Article Preview
TopIntroduction
In recent years, more attention has been paid to the role played by artificial intelligence applications in numerous industries. Based on data from the Gravity Recovery and Climate Experiment (GRACE), this study uses LSTM networks to track and predict TWSC and GWSC during the period from 2003 to 2025 for five basins in Saudi Arabia (Haq, Azam, et al., 2021). By analyzing large amounts of remotely sensed Moderate Resolution Imaging Spectroradiometer (MODIS) data, meteorological records, and simulated global climate data, this study provides a comprehensive overview of changes in vegetation, snow cover, and temperature patterns in Uttarakhand State, India. To further investigate the potential for predicting these environmental variables, we performed regression using machine learning (ML) methods, such as support vector machines (SVMs) and long short-term memory (LSTM) networks. In the present study, we used high-resolution PS nanosatellite data to evaluate agricultural activity in the Al-Qassim region of Saudi Arabia (KSA) across time. To assess the impact of time on the vegetation pattern, we generated an NDVI time series. Sparse foliage and brilliant exposed soil due to poor soil moisture limit NDVI in the current study area. Therefore, to evaluate the relative computing cost of NDVI and the area of vegetation, we employed an ML-based random forest (RF) classification model (Haq, Baral, et al., 2021). The study’s goals include filling in these blanks and providing useful information for classifying and forecasting air pollution.
To classify air pollution, we created five ML models, one of which was new and was given the term synthetic minority oversampling technique with deep neural network (SMOTEDNN). Hyperparameter tuning and effective data pre-processing were used across all five models. We created three statistical autoregression-based models for air pollution forecasting. The accuracy of all the models developed in this study improved overall. The unique SMOTEDNN model significantly outperformed the other models used in this study and earlier research in terms of accuracy (99.90%). The regionally distributed likelihood of permafrost occurrence might be reliably estimated using the results of logistic regression models. However, the results from these models varied depending on which topoclimatic and topographic variables were employed as predictors during the model’s computation (Haq & Baral, 2019). This was especially true for the location points chosen for rock glaciers’ initiation lines. This study focuses on creating a novel automated weed-identifying method for a real-world dataset consisting of 4,400 UAV images and 153,360 discrete weed features using convolutional neural network (CNN) classification. The best parameters for the proposed CNN LVQ model were determined with the help of snapshots. After extensive hyperparameter tuning, the generated CNN LVQ model achieved an overall accuracy of 99.44 percent in weed detection, a substantial improvement over previously reported studies.
We built and fine-tuned the climate deep long short-term memory (CDLSTM) model to make accurate temperature and precipitation predictions for all Himalayan states. To make predictions and evaluate the established CDLSTM model’s efficacy, we implemented the Facebook Prophet (FB-Prophet) model. Both models were assessed using a wide range of performance indicators, and the results showed that they were very accurate and had low error rates. We developed the models using GridsearchCV for precise hyperparameter tuning. Overfitting and underfitting were prevented in both DNN models by using early stopping. They were among the best-performing models in terms of accuracy and efficiency after extensive testing on the produced suite of tools. The study’s novelty lies in the generated models’ ability to fill these gaps using a real-world dataset while maintaining a low false alarm rate. The ANN model estimates align well with ground-penetrating radar measurements of ice thickness for five transverse cross-sections of Chhota Shigri Glacier.