectiveness of applying biopolymers in leachate treatments at different concentration levels. The laboratory results were taken as input variables for the algorithms used; after the validation and calibration process, we proceeded to estimate predicted data with the computational model, obtaining predictions of optimal doses for treatment with high statistical adjustment indicators. It is verified that the applied coagulation-flocculation treatments reduce the turbidity values in the leachate and contaminants associated with suspended solids. In this way, the jar tests are optimized so that the operational costs decrease without affecting the results of adequate dosing.
La matriz energética del Ecuador supera el 75% de energía renovable, sin embargo, la fuente principal es la energía hidroeléctrica. Como primer paso hacia la obtención de una matriz más amigable con el ambiente y más diversa es fundamental identificar zonas con potencial para la instalación de energías renovables no convencionales, los Sistemas de Información Geográfico (SIG) son de gran ayuda para identificar zonas con este potencial. En este estudio se identifican los posibles sitios para la implantación de centrales solares fotovoltaicas (CSF) en la Provincia del Azuay mediante los SIG y la Evaluación Multicriterio (EM). Para valorar la importancia de los criterios se empleó el Método Analítico Jerárquico (AHP). Para la obtención de un modelo de capacidad de acogida se integró el modelo de aptitud en el que se analizaron criterios económicos y técnicos; y un modelo de impacto que analizo variables ambientales. Al integrar los modelos esta metodología permitirá la identificación de zonas para el emplazamiento de estaciones que permitan el monitoreo de los recursos y el análisis del comportamiento previo a la implementación de las CSF. Una vez ejecutada la metodología propuesta se obtiene como resultado dos posibles sitios con características medias para el emplazamiento de CSF. Como conclusión en base a los indicadores analizados el Azuay no cuenta con una zona potencialmente adecuada para la instalación de esta tecnología.
Rainfall nowcasting is a challenging task due to the time-dependencies of the variables and the stochastic behavior of the process. The difficulty increases when the geographic area of interest is characterized by a large spatio-temporal variability of its meteorological variables, causing large variations of rainfall even within a small area in places such as the Tropical Andes. To address this problem, we propose a methodology for A Bayesian framework for rainfall nowcasting building a group of models based on Long Short-Term Memory (LSTM) neural networks using Bayesian optimization. We optimize the model hyperparameters using accumulated experience to reduce the hyperparameter search space over successive iterations. The result is a large reduction in modeling time that allows the building of specialized LSTM models for each zone and nowcasting time. We evaluated the method by nowcasting rain events in the urban area of Cuenca City in Ecuador, a city with large spatio-temporal variability. The results show that our proposed model offers better performance over the trivial forecaster for up to 9 hours of future forecasts with an accuracy of up to 84.4%.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.
hi@scite.ai
10624 S. Eastern Ave., Ste. A-614
Henderson, NV 89052, USA
Copyright © 2024 scite LLC. All rights reserved.
Made with 💙 for researchers
Part of the Research Solutions Family.