Mehdi Jamei

Water temperature plays a pivotal role in shaping riverine ecosystems, exerting significant influence on a range of water quality parameters. However, accurately forecasting multi-temporal daily data remains challenging due to the non-stationary and nonlinear characteristics of hydrological time series. To address these challenges, the study proposes a novel deep learning framework that integrates Recursive Feature Elimination (RFE) with Multivariate Variational Mode Decomposition (MVMD), a multi-channel decomposition scheme that extracts meaningful sub-signals across multiple correlated variables. The decomposed features are processed using an Elman neural network integrated with a Bidirectional Gated Recurrent Unit (BIGRU) to capture both bidirectional and feedback-driven temporal dynamics. The model is applied to Fanno Creek and the McKenzie River in the western United States, demonstrating superior predictive performance under dynamically evolving hydrological conditions. RFE selected key input variables (discharge, pH, specific conductance, dissolved oxygen) from five years of data (2017-2021). The MVMD multi-channel scheme decomposes input lags into sub-sequences for each forecast horizon. The primary model (MVMD-ELMAN-BIGRU) was validated using elastic net (ELNET) regression and a Convolutional neural network coupled with BIGRU (CNN-BIGRU) as comparative machine learning (ML) models. Evaluation facilities include statistical indices, vulnerability assessments, and diagnostic visualizations. Additionally, for a reasonable evaluation of the models, a novel Multi-criteria decision-making (MCDM) method, namely the Multi-Objective Optimisation method based on Ratio Analysis (MOORA), was adopted to consolidate the metric performance across scenarios. The results indicated that MVMD-ELMAN-BIGRU, regarding the least value of MOORA (T+1:0.1096; T+3: 0.0096; T+7: 0.0478) for the Fanno Creek Rivers (T+1:0.00; T+3: 0.0296; T+7: 0.0932) for the McKenzie River, was superior to the MVMD-CNN-BIGRU and MVMD-ELNET models, respectively. This approach presents a promising solution for multi-temporal water temperature forecasting, which is crucial for effectively managing river ecosystems and water resources.

Agricultural drought occurs when soil moisture available to crops is inadequate to meet their water requirements, resulting in reduced crop yields and agricultural production. Therefore, it is crucial to accurately monitor agricultural drought for effective irrigation planning and scheduling, water resources management and allocation, and food security protection. Soil moisture is a fundamental variable for estimating agricultural drought indicators; thus, reliable, accurate, continuous, and long-term datasets are essential for effective drought monitoring. L-band radiometry is the most effective passive microwave remote sensing technique for estimating global soil moisture. The ESA Soil Moisture and Ocean Salinity (SMOS) and NASA Soil Moisture Active Passive (SMAP) missions, launched in 2009 and 2015, respectively, were developed specifically to retrieve Surface Soil Moisture (SSM) at L-band (1.4 GHz) for the top 5 cm of soil, with a target accuracy of 0.04 m³ m⁻³. These missions also generate Root Zone Soil Moisture (RZSM) at 0–100 cm depth through assimilation of SSM into land surface models. Agricultural drought can be detected using the Soil Water Deficit Index (SWDI) and Soil Moisture Drought Index (SMDI) derived from SMAP and SMOS satellite-based soil moisture products. The Karkheh River Basin is one of the most important watersheds and agricultural regions in southwestern Iran, with a semi-arid to arid climate, which has experienced frequent droughts in recent years. This study evaluates the efficiency of SMOS and SMAP products for agricultural drought monitoring using the SMDI and SWDI indices over the Karkheh River Basin.

