Scholarlink

<p>The late 20th and early 21st centuries have witnessed significant increases in land-use/land-cover (LULC) changes due to rapid population growth, economic expansion, and industrial development, especially in emerging nations. Assessing the accuracy of various LULC mapping algorithms is crucial for identifying the best classifier for earth observation applications, as quantifying LULC changes is vital for understanding and managing land transformation. This study focuses on Cuddalore Taluk in Tamil Nadu, which is experiencing notable LULC changes. Urbanisation and reductions in agricultural land have significantly contributed to these transformations. Using Landsat series satellite data, we evaluated two machine-learning methods: Support Vector Machine (SVM) and Random Forest (RF). The National Remote Sensing Centre (NRSC) Level 1 classification was employed to categorize LULC features, including water bodies, agricultural land, built-up land, barren land, grassland, and mining areas. Accuracy assessments revealed that the RF algorithm achieved a higher accuracy score of 0.92 compared to the SVM method, which had an accuracy score of 0.81. From 1993 to 2023, LULC features changed as follows: water bodies decreased from 3.89 to 0.90%, agricultural land decreased from 63.96 to 56.02%, grassland decreased from 16.98 to 13.30%, built-up land increased from 1.72 to 7.43%, mining increased from 0.72 to 2.65%, and barren land increased from 12.78 to 19.69%. This study concludes that the RF algorithm is the superior LULC classifier. However, further testing of the RF algorithm in various morphoclimatic conditions is recommended. Detailed studies and ongoing monitoring are essential for informed decision-making and effective land use planning.</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

Land use and land cover change detection by machine learning classifiers (SVM and RF) using satellite remote sensing observations for Cuddalore Taluk, Tamil Nadu, India

  • 期刊论文

  • 发表日期：

    2025年1月11日

查看全文
  • N Gobika Shree,
  • C Meiaraj

摘要

The late 20th and early 21st centuries have witnessed significant increases in land-use/land-cover (LULC) changes due to rapid population growth, economic expansion, and industrial development, especially in emerging nations. Assessing the accuracy of various LULC mapping algorithms is crucial for identifying the best classifier for earth observation applications, as quantifying LULC changes is vital for understanding and managing land transformation. This study focuses on Cuddalore Taluk in Tamil Nadu, which is experiencing notable LULC changes. Urbanisation and reductions in agricultural land have significantly contributed to these transformations. Using Landsat series satellite data, we evaluated two machine-learning methods: Support Vector Machine (SVM) and Random Forest (RF). The National Remote Sensing Centre (NRSC) Level 1 classification was employed to categorize LULC features, including water bodies, agricultural land, built-up land, barren land, grassland, and mining areas. Accuracy assessments revealed that the RF algorithm achieved a higher accuracy score of 0.92 compared to the SVM method, which had an accuracy score of 0.81. From 1993 to 2023, LULC features changed as follows: water bodies decreased from 3.89 to 0.90%, agricultural land decreased from 63.96 to 56.02%, grassland decreased from 16.98 to 13.30%, built-up land increased from 1.72 to 7.43%, mining increased from 0.72 to 2.65%, and barren land increased from 12.78 to 19.69%. This study concludes that the RF algorithm is the superior LULC classifier. However, further testing of the RF algorithm in various morphoclimatic conditions is recommended. Detailed studies and ongoing monitoring are essential for informed decision-making and effective land use planning.