Electronic Theses and Dissertation

Kelapa sawit memiliki peran strategis dalam perekonomian indonesia, terutama dalam industri pangan, energi, dan manufaktur, sehingga memerlukan manajemen perkebunan yang optimal. salah satu aspek kunci dalam pengelolaan adalah pemantauan umur tanaman, yang berdampak langsung pada produktivitas dan perencanaan panen. namun, pemantauan umur tanaman sawit masih menjadi tantangan karena metode konvensional seperti survei lapangan memerlukan waktu dan biaya tinggi. penelitian ini bertujuan untuk mendeteksi dan memetakan umur kelapa sawit menggunakan citra satelit landsat 9 dan algoritma extreme gradient boosting (xgboost) berbasis google earth engine (gee) dan google collaboratory. citra landsat 9 diproses dengan teknik analisis spektral, termasuk penggunaan kanal b2, b3, b4, b5, b6 dan b11 serta indeks vegetasi seperti ndvi, evi, savi, mndwi, ndbi, avi, bsi, si dan ti. model xgboost diterapkan untuk klasifikasi umur sawit berdasarkan karakteristik spektral, dengan evaluasi performa menggunakan precision, recall, dan f1-score. hasil penelitian menunjukkan bahwa model xgboost mampu mengklasifikasikan umur kelapa sawit dengan akurasi keseluruhan sebesar 98,35%, dengan precision sebesar 97,2% untuk kelas sawit muda dan 97,3% untuk kelas sawit tua. peta tematik yang dihasilkan berhasil menggambarkan persebaran spasial umur tanaman kelapa sawit di kecamatan kaway xvi, kabupaten aceh barat, serta memberikan informasi luas tutupan lahan untuk setiap kelas umur. perbedaan karakteristik spektral, terutama pada nilai indeks vegetasi dan termal, berhasil diidentifikasi secara konsisten dan signifikan antara tanaman muda dan tua. kombinasi citra landsat 9 dan xgboost mampu meningkatkan akurasi klasifikasi umur kelapa sawit, bahkan di area yang sulit dijangkau. pendekatan berbasis remote sensing dan machine learning ini menawarkan solusi yang lebih efisien untuk pemantauan perkebunan sawit, mendukung pengelolaan yang lebih baik, serta berkontribusi pada kebijakan keberlanjutan lingkungan.

Oil palm plays a strategic role in Indonesia's economy, particularly in the food, energy, and manufacturing industries, thus requiring optimal plantation management. One key aspect of management is monitoring the age of plants, which directly impacts productivity and harvest planning. However, monitoring the age of oil palm plants remains a challenge because conventional methods such as field surveys are time-consuming and costly. This study aims to detect and map the age of oil palms using Landsat 9 satellite imagery and the Extreme Gradient Boosting (XGBoost) algorithm based on Google Earth Engine (GEE) and Google Collaboratory. Landsat 9 images were processed using spectral analysis techniques, including the use of channels B2, B3, B4, B5, B6 and B11, as well as vegetation indices such as NDVI, EVI, SAVI, MNDWI, NDBI, AVI, BSI, SI and TI. The XGBoost model was applied for oil palm age classification based on spectral characteristics, with performance evaluation using precision, recall, and F1-score. The study results showed that the XGBoost model could classify oil palm age with an overall accuracy of 98.35%, with precision of 97.2% for young oil palm class and 97.3% for old oil palm class. The thematic map successfully depicted the spatial distribution of oil palm age in Kaway XVI Subdistrict, Aceh Barat Regency, and provided information on land cover area for each age class. Differences in spectral characteristics, particularly in vegetation and thermal indices, were consistently and significantly identified between young and old plants. The combination of Landsat 9 imagery and XGBoost was able to improve the accuracy of oil palm age classification, even in hard-to-reach areas. This remote sensing and machine learning-based approach offers a more efficient solution for oil palm plantation monitoring, supports better management, and contributes to environmental sustainability policies.