Browsing by Author "Breve, Murilo Montanini"
Now showing 1 - 7 of 7
Results Per Page
Sort Options
- An exploratory study on hindcasting with analogue ensembles of principal componentsPublication . Balsa, Carlos; Breve, Murilo Montanini; Rodrigues, Carlos Veiga; Costa, Luís S.; Rufino, JoséThe aim of this study is the reconstruction of meteorological data that are missing in a given station by means of the data from neighbouring stations. To achieve this, the Analogue Ensemble (AnEn) method was applied to the Principal Components (PCs) of the time series dataset, computed via Principal Component Analysis. This combination allows exploring the possibility of reducing the number of meteorological variables used in the reconstruction. The proposed technique is greatly influenced by the choice of the number of PCs used in the data reconstruction. The number of favorable PC varies according to the predicted variable and weather station. This choice is directly linked to the variables correlation. The application of AnEn using PCs leads to improvements of 8% to 21% in the RMSE of wind speed.
- Applications of the analog ensembles method to meteorological data reconstruction in the Northeast of PortugalPublication . Breve, Murilo Montanini; Rufino, José; Balsa, Carlos; Costa, LuísThe observation of weather states has always been a human need. Our most distant ancestors already tried to understand and predict the weather, but did not have reliable methods. In the 19th century, modern meteorology took its first steps: the French government, motivated by the sinking of ships near the coast of Crimea, because of a heavy rainstorm, created a network of 24 stations spread across Europe, which began to observe the weather. In recent years, due to computational advances, different methods of predicting weather states have begun to emerge, increasing the forecast extent and its accuracy. The Analog Ensembles method (AnEn), introduced by Luca Delle Monache in 2011 [1], is a post-processing tool that has shown good results to improve whether predictions or perform hindcasting (reconstruction of missing meteorological data). The goal of this study is to use the AnEn method to perform hindcasting, in order to reconstruct past weather conditions in a specific area of the northeeast of Portugal and verify its similarity with the actual forecast.
- Applications of the analog ensembles method to meteorological data reconstruction in the Northeast of PortugalPublication . Breve, Murilo Montanini; Rufino, José; Balsa, Carlos; Costa, LuísIn recent years, due to computational advances, different methods of predicting weather states have begun to emerge, increasing the forecast extent and its accuracy. The Analog Ensembles method (AnEn), introduced by Luca Delle Monache in 2011 [1], is a post-processing tool that has shown good results to improve whether predictions or perform hindcasting (reconstruction of missingmeteorological data).
- Enhancing weather data reconstruction through hybridmethods with dimensionality reductionPublication . Breve, Murilo Montanini; Balsa, Carlos; Rufino, José; Martins, FabricioAccurate weather analysis and forecasting rely on complete historical data. However, missing weather data often occurs due to sensor failures, data transmission issues, or limited monitoring capabilities. Reconstructing this missing data is crucial for reliableweather analysis. The Analog Ensemble (AnEn) method leverages past weather events and information from nearby stations to reconstruct and forecast data. However, incorporating nearby stations significantly increases computational costs, making the reconstruction process time consuming. To address this challenge, this dissertation integrates AnEn with dimension reduction techniques: Principal Component Analysis (PCA) and Partial Least Squares (PLS). Four hybrid methods—PCAnEn, PLSAnEn, PCClustAnEn, and PLSClustAnEn—are developed to enhance computational performance while maintaining or improving accuracy. Through four studies using three datasets, this research focuses on reconstructing six variables: wind-related variables, temperature, pressure, and humidity. The hybrid methods improved accuracy compared to the original AnEn. Notably, PLSAnEn achieves the highest reconstruction accuracy, while PLSR exhibits the fastest processing times. Additionally, PLSClustAnEn also proves to be a alternative for data reconstruction. The findings of this research contribute to the portfolio of strategies for addressing missing weather data.
- Optimal Latent Variables Number for the Reconstruction of Time Series with PLSRPublication . Balsa, Carlos; Dupuis, Hugo; Breve, Murilo Montanini; Guivarch, Ronan; Rufino, JoséThe Partial Least Squares Regression is an efficient method for the filling of gaps in meteorological time series. It enables to reduce the dimension of the predictor dataset to a reduced number of latent variables, without loss of significant information. Defining the number of latent variables to be used is an essential aspect of the success of the method. This study is about the comparison between eight different criteria, used in the choice of latent variables. The results indicate that the criteria based on cross-validation are the most efficient, being, however, more computationally demanding.
- PCAnEn - hindcasting with analogue ensembles of principal componentsPublication . Balsa, Carlos; Breve, Murilo Montanini; André, Baptiste; Rodrigues, Carlos Veiga; Rufino, JoséThe focus of this study is the reconstruction of missing meteorological data at a station based on data from neighboring stations. To that end, the Principal Components Analysis (PCA) method was applied to the Analogue Ensemble (AnEn) method to reduce the data dimensionality. The proposed technique is greatly influenced by the choice of stations according to proximity and correlation to the predicted one. PCA associated with AnEn decreased the errors in the prediction of some meteorological variables by 30% and, at the same time, decreased the prediction time by 48%. It was also verified that our implementation of this methodology in MATLAB is around two times faster than in R.
- Reconstruction of meteorological records by methods based on dimension reduction of the predictor datasetPublication . Balsa, Carlos; Breve, Murilo Montanini; Rodrigues, Carlos Veiga; Rufino, JoséThe reconstruction or prediction of meteorological records through the Analog Ensemble (AnEn) method is very efficient when the number of predictor time series is small. Thus, in order to take advantage of the richness and diversity of information contained in a large number of predictors, it is necessary to reduce their dimensions. This study presents methods to accomplish such reduction, allowing the use of a high number of predictor variables. In particular, the techniques of Principal Component Analysis (PCA) and Partial Least Squares (PLS) are used to reduce the dimension of the predictor dataset without loss of essential information. The combination of the AnEn and PLS techniques results in a very efficient hybrid method (PLSAnEn) for reconstructing or forecasting unstable meteorological variables, such as wind speed. This hybrid method is computationally demanding but its performance can be improved via parallelization or the introduction of variants in which all possible analogs are previously clustered. The multivariate linear regression methods used on the new variables resulting from the PCA or PLS techniques also proved to be efficient, especially for the prediction of meteorological variables without local oscillations, such as the pressure.