loading page

Stochastic and Artificial Intelligence Models for Climate Change Investigation and Groundwater Level Assessment of Gaza Coastal Aquifer (Palestine)
  • +4
  • Hassan Al-Najjar,
  • Gokmen Ceribasi,
  • Emrah Dogan,
  • Khalid Qahman,
  • Mazen Abualtayef,
  • Ahmed Shaqfa,
  • Iyad Ahmet
Hassan Al-Najjar
IUG
Author Profile
Gokmen Ceribasi
Sakarya Universitesi
Author Profile
Emrah Dogan
Sakarya Universitesi
Author Profile
Khalid Qahman
IUG
Author Profile
Mazen Abualtayef
IUG
Author Profile
Ahmed Shaqfa
IUG
Author Profile
Iyad Ahmet
Sakarya Universitesi
Author Profile

Abstract

The Gaza coastal aquifer is a critical resource for the supply of water to the Gaza Strip and continues to be depleted as a result of the effects of climate change and the anthropogenic activities. Therefore, this study tends to investigate the impact of climate change and groundwater withdrawal practices on the oscillation of the Gaza Coastal Aquifer water table level by recruiting the power of the stochastic time-series models in exemplifying the autoregression of data and by leveraging the efficiency of the artificial neural networks (ANNs) in expressing the nonlinear regression between the different meteorological and hydrological factors. The climate stochastic models reveal that the Gaza Strip region will face a decline in the precipitation by -5.2% and an increase in the temperature by +1˚C in the timeframe of 2020-2040. The potential evaporation and the sunshine period will 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 stochastic models developed for the groundwater abstraction time series show that the groundwater pumping processes would increase by about 55 % by 2040, compared to the 124 million cubic meters of groundwater that was withdrawn in 2020. The stochastic model of structure (2,1,5) (4,1,2)12 was defined to extend the time series of the groundwater level up to 2040. In order to form an integrated stochastic-ANN model, the combination of the time series of climate factors, groundwater abstraction and groundwater level were emerged into a one hidden layer ANN of 20-neurons. The performance of the model was high in term of training and in forecasting the future where the correlation coefficient (r) = 0.95-0.99 and the root mean square error (RMSE) = 0.09-0.21.