Please use this identifier to cite or link to this item: http://repository.ipb.ac.id/handle/123456789/171999
Title: Tingkat Akurasi Kombinasi Band Citra Sentinel-2 dan Integrasinya dengan Sentinel-1 dalam Klasifikasi Tutupan Lahan
Other Titles: The Accuracy Level of the Citra Sentinel-2 Band Combination and its Integration with Sentinel-1 in Classification Land Cover
Authors: Priyanto
Fayiz, M. Zhafran
Issue Date: 2026
Publisher: IPB University
Abstract: Informasi klasifikasi tutupan lahan yang akurat sangat penting dalam pemantauan perubahan penggunaan lahan dan pengelolaan sumber daya secara berkelanjutan. Penelitian ini bertujuan membandingkan skenario penggunaan band dan integrasi citra Sentinel serta pembuatan klasifikasi tutupan lahan menggunakan algoritma random forest dan evaluasinya terhadap perubahan tutupan. Data dikumpulkan untuk dua periode (2018 dan 2024) pada musim kemarau dan hujan, dilengkapi dengan 180 titik verifikasi lapangan. Integrasi informasi spektral dan tekstural menunjukkan peningkatan akurasi dengan nilai akurasi keseluruhan sebesar 95,56% dan akurasi Kappa 94,67%. Temuan penting menunjukkan bahwa integrasi multi-sensor meningkatkan ketepatan klasifikasi, terutama pada kelas hutan, pertambangan, dan pertanian. Analisis perubahan tutupan lahan menunjukkan konversi positif seluas 644,13 ha didominasi dari tutupan lahan pertambangan menjadi tutupan hutan, sedangkan perubahan negatif seluas 660,3 ha didominasi oleh alih konversi sawah menjadi permukiman. Pendekatan ini memberikan implikasi penting dalam pemantauan dinamis lanskap dan mendukung perencanaan tata ruang berbasis integrasi data citra berbeda.
Accurate land cover classification information is essential for monitoring land use change and sustainable resource management. This study aims to compare the scenario of band use and the integration of Sentinel imagery as well as the creation of land cover classification using random forest algorithm and its evaluation of cover changes. The data was collected for two periods (2018 and 2024) in the dry and rainy seasons, complemented by 180 field verification points. The integration of spectral and textural information showed an increase in accuracy with an Overall Accuracy value of 95,56% and a Kappa of 94,67%. Important findings show that multi-sensor integration improves classification accuracy, especially in forest, mining, and agricultural classes. Analysis of land cover change showed a positive conversion of 644,13 ha were dominated from mining land cover becomes forest cover, while negative changes of 660,3 ha were dominated by conversion of rice fields into settlements. This approach provides important implications in dynamic monitoring of landscapes and supports spatial planning based on the integration of different image data
URI: http://repository.ipb.ac.id/handle/123456789/171999
Appears in Collections:UT - Forest Management

Files in This Item:
File Description SizeFormat 
cover_E1401211078_4fea60af64434b07bc5fa70755bc89c9.pdfCover445.37 kBAdobe PDFView/Open
fulltext_E1401211078_d4b0a5e0713040a9b77d00292d1b55fe.pdf
  Restricted Access
Fulltext1.2 MBAdobe PDFView/Open
lampiran_E1401211078_a9ff24de91b84977bffc856f3753a493.pdf
  Restricted Access
Lampiran179.18 kBAdobe PDFView/Open


Items in DSpace are protected by copyright, with all rights reserved, unless otherwise indicated.