Modeling response of runoff and evapotranspiration for predicting water table depth in arid region using dynamic recurrent neural network

Ghose, Dillip K.; Das, Umesh Kumar; Roy, Parthajit

doi:10.1016/j.gsd.2018.01.007

Cited by 39 publications

(16 citation statements)

References 15 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We set the learning rate LR = 0.001, and the batch size is set at 256. Similar to the Encoder-Decoder model, we set the length of the encoder to [2,4,6,8,10,12], and then we find the optimal settings of the parameters by testing different parameters. For ARIMA, MLP, and SVM, we use the previous 6-day data as input, and for deep learning methods including DBN, LSTM, and CA-LSTM, we use the default settings in their paper [2], [22], [23].…”

Section: E Details On Training Proceduresmentioning

confidence: 99%

“…Other commonly used deep learning models include Recurrent Neural Networks (RNN) and Long Short-Term Memory (LSTM), which have been applied to many different tasks in the hydrological field. For example, RNN has been used to predict groundwater levels [12], wavelet-coupled RNN has be used to simulate rainfall-runoff [13], and LSTM-based water table depth prediction has been used in agricultural areas [14].…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Stream-Flow Forecasting Based on Dynamic Spatio-Temporal Attention

2019

View full text Add to dashboard Cite

Recent advances in stream-flow prediction using Recurrent Neural Networks (RNN) or Long Short-Term Memory (LSTM) have prompted their exploration in the field of hydrology. Predicting hydrological time series is of great importance, but is very challenging as it is influenced by many complex factors such as dynamic spatio-temporal feature correlation and nonlinear meteorological-hydrological. In this context, we propose a framework based on dynamic spatio-temporal attention (DSTA) to predict the stream-flow of a hydrological station over several days. It consists of three main modules: a spatial module, a temporal module, and a trend input module. Experiments were carried out on real data sets from four watersheds in the mainstream of the Yangtze River and proved that our method is superior to the six baseline methods.INDEX TERMS Spatio-temporal attention, spatio-temporal data mining, stream-flow predicting, LSTM.

show abstract

Section: E Details On Training Proceduresmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Stream-Flow Forecasting Based on Dynamic Spatio-Temporal Attention

2019

View full text Add to dashboard Cite

show abstract

“…The hydrological implications of the dynamics of water availability in reservoirs are often assessed according to the principles of inputoutput equilibrium [Araujo et al, 2006;Pandey et al, 2011;Fowe et al, 2015;Ali et al, 2017], or the concept of stock flow [Alifujiang et al, 2017]. Some of the mathematical models used to explain the dynamics of water volume growth in reservoirs are the water balance models [Bonacci and Roje, 2008 Ghose et al, 2018], the soil and water assessment tool (SWAT) model [Desta andLemma, 2017: Hallauz et al, 2018;Anand and Oinam, 2019], or the Muskingum equation [PJRRC, 2015]. The intrinsic water volume growth rate as an ecological parameter [Cortes, 2016] is not studied in these models.…”

Section: Introductionmentioning

confidence: 99%

Ecological Implications of the Dynamics of Water Volume Growth in a Reservoir

Kartono¹,

Purwanto²,

Suripin³

2021

Ecol. Eng. Environ. Technol.

View full text Add to dashboard Cite

The rate of growth of the water volume in the reservoir varies with each charging season. The accuracy of the predictions is required in sustainable reservoir management. Its intrinsic growth rate as an ecological parameter plays a role in determining this speed. This study aimed to analyze the dynamics of water volume growth based on its intrinsic growth rate to assess the potential for hydrometeorology disasters. The population growth models proposed to be tested for suitability and goodness is the Verhulst, Richards, Comperzt, and modified Malthus model. Test suitability and model goodness were subjected to stages of verification, parameter estimation and model validation based on daily water volume data in the Gembong Reservoir, Pati, Indonesia for the period 2007-2020. A good model is determined based on the Mean Average Percentage Error (MAPE) criteria. The Richards model with b = 2 and r = 0.063/day had consistently low MAPE values during training and testing. This model was chosen as a new approach to understand the dynamics of water volume growth in a reservoir. The ecological implication of these dynamics of water volume growth is that reservoirs experience an abundance of water during the charging season. Reservoir normalization can be prioritized as a mitigation strategy for potential flood disasters.

show abstract

“…The stochastic ARIMA models are widely used in water resources management applications, especially for modeling hydrological stream flows, groundwater level fluctuations, and drought patterns (Myronidis et al 2018;Takafuji et al 2018;Sakizadeh et al 2019;. Moreover, the ability of AI in hydrology and water resources management and for groundwater level modeling has been examined by many studies (Rakhshandehroo et al 2012;Ghose et al 2018;Kouziokas et al 2018;Guzman et al 2019;Lee et al 2019;Tang et al 2019). In particular, this study aims to simulate the fluctuations in the groundwater level of the Gaza coastal aquifer in light of the climate change consequences.…”

Section: Introductionmentioning

confidence: 99%

Statistical modeling of spatial and temporal vulnerability of groundwater level in the Gaza Strip (Palestine)

et al. 2021

View full text Add to dashboard Cite

The water supply in the Gaza Strip substantially depends on the groundwater resource of the Gaza coastal aquifer. The climate changes and the over-exploiting processes negatively impact the recovery of the groundwater balance. The climate variability is characterized by the decline in the precipitation of −5.2% and an increase in temperature of +1 °C in the timeframe of 2020–2040. The potential evaporation and the sunshine period are expected to increase by about 111 mm and 5 hours, respectively, during the next 20 years. However, the atmosphere is predicted to be drier where the relative humidity will fall by a trend of −8% in 20 years. The groundwater abstraction is predicted to increase by 55% by 2040. The response of the groundwater level to climate change and groundwater pumping was evaluated using a model of a 20-neuron ANN with a performance of the correlation coefficient (r)=0.95–0.99 and the root mean square error (RMSE)=0.09–0.21. Nowadays, the model reveals that the groundwater level ranges between −0.38 and −18.5 m and by 2040 it is expected to reach −1.13 and −28 m below MSL at the northern and southern governorates of the Gaza Strip, respectively.

show abstract

Modeling response of runoff and evapotranspiration for predicting water table depth in arid region using dynamic recurrent neural network

Cited by 39 publications

References 15 publications

Stream-Flow Forecasting Based on Dynamic Spatio-Temporal Attention

Stream-Flow Forecasting Based on Dynamic Spatio-Temporal Attention

Ecological Implications of the Dynamics of Water Volume Growth in a Reservoir

Statistical modeling of spatial and temporal vulnerability of groundwater level in the Gaza Strip (Palestine)

Contact Info

Product

Resources

About