Scholarlink

<p>Groundwater is essential for supporting agriculture, home supplies, and ecosystem stability, particularly in monsoon-dependent areas like India, where surface water availability is markedly seasonal. Notwithstanding its importance, precisely forecasting groundwater levels remains a formidable challenge due to the intrinsic nonlinearity, noise, and seasonal fluctuations of hydrological systems. Traditional statistics and machine learning models often fail to adequately capture these intricate temporal-spatial dynamics, leading to diminished predictive accuracy and incorrect management insights. This paper presents a novel hybrid forecasting framework, the Polar Fox-based Sequential Neural Network (PFbSNN), which combines the optimisation efficiency of the Polar Fox algorithm with the temporal learning capabilities of a sequential neural network. The model was trained and validated using a comprehensive dataset of 2196 groundwater observations collected from 10 districts in Telangana, India, for the period 2018–2020. The procedure encompassed methodical data preprocessing for noise attenuation, feature extraction to enhance input significance, and temporal modelling using PFbSNN to forecast groundwater levels for both pre- and post-monsoon periods. The performance evaluation demonstrated that the PFbSNN achieved remarkable prediction accuracy, as evidenced by an R<sup>2</sup> value of 0.99, a Root Mean Square Error of 0.18&#xa0;m, and a Mean Absolute Error of 0.05&#xa0;m, significantly exceeding those of traditional and independent neural network models. The model accurately captured seasonal variations and regional differences in groundwater dynamics, demonstrating its ability to manage intricate hydrological relationships. The findings underscore the PFbSNN model as a viable decision-support instrument for groundwater monitoring, resource management, and sustainable planning in semi-arid and monsoon-influenced environments. Future endeavours will focus on expanding the model to encompass broader spatio-temporal scales, integrating climatic and land-use predictors, and evaluating its transferability across various hydrogeological contexts.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Ground Water Modeling for Analysing the Water Level of Wells Before and After Monsoon

  • 期刊论文

  • 发表日期：

    2026年6月22日

查看全文
  • Deshmukh Sachin Chandrakant,
  • Goushiya Sayyed,
  • Ubale Mayur Ashok,
  • Gosavi Shrikrishna Avinash,
  • Bhaganagare Prashant Balu,
  • Dhulappa Balu Borkar

摘要

Groundwater is essential for supporting agriculture, home supplies, and ecosystem stability, particularly in monsoon-dependent areas like India, where surface water availability is markedly seasonal. Notwithstanding its importance, precisely forecasting groundwater levels remains a formidable challenge due to the intrinsic nonlinearity, noise, and seasonal fluctuations of hydrological systems. Traditional statistics and machine learning models often fail to adequately capture these intricate temporal-spatial dynamics, leading to diminished predictive accuracy and incorrect management insights. This paper presents a novel hybrid forecasting framework, the Polar Fox-based Sequential Neural Network (PFbSNN), which combines the optimisation efficiency of the Polar Fox algorithm with the temporal learning capabilities of a sequential neural network. The model was trained and validated using a comprehensive dataset of 2196 groundwater observations collected from 10 districts in Telangana, India, for the period 2018–2020. The procedure encompassed methodical data preprocessing for noise attenuation, feature extraction to enhance input significance, and temporal modelling using PFbSNN to forecast groundwater levels for both pre- and post-monsoon periods. The performance evaluation demonstrated that the PFbSNN achieved remarkable prediction accuracy, as evidenced by an R2 value of 0.99, a Root Mean Square Error of 0.18 m, and a Mean Absolute Error of 0.05 m, significantly exceeding those of traditional and independent neural network models. The model accurately captured seasonal variations and regional differences in groundwater dynamics, demonstrating its ability to manage intricate hydrological relationships. The findings underscore the PFbSNN model as a viable decision-support instrument for groundwater monitoring, resource management, and sustainable planning in semi-arid and monsoon-influenced environments. Future endeavours will focus on expanding the model to encompass broader spatio-temporal scales, integrating climatic and land-use predictors, and evaluating its transferability across various hydrogeological contexts.