TY - JOUR
T1 - A novel machine learning application
T2 - Water quality resilience prediction Model
AU - Imani, Maryam
AU - Hasan, Md Mahmudul
AU - Bittencourt, Luiz Fernando
AU - McClymont, Kent
AU - Kapelan, Zoran
N1 - Accepted Author Manuscript
PY - 2021
Y1 - 2021
N2 - Resilience-informed water quality management embraces the growing environmental challenges and provides greater accuracy by unpacking the systems' characteristics in response to failure conditions in order to identify more effective opportunities for intervention. Assessing the resilience of water quality requires complex analysis of influential parameters which can be challenging, time consuming and costly to compute. It may also require building detailed conceptual and/or physically process-based models that are difficult to build, calibrate and validate. This study utilises Artificial Neural Network (ANN) to develop a novel application to predict water quality resilience to simplify resilience evaluation. The Fuzzy Analytic Hierarchy Process method is used to rank water basins based on their level of resilience and to identify the ones that demand prompt restoration strategies. The commonly used ‘magnitude * duration of being in failure state’ quantification method has been used to formulate and evaluate resilience. A 17-years long water quality dataset from the 22 water basins in the State of São Paulo, Brazil, was used to train and test the ANN model. The overall agreement between the measured and simulated WQI resilience values is satisfactory and hence, can be used by planners and decision makers for improved water management. Moreover, comparative analyses show similarities and differences between the ‘level of criticalities’ reported in each zone by Environment Agency of the state of São Paulo (CETESB) and by the resilience model in this study.
AB - Resilience-informed water quality management embraces the growing environmental challenges and provides greater accuracy by unpacking the systems' characteristics in response to failure conditions in order to identify more effective opportunities for intervention. Assessing the resilience of water quality requires complex analysis of influential parameters which can be challenging, time consuming and costly to compute. It may also require building detailed conceptual and/or physically process-based models that are difficult to build, calibrate and validate. This study utilises Artificial Neural Network (ANN) to develop a novel application to predict water quality resilience to simplify resilience evaluation. The Fuzzy Analytic Hierarchy Process method is used to rank water basins based on their level of resilience and to identify the ones that demand prompt restoration strategies. The commonly used ‘magnitude * duration of being in failure state’ quantification method has been used to formulate and evaluate resilience. A 17-years long water quality dataset from the 22 water basins in the State of São Paulo, Brazil, was used to train and test the ANN model. The overall agreement between the measured and simulated WQI resilience values is satisfactory and hence, can be used by planners and decision makers for improved water management. Moreover, comparative analyses show similarities and differences between the ‘level of criticalities’ reported in each zone by Environment Agency of the state of São Paulo (CETESB) and by the resilience model in this study.
KW - Analytic hierarchy process
KW - Artificial neural network
KW - Fuzzy logic
KW - Machine learning
KW - Resilience
KW - Triangular fuzzy number
KW - Water quality
UR - http://www.scopus.com/inward/record.url?scp=85099456457&partnerID=8YFLogxK
U2 - 10.1016/j.scitotenv.2020.144459
DO - 10.1016/j.scitotenv.2020.144459
M3 - Article
AN - SCOPUS:85099456457
SN - 0048-9697
VL - 768
JO - Science of the Total Environment
JF - Science of the Total Environment
M1 - 144459
ER -