The current climate change panorama and the impacts that it has on water resources have increased the need to study the hydrologic system and analyze its dynamics, intending to better understand the changes expected on it and to underline the water resources management needed. Given the complexity involved in the parameterization and prediction from groundwater transport models, an alternative data-driven modeling has been implemented in the present study. For an aquifer located in the northern Italy region, among the several factors that may interact in this natural process, two variables have been selected to perform the dynamic’s system analysis, i.e. hydraulic head and rainfall levels. Sets of data of both are available for around ten years, making them suitable for the application of a time series analysis after the proper transformations required. An ARDL model allows introducing the effects of the external variable, rainfall levels, on the system’s behavior, evaluated through changes in the hydraulic head levels. For the forecasting purpose of the phreatic levels, future values of precipitation have been necessary in advanced. To this aim, an ARIMA model has been initially estimated letting to predict the future values subsequently introduce in the forecasting by the ARDL model. R and EViews® software have been used to estimate the models based on statistical tools and codes already developed. Both models have been successfully estimated and the final forecasting of hydraulic head values has been obtained. Even the uncertainty that the estimated values may attain, good performance of the model gives an expected trend and allows driving relevant conclusions about the dynamic of the aquifer studied and emphasizes the strong interaction among the two variables in the area considered.
L'attuale panorama del cambiamento climatico e gli impatti che ha sulle risorse idriche aumentano la necessità di studiare i sistemi idrologici e di analizzare le proprie dinamiche, intese a comprendere meglio i cambiamenti attesi e sottolineare la necessaria gestione delle risorse idriche. Data la complessità che riguarda la parametrizzazione e la predizione dei modelli di trasporto delle acque sotterranee, nel presente studio è stata implementata una alternativa basata sui modellazione dei dati. Per un acquifero situato nella regione settentrionale dell'Italia, tra i diversi fattori che possono interagire in questo processo naturale, sono state selezionate due variabili per eseguire l'analisi della dinamica del sistema, cioè i livelli idraulici e le precipitazioni. I set di dati di entrambi sono disponibili per circa dieci anni, rendendoli adatti per l'applicazione di un'analisi della serie temporale dopo le adeguate trasformazioni richieste. Un modello ARDL consente di introdurre gli effetti della variabile esterna, i livelli di pioggia, sul comportamento del sistema valutato attraverso i cambiamenti nei livelli della testa idraulica. Per lo scopo di previsione dei livelli freatici, i valori futuri delle precipitazioni sono stati necessari in avanti. A questo scopo, è stato inizialmente stimato un modello ARIMA che permettere di prevedere i valori futuri introdotti successivamente nella previsione dal modello ARDL. R e EViews® sono stati utilizzati per stimare i modelli basati su strumenti statistici e codici già sviluppati. Entrambi i modelli sono stati accuratamente valutati e si è ottenuta la previsione finale dei valori delle teste idrauliche. Anche l'incertezza che i valori stimati possono raggiungere, le buone prestazioni del modello danno una tendenza attesa e permettono di trarre conclusioni pertinenti sulla dinamica dell'acquifero studiato e sottolinea la forte interazione tra le due variabili nella zona considerata.
Study of groundwater flow by means of ARIMA and ARDL models for the analysis of the relationship between hydraulic head and rainfall time series
ADARME MEJIA, YULLY PAOLA
2016/2017
Abstract
The current climate change panorama and the impacts that it has on water resources have increased the need to study the hydrologic system and analyze its dynamics, intending to better understand the changes expected on it and to underline the water resources management needed. Given the complexity involved in the parameterization and prediction from groundwater transport models, an alternative data-driven modeling has been implemented in the present study. For an aquifer located in the northern Italy region, among the several factors that may interact in this natural process, two variables have been selected to perform the dynamic’s system analysis, i.e. hydraulic head and rainfall levels. Sets of data of both are available for around ten years, making them suitable for the application of a time series analysis after the proper transformations required. An ARDL model allows introducing the effects of the external variable, rainfall levels, on the system’s behavior, evaluated through changes in the hydraulic head levels. For the forecasting purpose of the phreatic levels, future values of precipitation have been necessary in advanced. To this aim, an ARIMA model has been initially estimated letting to predict the future values subsequently introduce in the forecasting by the ARDL model. R and EViews® software have been used to estimate the models based on statistical tools and codes already developed. Both models have been successfully estimated and the final forecasting of hydraulic head values has been obtained. Even the uncertainty that the estimated values may attain, good performance of the model gives an expected trend and allows driving relevant conclusions about the dynamic of the aquifer studied and emphasizes the strong interaction among the two variables in the area considered.File | Dimensione | Formato | |
---|---|---|---|
2017_07_Adarme.pdf
accessibile in internet per tutti
Descrizione: Thesis text
Dimensione
54.46 MB
Formato
Adobe PDF
|
54.46 MB | Adobe PDF | Visualizza/Apri |
I documenti in POLITesi sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/10589/135282