Developing an SSA-optimized attention-ConvGRU model for predicting and assessing soil contaminant distribution

Abstract

Introduction: Soil pollution, which includes a variety of contaminants such as heavy metals and organic compounds, poses significant environmental and health risks, making effective prediction and assessment techniques essential. Current predictive models often struggle with the complexity and diversity of soil contaminant behaviors, leading to limitations in their accuracy and applicability.Methods: Recognizing the importance of capturing the temporal dynamics influenced by seasonal variations and agricultural practices, our study introduces an SSA-optimized Attention-ConvGRU model. This model integrates convolutional neural networks, gated recurrent units, and attention mechanisms, enhanced through optimization with the Sparrow Search Algorithm to improve predictive performance.Results: Experimental results confirm that our model significantly outperforms traditional methods, demonstrating over 30% improvement in prediction accuracy across multiple datasets.Discussion: This research underscores the potential of advanced machine learning techniques to revolutionize the assessment of soil pollution, providing substantial benefits for environmental management and public health protection.