Land Use and Land
Cover Classification using Support Vector Machine Technique: A Case Study of Kaddam
Watershed
Sridhar P., Anil K., Raju J.N.S.S. and Ramana K.V.
Disaster Advances; Vol. 18(6); 1-6; doi:
https://doi.org/10.25303/186da0106; (2025)
Abstract
Land use and land cover (LU/LC) maps, along with their temporal dynamics, are essential
for flood prediction, seasonal water quality monitoring, environmental sustainability
planning and ecological assessments. Accurate classification of satellite image
datasets presents a challenging task due to the complexity of land use patterns
and the need for precise methods. To address these challenges, this study utilized
multi-temporal satellite image datasets to perform LU/LC classification and analyze
temporal changes within the Kaddam watershed. The Support Vector Machine (SVM) classification
technique was employed, using training samples that represent critical land use
classes including water bodies, agricultural land, forests, urban areas and barren
land. Representative polygons were digitized to train the SVM model and key parameters
such as kernel type and gamma values were optimized to enhance classification accuracy.
Performance evaluation was conducted using a confusion matrix to derive metrics
such as overall accuracy and Kappa statistics. Ground truth data comparisons further
validated the classification results. The high accuracy and robustness of the SVM-based
approach demonstrate its potential as a reliable tool for LU/LC classification and
its applicability to other regions for effective land use management and planning.
