Estimativa de recarga de águas subterrâneas por sensoriamento remoto e computação em nuvem
Carregando...
Data
2020-09-11
Autores
Magnoni, Pedro Henrique Jandreice
Título da Revista
ISSN da Revista
Título de Volume
Editor
Universidade Estadual Paulista (Unesp)
Resumo
A popularização do sensoriamento remoto via computação em nuvem em plataformas como o Google Earth Engine (GEE) trouxe novas perspectivas aos modeladores, facilitando acesso e processamento de dados espacializados em diferentes localidades. No entanto, estas ferramentas ainda são pouco exploradas no processo de planejamento e tomada de decisões sobre recursos hídricos, especialmente tratando-se das águas subterrâneas. O Sistema Aquífero Guaraní (SAG) é o maior reservatório transfronteiriço de água subterrânea da América do Sul, mas a recarga em suas zonas de afloramento é uma das variáveis hidrológicas menos conhecidas. Portanto, o objetivo deste trabalho foi contribuir para a compreensão do processo de recarga das águas subterrâneas em uma área de afloramento do SAG. Avaliou-se a adequação do uso de dados de sensoriamento remoto na equação do balanço hídrico para estimar a recarga (GWR) e o impacto nos níveis freáticos (WTD), por meio da utilização do GEE. Um modelo de estimativa da WTD foi adaptado a partir do método de flutuação do nível freático (WTF). Séries temporais estimadas de precipitação, evapotranspiração e escoamento superficial de três bases de dados hidroclimáticos baseados em sensoriamento remoto (FLDAS, GLDAS e TERRACLIMATE) foram obtidas ao longo de 2014 a 2017, no GEE na localização da Estação Ecológica de Angatuba (EEcA), no Estado de São Paulo. As estimativas da GWR e da WTD foram calculadas na localização de oito poços de monitoramento do nível freático nas imediações da EEcA. Uma análise do viés das estimativas de precipitação foi realizada usando dados meteorológicos observados. Dois cenários de estimativa da GWR foram avaliados, bem como o efeito de diferentes faixas de rendimento específico (Sy) adotados no modelo. A GWR no cenário 1 (C1) variou de 10,8% a 19,69% da precipitação observada para o FLDAS, de 18,65% a 38,09% no GLDAS e de 0,36% a 6,11% no TERRACLIMATE, enquanto que no cenário 2 (C2) as variações foram de 0,9 a 13%, de 3,76% a 27,36% e de 0,13% 3,25%, respectivamente. Quanto as estimativas da WTD, o FLDAS (base com melhor resultado nas métricas de validação do modelo) obteve a Raíz do Erro Médio Quadrático (RMSE) com variação de 0,36 a 1,12 m, e melhores resultados em poços de profundidade média até 15 m.
The advent, improvement and popularization of remote sensing by cloud computing such as Google Earth Engine (GEE) has brought new perspectives to modelers, allowing access to reliable and representative data over vast areas. However, several tools are still under-explored and actually used in the water resources planning and decision-making process. The Guarani Aquifer System (GAS) is the largest transboundary groundwater reservoir in South America, yet recharge in the GAS outcrop zones is one of the least known hydrological variables. This work aimed to contribute to understanding of the groundwater recharge process in an outcrop area of the SAG. Through GEE, the use of remote sensing data in the water-budget equation was evaluated to recharge estimation (GWR) and groundwater levels modelling (WTD). The WTD model was adapted from the water table fluctuation method (WTF). Estimated series of precipitation, evapotranspiration and runoff from three hydroclimatic remote sensing based databases (FLDAS, GLDAS and TERRACLIMATE) were obtained from GEE at the location of the Ecological Station of Angatuba (EEcA), in State of São Paulo, from 2014 to 2017. GWR and WTD were obtained at the location of eight monitoring wells near to EEcA. Bias analysis of precipitation estimates was performed using observed meteorological data. Two GWR scenarios were evaluated, as well as the effect of different ranges of specific yield (Sy) adopted. The GWR in scenario 1 (C1) ranged from 10.8% to 19.69% of the observed rainfall for FLDAS, from 18.65% to 38.09% for GLDAS and from 0.36% to 6.11% for TERRACLIMATE, while in scenario 2 (C2), the results ranged from 0.9 to 13%, from 3.76% to 27.36% and 0.13% 3.25%, respectively. For WTD modelling, using FLDAS data (database with better validation results) the Root Mean Square Error (RMSE) ranged from 0.36 to 1.12 m, with better agreement in wells of medium water depth up to 15 m.
The advent, improvement and popularization of remote sensing by cloud computing such as Google Earth Engine (GEE) has brought new perspectives to modelers, allowing access to reliable and representative data over vast areas. However, several tools are still under-explored and actually used in the water resources planning and decision-making process. The Guarani Aquifer System (GAS) is the largest transboundary groundwater reservoir in South America, yet recharge in the GAS outcrop zones is one of the least known hydrological variables. This work aimed to contribute to understanding of the groundwater recharge process in an outcrop area of the SAG. Through GEE, the use of remote sensing data in the water-budget equation was evaluated to recharge estimation (GWR) and groundwater levels modelling (WTD). The WTD model was adapted from the water table fluctuation method (WTF). Estimated series of precipitation, evapotranspiration and runoff from three hydroclimatic remote sensing based databases (FLDAS, GLDAS and TERRACLIMATE) were obtained from GEE at the location of the Ecological Station of Angatuba (EEcA), in State of São Paulo, from 2014 to 2017. GWR and WTD were obtained at the location of eight monitoring wells near to EEcA. Bias analysis of precipitation estimates was performed using observed meteorological data. Two GWR scenarios were evaluated, as well as the effect of different ranges of specific yield (Sy) adopted. The GWR in scenario 1 (C1) ranged from 10.8% to 19.69% of the observed rainfall for FLDAS, from 18.65% to 38.09% for GLDAS and from 0.36% to 6.11% for TERRACLIMATE, while in scenario 2 (C2), the results ranged from 0.9 to 13%, from 3.76% to 27.36% and 0.13% 3.25%, respectively. For WTD modelling, using FLDAS data (database with better validation results) the Root Mean Square Error (RMSE) ranged from 0.36 to 1.12 m, with better agreement in wells of medium water depth up to 15 m.
Descrição
Palavras-chave
FLDAS, GLDAS, TERRACLIMATE, Águas subterrâneas, Balanço hídrico, Sensoriamento remoto, Groundwater Recharge, Remote sensing, Waterbudget, Google Earth Engine