Scholarlink

<p>Accurate prediction of soil properties is essential for sustainable agriculture and land management in arid regions where water scarcity and soil degradation pose critical challenges. This study evaluates and compares multiple interpolation techniques for mapping surface and subsurface soil attributes in Egypt’s New Delta, including soil reaction pH, electrical conductivity (EC), calcium carbonate (CaCO₃), and gravel content. A total of 468 soil samples were collected from 234 profiles (0–30&#xa0;cm and 30–60&#xa0;cm depths) under arid climatic conditions characterized by hot summers (mean 27.8&#xa0;°C, maxima &gt; 38&#xa0;°C), mild winters (mean 13.7&#xa0;°C), and variable soil moisture regimes. Deterministic methods Inverse Distance Weighting (IDW), Global Polynomial Interpolation (GPI), Radial Basis Functions (RBF), and Local Polynomial Interpolation (LPI) and geostatistical approaches Ordinary Kriging (OK), Simple Kriging (SK), Universal Kriging (UK), Indicator Kriging (IK), Probability Kriging (PK), Disjunctive Kriging (DsK), Empirical Bayesian Kriging (EBK), Kernel Smoothing (KS), and Diffusion Kernel (DfK) were tested, with accuracy assessed using cross-validation metrics (Root Mean Square Error, RMSE; coefficient of determination, R<sup>2</sup>). Results revealed pronounced spatial variability: surface soils ranged from slightly to very strongly alkaline, non-saline to highly saline, and non-calcareous to very calcareous, while subsurface soils exhibited lower pH and CaCO₃ but higher salinity. Comparative evaluation showed that the exponential semivariogram model under OK was most effective for pH and gravel prediction, the J-Bessel model under OK and UK performed best for EC, and the Penta-spherical model under OK was optimal for CaCO₃. Among deterministic methods, RBF with inverse multi-quadratic functions achieved superior accuracy across properties. EBK with circular semivariograms consistently provided robust predictions for all soil attributes. These findings demonstrate that model performance varies by property and method, and the generated soil maps provide actionable insights for irrigation scheduling, soil improvement, and crop selection. The study highlights the importance of integrating geostatistical interpolation with sustainability frameworks, offering transferable lessons for precision land management in arid regions globally.</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

Comparative geospatial interpolation of soil properties in arid lands: insights for sustainable agriculture and land management in Egypt’s Northwestern desert

  • 期刊论文

  • 发表日期：

    2026年7月7日

查看全文
  • Mohamed A. E. AbdelRahman,
  • Mohamed M. Metwaly,
  • Taher M. H. Yossif,
  • Ali R. A. Moursy

摘要

Accurate prediction of soil properties is essential for sustainable agriculture and land management in arid regions where water scarcity and soil degradation pose critical challenges. This study evaluates and compares multiple interpolation techniques for mapping surface and subsurface soil attributes in Egypt’s New Delta, including soil reaction pH, electrical conductivity (EC), calcium carbonate (CaCO₃), and gravel content. A total of 468 soil samples were collected from 234 profiles (0–30 cm and 30–60 cm depths) under arid climatic conditions characterized by hot summers (mean 27.8 °C, maxima > 38 °C), mild winters (mean 13.7 °C), and variable soil moisture regimes. Deterministic methods Inverse Distance Weighting (IDW), Global Polynomial Interpolation (GPI), Radial Basis Functions (RBF), and Local Polynomial Interpolation (LPI) and geostatistical approaches Ordinary Kriging (OK), Simple Kriging (SK), Universal Kriging (UK), Indicator Kriging (IK), Probability Kriging (PK), Disjunctive Kriging (DsK), Empirical Bayesian Kriging (EBK), Kernel Smoothing (KS), and Diffusion Kernel (DfK) were tested, with accuracy assessed using cross-validation metrics (Root Mean Square Error, RMSE; coefficient of determination, R2). Results revealed pronounced spatial variability: surface soils ranged from slightly to very strongly alkaline, non-saline to highly saline, and non-calcareous to very calcareous, while subsurface soils exhibited lower pH and CaCO₃ but higher salinity. Comparative evaluation showed that the exponential semivariogram model under OK was most effective for pH and gravel prediction, the J-Bessel model under OK and UK performed best for EC, and the Penta-spherical model under OK was optimal for CaCO₃. Among deterministic methods, RBF with inverse multi-quadratic functions achieved superior accuracy across properties. EBK with circular semivariograms consistently provided robust predictions for all soil attributes. These findings demonstrate that model performance varies by property and method, and the generated soil maps provide actionable insights for irrigation scheduling, soil improvement, and crop selection. The study highlights the importance of integrating geostatistical interpolation with sustainability frameworks, offering transferable lessons for precision land management in arid regions globally.