Accurate and continuous root zone soil moisture (RZSM) data are crucial for important applications such as water resources management, flood and drought monitoring, irrigation planning and timing, agricultural productivity, climate change adaptation, and climate modeling. Direct measurement of RZSM using in-situ sensors can be time-consuming and costly, particularly over large areas. The Soil Moisture and Ocean Salinity (SMOS) and Soil Moisture Active Passive (SMAP) satellites provide surface soil moisture (SSM) data at L-Band and on a global scale. Due to the relationship between SSM and RZSM, satellite SSM data can be assimilated into land surface models to estimate RZSM data. The SMAP Level-4 Surface and Root Zone Soil Moisture (SMAP RZSM) product includes RZSM (0–100 cm), 3-hourly, 9 km on EASE-Grid 2; and the SMOS Level 4 RZSM CATDS (SMOS RZSM) product provides daily RZSM (0–100 cm), 25 km spatial sampling on EASE-Grid 2 on a global scale. This study aimed to evaluate the accuracy and efficiency of RZSM data derived from SMAP RZSM and SMOS RZSM products compared to in-situ measurements collected from IRIMO (Islamic Republic of Iran Meteorological Organization) sites from 2017 to 2020 in different regions of Iran. The evaluation results indicate that the SMOS RZSM data has higher accuracy compared to the SMAP RZSM data. The SMOS RZSM data has a root mean square error (RMSE) ranging from 0.03 to 0.09 m3m−3 and an unbiased root mean square error (ubRMSE) ranging from 0.03 to 0.05 m3m−3. The SMAP RZSM data has an RMSE range of 0.03 to 0.21 m3m−3 and an ubRMSE range of 0.02 to 0.08 m3m−3.

The rainfall pattern plays a vital role in agriculture and overall climate resilience in the Atlantic provinces of Canada. Prince Edward Island and New Brunswick, two Atlantic provinces, have great potential to grow potatoes, grains, and blueberries. Thus, accurate forecasting of the rainfall pattern helps farmers determine the optimal time for planting, irrigation, fertilisation, and harvesting based on predicted rainfall patterns. Here, a new complementary multi-level intelligent framework comprised of the recursive feature elimination (RFE), Empirical Fourier Decomposition (EFD), and deep ensemble random vector functional link (Deep RVFL) has been developed to forecast the monthly (one month ahead) rainfall pattern in Charlottetown and Fredericton stations. Aiming for this, first, the significant antecedent information (lag sequences) was indicated using the RFE scheme. Then, all the optimal lags were decomposed using the EDF scheme to deduce the complexities of rainfall sub-component signals before feeding the Deep RVFL algorithm. Two comparative deep learning models, namely, RVFL and CNN-LSTM, were incorporated with the implemented multi-level pre-processing scheme in hybrid and standalone forms. In order to validate the models, several statistical indices, such as correlation coefficient (R), root mean square error (RMSE), and Kling-Gupta efficiency (KGE), scatter plots, signal trend analysis, and diagnostic assessment, were utilized. The outcomes of the results ascertained that the RFE-EDF-Deep RVFL framework, owing to superior forecasting performance, outperformed the RFE-EDF-Deep CNN-LSTM, RFE-EDF-RVFL and all the standalone counterpart models.

The watershed dams water harvesting establishment is a complex decision-making process due to the existence of several critical factors on which the project is based adopted. Such complexity renders consideration for the priority scale of development, especially when the project considers several possible sites against a plethora of criteria. These criteria may be of a technical, economic, social, cultural, environmental, legal, and political nature, thereby increasing the complexity of the framework under which the decision has to be made. Appropriate solutions for the challenge have to consider Multi-Criteria Decision-Making (MCDM) approaches. MCDM is defined as an organized approach to decision-making designed to help in site selection from amongst alternative feasible sites based on a range of criteria. Most of the prominent MCDM methods used over the literature are including Analytic Hierarchy Process (AHP), Weighted Linear Combination (WLC), Fuzzy logic (FL), Boolean Methods, Technique for Order Preference by Similarity to Ideal Solution (TOPSIS). The MCDM methods represent a different strategy of integration and analysis of multi-dimensional criteria that influence the choice of the dam site. The advantages of the MCDM methods are not only help to undertake a detailed study of priorities for the construction of the dam; however, they devise a methodological approach to the selection of the most appropriate location of dams for their construction. The present survey exhibited and evaluated all the conducted researches to rank the dam construction projects using the MCDM methods. This review provides an comprehensive assessment and evaluation for the literature and recognizes the opportunities for the improvement of the MCDM methods effectiveness under complex dam site selection problems. The survey is expected to contribute toward refining strategic decision-making processes required for sustainable infrastructure development.