Groundwater Level Prediction/Forecasting and Assessment of Uncertainty Using SGS and ARIMA Models: A Case Study in the Bauru Aquifer System (Brazil)

• Autor(a) principal: Takafuji, Eduardo Henrique de Moraes
• Data de Publicação: 2018
• Outros Autores: Rocha, Marcelo Monteiro da, Manzione, Rodrigo Lilla [UNESP]
• Tipo de documento: Artigo
• Idioma: eng
• Título da fonte: Repositório Institucional da UNESP
• Texto Completo: http://dx.doi.org/10.1007/s11053-018-9403-6; http://hdl.handle.net/11449/171373
• topic: Bauru aquifer system; Ecological Station of Santa Barbara (Brazil); Geostatistics; Groundwater monitoring; Time series

Best water management practices should involve the prediction of the availability of groundwater resources. To predict/forecast and consequently manage these water resources, two known methods are discussed: a time series method using the autoregressive integrated moving average (ARIMA) and a geostatistical method using sequential Gaussian simulation (SGS). This study was conducted in the Ecological Station of Santa Barbara (EEcSB), located at the Bauru Aquifer System domain, a substantial water source for the countryside of São Paulo State, Brazil. The relevance of this study lies in the fact that the 2013/2014 hydrological year was one of the driest periods ever recorded in São Paulo State, which was directly reflected in the groundwater table level behavior. A hydroclimatological network comprising 49 wells was set up to monitor the groundwater table depths at EEcSB to capture this response. The traditional time series has the advantage that it has been created to forecast and the disadvantage that an interpolation method must also be used to generate a spatially distributed map. On the other hand, a geostatistical approach can generate a map directly. To properly compare the results, both methods were used to predict/forecast the groundwater table levels at the next four measured times at the wells’ locations. The errors show that SGS achieves a slightly higher level of accuracy and considered anomalous events (e.g., severe drought). Meanwhile, the ARIMA models are considered better for monitoring the aquifer because they achieved the same accuracy level as SGS in the 2-month forecast and a higher precision at all periods and can be optimized automatically by using the Akaike information criterion.