- You are here:
-
-
All publications
- Groundwater level prediction using artificial intelligence based data driven modelling
All publications from your selection
Groundwater level prediction using artificial intelligence based data driven modelling
Authors: Harsh Tusharkumar Parikh
Identifier: CSBE22266
Download file:
Published in: CSBE-SCGAB Technical Conferences » AGM Charlottetown 2022
Download RAW file:
DOI:
Description: Groundwater is a major source of freshwater, which fulfills the households, agricultural domestic and industrial water needs. It is very important and challenging to accurately estimate the groundwater levels (GWLs) without physical wells. This study employed novel machine learning based data driven modeling techniques including Xgboost, bidirectional long short term memory (BiLSTM), gated recurrent unit (GRU), and prophet to estimate GWLs for five different watersheds of Prince Edward Island (PEI), Canada. These five watersheds were selected among three counties of PEI to capture the diverse ecosystem of the study area. Selected input variables were the stream level, stream velocity, precipitation, relative humidity, mean temperature, heat degree days, and dew point temperature. The variable importance was determined using information theory-based feature engineering method. Daily groundwater level data (2015-2019) of five monitoring wells was retrieved from the Land and Environment, PEI. Weather data for this period was collected from a local weather station in Charlottetown. Data was split in training period (2015-2017) and testing period (2018-2019) to assess the machine learning model accuracy with statistical evaluations including coefficient of determination (R2), root mean square error (RMSE) and Nash–Sutcliffe model efficiency coefficient (NSE). The results suggested that the machine learning models were able to accurately predict the GWLs of all watersheds in PEI. The performance of the Prophet model remained better in comparison with other machine learning models with lower RMSE and higher NSE and R2 scores. The detailed results of this study will be presented in CSBE conference.
Keywords: Ground water levels, Machine learning, XGboost, Data driven modeling, Hydrology, statistical modeling.
Résumé:
Mots-clés:
Citation:
Volume:
Issue:
Pages -
Contributor:
Date: 2022-07-24
Technical field:
Conference name: CSBE/SCGAB 2022 Annual Conference, Charlottetow, PEI, 24-27 July 2022.
Session name: Poster Sessions
Other information:
Type: Text.Abstract
Format:
Publication type: Poster
Source:
Relation:
Coverage: India
Language 1: en
Language 2:
Rights: Canadian Society for Bioengineering
Notes:
Other files:
DOI:
Description: Groundwater is a major source of freshwater, which fulfills the households, agricultural domestic and industrial water needs. It is very important and challenging to accurately estimate the groundwater levels (GWLs) without physical wells. This study employed novel machine learning based data driven modeling techniques including Xgboost, bidirectional long short term memory (BiLSTM), gated recurrent unit (GRU), and prophet to estimate GWLs for five different watersheds of Prince Edward Island (PEI), Canada. These five watersheds were selected among three counties of PEI to capture the diverse ecosystem of the study area. Selected input variables were the stream level, stream velocity, precipitation, relative humidity, mean temperature, heat degree days, and dew point temperature. The variable importance was determined using information theory-based feature engineering method. Daily groundwater level data (2015-2019) of five monitoring wells was retrieved from the Land and Environment, PEI. Weather data for this period was collected from a local weather station in Charlottetown. Data was split in training period (2015-2017) and testing period (2018-2019) to assess the machine learning model accuracy with statistical evaluations including coefficient of determination (R2), root mean square error (RMSE) and Nash–Sutcliffe model efficiency coefficient (NSE). The results suggested that the machine learning models were able to accurately predict the GWLs of all watersheds in PEI. The performance of the Prophet model remained better in comparison with other machine learning models with lower RMSE and higher NSE and R2 scores. The detailed results of this study will be presented in CSBE conference.
Keywords: Ground water levels, Machine learning, XGboost, Data driven modeling, Hydrology, statistical modeling.
Résumé:
Mots-clés:
Citation:
Volume:
Issue:
Pages -
Contributor:
Date: 2022-07-24
Technical field:
Conference name: CSBE/SCGAB 2022 Annual Conference, Charlottetow, PEI, 24-27 July 2022.
Session name: Poster Sessions
Other information:
Type: Text.Abstract
Format:
Publication type: Poster
Source:
Relation:
Coverage: India
Language 1: en
Language 2:
Rights: Canadian Society for Bioengineering
Notes:
Other files:
©
2024
CSBE-SCGAB