Using GAN for Imputation of Missing Recorded Data to Improve Groundwater Level Prediction Based on Deep Learning Methods
- Department of Hydraulics and Ocean Engineering, National Cheng Kung University, Tainan City, Taiwan (cindy23322332@gmail.com)
The development of civilization and the preservation of environmental ecosystems are strongly dependent on water resources. Typically, the insufficient supply of surface water resources for domestic, industrial, and agriculture needs is often supplemented by the ground water resources. However, the groundwater is a natural resource that must be accumulated over many years and cannot be recovered after a short period of recharge. Therefore, the long-term management of groundwater resources is an important issue for the sustainable development. The accurate prediction of groundwater levels is the first step to evaluate the total water resources and its allocation.
However, in the process of data collection, data may be missing due to various factors. Thus, retracting the missing data is a main problem which any research field must deal with. It has been well known that to maintain the data integrity, one of the effective approaches is to choose missing value imputation (MVI) for tackling the problem. In addition, it has been demonstrated that the method of the machine learning may be a better tool. Therefore, the main purpose of this study is to utilize a generative adversarial network (GAN) that consists of a generative model and a discriminative model for imputation. Our result shows that GAN can improve the accuracy of water resource evaluations.
In the current study, two interdisciplinary deep learning methods, Univariate and Seq2val, are used for groundwater level estimation. In addition to addressing the significance of the parameter conditions, the advantages and disadvantages of these two models in hydrological simulations are also discussed and compared. Finally, Seq2seq is employed to examine the limit of the models in long-term water level simulations. Our result suggests that the interdisciplinary deep learning approach may be beneficial for providing a better evaluation of water resources.
Keywords: GAN，CNN，LSTM，Imputation，Groundwater prediction
How to cite: Chen, H. Y., Lo, W.-C., and Huang, C.-T.: Using GAN for Imputation of Missing Recorded Data to Improve Groundwater Level Prediction Based on Deep Learning Methods, EGU General Assembly 2023, Vienna, Austria, 24–28 Apr 2023, EGU23-10611, https://doi.org/10.5194/egusphere-egu23-10611, 2023.
