Show simple item record

dc.contributor.advisorLiyantono
dc.contributor.authorErsaputra, Dionisius Ariel Elazaro
dc.date.accessioned2026-08-14T04:52:18Z
dc.date.available2026-08-14T04:52:18Z
dc.date.issued2026
dc.identifier.urihttp://repository.ipb.ac.id/handle/123456789/178638
dc.description.abstractLahan sawah di Kota Sukabumi menyusut dari 1.365 hektar (2019) menjadi 1.272 hektar (2025), sementara realisasi kawasan Lahan Pertanian Pangan Berkelanjutan (LP2B) dalam Rencana Tata Ruang Wilayah (RTRW) 2022-2042 baru mencapai 11,13% dari target 425 hektar. Penelitian ini bertujuan memetakan dinamika perubahan lahan sawah periode 2019-2025 serta mengevaluasi dampaknya terhadap kawasan LP2B. Metode yang diterapkan adalah klasifikasi multitemporal berbasis Random Forest menggunakan fusi citra Sentinel-1, Sentinel-2, GLO-30, dan FABDEM pada platform Google Earth Engine, divalidasi melalui confusion matrix, ground check lapangan, dan validasi eksternal lintas wilayah di Denpasar. Model menghasilkan Overall Accuracy 97,5% dan Koefisien Kappa 0,962, meskipun validasi eksternal menunjukkan penurunan akurasi menjadi 69,0% akibat keterbatasan representasi sampel training pada satu domain geografis. Alih fungsi permanen teridentifikasi seluas 57,28 hektar, alih fungsi reversibel 22,5 hektar, dan lahan tidak aktif berisiko laten 61,30 hektar. Pengujian kombinasi citra mengungkap temuan bahwa data Sentinel-1 tunggal justru lebih unggul (79,58%) untuk deteksi perubahan dibanding kombinasi multisensor. Prototipe Early Warning System berbasis Google Earth Engine dikembangkan dengan interval 4 bulan (73,75%, optimal untuk petak kecil) dan 6 bulan (76,25%, akurasi keseluruhan terbaik) sebagai instrumen operasional deteksi dini mendukung penguatan kebijakan LP2B.
dc.description.abstractPaddy fields in Kota Sukabumi shrank from 1,365 hectares (2019) to 1,272 hectares (2025), while realization of the Sustainable Food Agricultural Land (LP2B) zone under the 2022-2042 Spatial Plan (RTRW) has reached only 11.13% of the 425- hectare target. This study aims to map paddy field change dynamics for 2019-2025 and evaluate their impact on LP2B areas. A multi-temporal Random Forest classification was applied, fusing Sentinel-1, Sentinel-2, GLO-30, and FABDEM imagery on Google Earth Engine, validated through a confusion matrix, field ground checks, and external cross-regional validation in Denpasar. The model achieved 97.5% Overall Accuracy and a Kappa Coefficient of 0.962, although external validation showed a decline to 69.0% accuracy, indicating limited generalization from single-domain training samples. Permanent conversion was identified across 57.28 hectares, reversible conversion across 22.5 hectares, and 61.30 hectares of inactive paddy fields carrying latent conversion risk. Image-combination testing revealed that single Sentinel-1 data (S1_Only) outperformed multi-sensor combinations (79.58%) for change detection, unlike its role in static classification. An Google Earth Engine based Early Warning System prototype was developed with 4-month (73.75%, optimal for small plots) and 6-month (76.25%, best overall accuracy) monitoring intervals as an operational instrument for early detection to strengthen LP2B policy implementation.
dc.description.sponsorship
dc.language.isoid
dc.publisherIPB Universityid
dc.titleEvaluasi Spasial Alih Fungsi Lahan Sawah Kota Sukabumi Tahun 2019-2025 dengan Metode Analisis Multitemporal dan Kombinasi Citra Satelitid
dc.title.alternativeSpatial Evaluation of Paddy Fields Land-Use Conversion in Sukabumi City Using Multi-temporal Methods and Combined Satellite Imagery
dc.typeSkripsi
dc.subject.keywordpaddy field conversionid
dc.subject.keywordearly warning systemid
dc.subject.keywordGoogle Earth Engineid
dc.subject.keywordLP2Bid
dc.subject.keywordrandom forestid
dc.subtypeUndergraduate Theses


Files in this item

Thumbnail
Thumbnail
Thumbnail

This item appears in the following Collection(s)

Show simple item record