Deep learning models for groundwater level prediction based on delay penalty
In irrigation agriculture, predicting groundwater level (GWL) using deep learning models can help decision-makers coordinate surface water and groundwater usage, thus aiding in the sustainable development and utilization of groundwater. However, when making a long sequence prediction, prediction seq...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
IWA Publishing
2024-02-01
|
| Series: | Water Supply |
| Subjects: | |
| Online Access: | http://ws.iwaponline.com/content/24/2/555 |
| _version_ | 1797200670286151680 |
|---|---|
| author | Zhang Chenjia Tianxin Xu Yan Zhang Daokun Ma |
| author_facet | Zhang Chenjia Tianxin Xu Yan Zhang Daokun Ma |
| author_sort | Zhang Chenjia |
| collection | DOAJ |
| description | In irrigation agriculture, predicting groundwater level (GWL) using deep learning models can help decision-makers coordinate surface water and groundwater usage, thus aiding in the sustainable development and utilization of groundwater. However, when making a long sequence prediction, prediction sequences often have severe delays affecting the availability of prediction results. In this paper, a new loss function is proposed to minimize the lag and oversmoothing on the prediction of GWLs. GWL, meteorology, and pumping data are collected via an irrigation Internet of Things system in Hutubi County, Xinjiang. Through Pearson's correlation analysis, historical potential evapotranspiration (ET0), groundwater extraction, and GWL were chosen to predict GWLs. Datasets were constructed through the proposed spatiotemporal data fusion method; then, the best model from the six deep learning models was selected by comparing the prediction capability of the datasets. Finally, the mean-squared error (MSE) loss function is replaced by the proposed loss function. Compared to the mean absolute error, MSE, and predicted sequence graphs, the new loss function significantly depresses the time delay with similar prediction accuracy.
HIGHLIGHTS
An integrated loss function considering the Euclidean error, time warping, and time delay is proposed to depress the time-delay problem of the predicted sequences.;
A novel data fusion approach was introduced by incorporating the groundwater extraction data from motive pumping wells into the prediction of groundwater levels.; |
| first_indexed | 2024-04-24T07:35:20Z |
| format | Article |
| id | doaj.art-97aa9f323a964f08b78b237e346bcdf3 |
| institution | Directory Open Access Journal |
| issn | 1606-9749 1607-0798 |
| language | English |
| last_indexed | 2024-04-24T07:35:20Z |
| publishDate | 2024-02-01 |
| publisher | IWA Publishing |
| record_format | Article |
| series | Water Supply |
| spelling | doaj.art-97aa9f323a964f08b78b237e346bcdf32024-04-20T06:39:25ZengIWA PublishingWater Supply1606-97491607-07982024-02-0124255556710.2166/ws.2024.009009Deep learning models for groundwater level prediction based on delay penaltyZhang Chenjia0Tianxin Xu1Yan Zhang2Daokun Ma3 College of Information and Electrical Engineering, China Agricultural University Haidian District, Beijing 100089, China College of Information and Electrical Engineering, China Agricultural University Haidian District, Beijing 100089, China School of Yi Language and Culture, Xichang University, Xichang, Sichaun 831000, China College of Information and Electrical Engineering, China Agricultural University Haidian District, Beijing 100089, China In irrigation agriculture, predicting groundwater level (GWL) using deep learning models can help decision-makers coordinate surface water and groundwater usage, thus aiding in the sustainable development and utilization of groundwater. However, when making a long sequence prediction, prediction sequences often have severe delays affecting the availability of prediction results. In this paper, a new loss function is proposed to minimize the lag and oversmoothing on the prediction of GWLs. GWL, meteorology, and pumping data are collected via an irrigation Internet of Things system in Hutubi County, Xinjiang. Through Pearson's correlation analysis, historical potential evapotranspiration (ET0), groundwater extraction, and GWL were chosen to predict GWLs. Datasets were constructed through the proposed spatiotemporal data fusion method; then, the best model from the six deep learning models was selected by comparing the prediction capability of the datasets. Finally, the mean-squared error (MSE) loss function is replaced by the proposed loss function. Compared to the mean absolute error, MSE, and predicted sequence graphs, the new loss function significantly depresses the time delay with similar prediction accuracy. HIGHLIGHTS An integrated loss function considering the Euclidean error, time warping, and time delay is proposed to depress the time-delay problem of the predicted sequences.; A novel data fusion approach was introduced by incorporating the groundwater extraction data from motive pumping wells into the prediction of groundwater levels.;http://ws.iwaponline.com/content/24/2/555data fusiondeep learninggroundwater level predictionloss function |
| spellingShingle | Zhang Chenjia Tianxin Xu Yan Zhang Daokun Ma Deep learning models for groundwater level prediction based on delay penalty Water Supply data fusion deep learning groundwater level prediction loss function |
| title | Deep learning models for groundwater level prediction based on delay penalty |
| title_full | Deep learning models for groundwater level prediction based on delay penalty |
| title_fullStr | Deep learning models for groundwater level prediction based on delay penalty |
| title_full_unstemmed | Deep learning models for groundwater level prediction based on delay penalty |
| title_short | Deep learning models for groundwater level prediction based on delay penalty |
| title_sort | deep learning models for groundwater level prediction based on delay penalty |
| topic | data fusion deep learning groundwater level prediction loss function |
| url | http://ws.iwaponline.com/content/24/2/555 |
| work_keys_str_mv | AT zhangchenjia deeplearningmodelsforgroundwaterlevelpredictionbasedondelaypenalty AT tianxinxu deeplearningmodelsforgroundwaterlevelpredictionbasedondelaypenalty AT yanzhang deeplearningmodelsforgroundwaterlevelpredictionbasedondelaypenalty AT daokunma deeplearningmodelsforgroundwaterlevelpredictionbasedondelaypenalty |