Precise forecasting of reference evapotranspiration (ET₀) is one of the critical initial steps in determining crop water requirements, which contributes to the reliable management and long-term planning of the world’s scarce water sources. This study provides daily prediction and multi-step forward forecasting of ET₀ utilizing a long short-term memory network (LSTM) and a bi-directional LSTM (Bi-LSTM) model. For daily predictions, the LSTM model’s accuracy was compared to that of other artificial intelligence-based models commonly used in ET0 forecasting, including support vector regression (SVR), M5 model tree (M5Tree), multivariate adaptive regression spline (MARS), probabilistic linear regression (PLR), adaptive neuro-fuzzy inference system (ANFIS), and Gaussian process regression (GPR). The LSTM model outperformed the other models in a comparison based on Shannon’s entropy-based decision theory, while the M5 tree and PLR models proved to be the lowest performers. Prior to performing a multi-step-ahead forecasting, ANFIS, sequence-to-sequence regression LSTM network (SSR-LSTM), LSTM, and Bi-LSTM approaches were used for one-step-ahead forecasting utilizing the past values of the ET₀ time series. The results showed that the Bi-LSTM model outperformed other models and that the sequence of models in ascending order in terms of accuracies was Bi-LSTM > SSR-LSTM > ANFIS > LSTM. The Bi-LSTM model provided multi-step (5 day)-ahead ET₀ forecasting in the next step. According to the results, the Bi-LSTM provided reasonably accurate and acceptable forecasting of multi-step-forward ET₀ with relatively lower levels of forecasting errors. In the final step, the generalization capability of the proposed best models (LSTM for daily predictions and Bi-LSTM for multi-step-ahead forecasting) was evaluated on new unseen data obtained from a test station, Ishurdi. The model’s performance was assessed on three distinct datasets (the entire dataset and the first and the second halves of the entire dataset) derived from the test dataset between 1 January 2015 and 31 December 2020. The results indicated that the deep learning techniques (LSTM and Bi-LSTM) achieved equally good performances as the training station dataset, for which the models were developed. The research outcomes demonstrated the ability of the developed deep learning models to generalize the prediction capabilities outside the training station.

This study evaluated the rainfall historical simulations of 15 Global Climate Models (GCMs) of the Coupled Model Intercomparison Project phase 6 (CMIP6) in replicating annual and seasonal rainfall climatology, their temporal variability and trends in Bangladesh for the period 1979–2014, considering ERA5 (ECMWF Reanalysis 5th Generation) reanalysis as the reference dataset. Shannon's Entropy decision‐analysis was employed for GCMs' rating based on eight statistical indicators and a comprehensive rating metric for the final grading of the GCMs. The majority of the CMIP6 GCMs accurately reproduced the spatial feature of ERA5 rainfall. However, the GCMs underestimated annual rainfall by an average of 190.5 mm, with the highest underestimation in monsoon (131.76 mm) and least in winter (3.52 mm) seasons. Most GCMs also underestimated rainfall variability for all seasons except winter. Besides, the GCMs showed an increasing trend in pre‐monsoon and a decreasing trend in post‐monsoon rainfall like ERA5, but an opposite (negative) to ERA5 trend (positive) in monsoon season rainfall. The ensemble mean of the GCMs showed higher skill in reconstructing rainfall climatology, temporal variability and trends than the individual GCMs. The study identified MPI‐ESM1‐2‐LR, MPI‐ESM1‐2‐HR, and GFDL‐ESM4 as the most effective GCMs in reproducing precipitation over Bangladesh. The selected models' simulation can be used for climate change impact assessment in Bangladesh after bias minimization.

Application of multivariate adaptive regression spline ensembles (En-MARS) in a coupled simulation-optimization methodology to derive multiple-objective optimal groundwater extraction strategies for a multilayered coastal aquifer system is demonstrated. Two conflicting objectives of groundwater extraction strategies are solved using a controlled elitist multiobjective genetic algorithm. A three-dimensional density-dependent coupled flow and salt-transport numerical simulation model is used to generate the training patterns of groundwater extraction strategies and resulting saltwater concentrations. Prediction capability of En-MARS is compared with that of the best multivariate adaptive regression spline (MARS) model in the ensemble. En-MARS is then linked externally within the optimization algorithm to develop the management model. The optimal solutions obtained from the En-MARS models are verified by running the numerical simulation model. The results indicate that MARS-based ensemble modeling approach is able to provide reliable solutions for a multilayered coastal aquifer management problem. The adaptive nature of MARS models and use of ensembles and parallel processing results in a computationally efficient, accurate, and reliable methodology for coastal aquifer management that also incorporates uncertainties in modeling.