Monitoring Perubahan Massa Batuan Di Lapangan Panas Bumi Dieng Menggunakan Data Gayaberat Satelit Dan Algoritma Machine Learning

Pertiwi, Angger Pandhu (2026) Monitoring Perubahan Massa Batuan Di Lapangan Panas Bumi Dieng Menggunakan Data Gayaberat Satelit Dan Algoritma Machine Learning. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5017221043-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (5MB) | Request a copy

Abstract

Lapangan Panas Bumi Dieng merupakan sistem panas bumi liquid -dominated yang dipengaruhi oleh dinamika fluida reservoir. Aktivitas produksi dan reinjeksi dapat menyebabkan redistribusi massa bawah permukaan yang berpengaruh terhadap perubahan nilai gayaberat. Penelitian ini bertujuan untuk menganalisis pola spasial dan temporalComplete Bouguer Anomaly (CBA) berbasis data gayaberat satelit serta mengevaluasikemampuan algoritma machine learning dalam mengenali pola perubahan CBA sebagai pendukung monitoring perubahan massa bawah permukaan. Data yang digunakan berupa data gayaberat satelit periode Agustus 2018 hingga November 2025 pada 16 titik pengamatan di Lapangan Panas Bumi Dieng. Data diolah hingga menghasilkan nilai CBA, kemudian dianalisis secara temporal dan spasial melalui grafik deret waktu serta peta interpolasi tahunan. Selanjutnya, nilai CBA diprediksi menggunakan dua algoritma, yaitu Long Short -Term Memory (LSTM) dan Extreme Gradient Boosting (XGBoost), dengan lima variasi rasio data latih dan dat a uji untuk mengevaluasi kestabilan model melalui pendekatan backtesting. Hasil penelitian menunjukkan bahwa pola CBA secara spasial relatif stabil dari tahun 2018 hingga 2025. Zona CBA tinggi dominan berada di bagian utara–barat laut, sedangkan zona CBA rendah berkembang ke arah tengah–tenggara. Pola ini mengindikasikan adanya variasi massa regional yang kemungkinan dipengaruhi oleh litologi vulkanik, kontras densitas batuan, dan sebaran alterasi hidrotermal. Secara temporal, nilai CBA menunjukkan fluktuasi non-linear yang mengindikasikan adanya dinamika massa relatif bawah permukaan. Hasil evaluasi model menunjukkan bahwa XGBoost menghasilkan nilai RMS/RMSE lebih rendah dibandingkan LSTM pada seluruh variasi data. Oleh karena itu, XGBoost dipilih sebagaimodelyang lebih stabil untuk mendukung monitoring temporal nilai CBA pada data terbatas.
==================================================================================================================================
The Dieng Geothermal Field is a liquid-dominated geothermal system influenced by reservoir fluid dynamics. Production and reinjection activities can cause subsurface mass redistribution, which affects changes in gravity values. This study aims to analyze the spatial and temporal patterns of the Complete Bouguer Anomaly (CBA) based on satellite gravity data and to evaluate the capability of machine learning algorithms in recognizing CBA change patterns to support subsurface mass change monitoring. The data used consist of satellite gravity data from August 2018 to November 2025 across 16 observation points in the Dieng Geothermal Field. The data were processed to generate CBA values, then analyzed temporally and spatially through time-series graphs and annual interpolation maps. Furthermore, the CBA values were predicted using two algorithms, namely Long Short-Term Memory (LSTM) and Extreme Gradient Boosting (XGBoost), with five variations of train-test data ratios to evaluate model stability using a backtesting approach. The results indicate that the spatial CBA pattern was relatively stable from 2018 to 2025. High CBA zones are dominantly located in the northern–northwestern part, while low CBA zones develop towards the central–southeastern direction. This pattern indicates regional mass variations likely influenced by volcanic lithology, rock density contrasts, and the distribution of hydrothermal alteration. Temporally, the CBA values exhibit non-linear fluctuations, indicating subsurface relative mass dynamics. Model evaluation results show that XGBoost yields lower RMS/RMSE values compared to LSTM across all data variations. Therefore, XGBoost is selected as the more stable model to support the temporal monitoring of CBA values under limited data conditions.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Lapangan Panas Bumi Dieng, GRACE-FO, Complete Bouguer Anomaly (CBA), LSTM, XGBoost, Dieng Geothermal Field, GRACE-FO, Complete Bouguer Anomaly (CBA), LSTM, XGBoost.
Subjects:	G Geography. Anthropology. Recreation > GB Physical geography > GB1199.5 Geothermal resources Q Science > Q Science (General) > Q180 Gravitation. Q Science > Q Science (General) > Q325.5 Machine learning. Support vector machines.
Divisions:	Faculty of Civil, Planning, and Geo Engineering (CIVPLAN) > Geophysics Engineering > 33201-(S1) Undergraduate Thesis
Depositing User:	Angger Pandhu Pertiwi
Date Deposited:	14 Aug 2026 06:04
Last Modified:	14 Aug 2026 06:04
URI:	http://repository.its.ac.id/id/eprint/144196

Actions (login required)

View Item