Estimasi Konsentrasi Klorofil-A di Perairan Danau Laguna Filipina menggunakan Model Extreme Gradient Boost

Putra, David Beta Putra (2025) Estimasi Konsentrasi Klorofil-A di Perairan Danau Laguna Filipina menggunakan Model Extreme Gradient Boost. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6016222004-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (3MB) | Request a copy

Abstract

Danau merupakan sumber daya air tawar yang penting dan secara historis telah menjadi sumber utama air untuk konsumsi manusia. Namun, kondisi ekologisnya semakin mengalami degradasi akibat meningkatnya aktivitas antropogenik di wilayah sekitarnya. Salah satu dampak utama dari degradasi ini adalah eutrofikasi, yang berkaitan erat dengan meningkatnya konsentrasi klorofil-a. Penelitian ini bertujuan untuk mengestimasi konsentrasi klorofil-a di Danau Laguna menggunakan algoritma Extreme Gradient Boosting (XGBoost), yaitu metode pembelajaran mesin berbasis pohon keputusan. XGBoost dipilih karena kemampuannya dalam menyediakan informasi pentingnya tiap variabel input (feature importance), sehingga memungkinkan evaluasi kontribusi masing-masing variabel terhadap performa model. Kerangka metodologi mencakup pra-pemrosesan citra satelit Sentinel-3 untuk mengekstraksi nilai reflektansi permukaan, kemudian dilanjutkan dengan pelatihan model XGBoost menggunakan pendekatan validasi silang 10-fold. Performa model dievaluasi menggunakan Root Mean Square Error (RMSE) dan koefisien determinasi (R²). Model XGBoost menunjukkan performa yang stabil dengan nilai rata-rata RMSE sebesar 1,8234 µg/L dan R² sebesar 0,2507. Analisis komparatif dengan algoritma lain, termasuk Decision Tree dan Decision Tree, menunjukkan bahwa XGBoost memiliki akurasi prediksi yang lebih baik. Hasil pemetaan spasial menunjukkan distribusi klorofil-a yang konsisten secara musiman, dengan konsentrasi yang lebih tinggi teramati pada musim hujan. Temuan ini menegaskan potensi algoritma berbasis pohon keputusan sebagai alternatif yang layak terhadap pendekatan jaringan saraf tiruan (neural network) yang lebih umum digunakan dalam pemodelan lingkungan berbasis citra satelit.
====================================================================================================================================
Lakes represent a vital freshwater resource and have historically served as key sources of water for human consumption. However, their ecological conditions have been increasingly degraded due to intensified anthropogenic activities in surrounding areas. One major impact of this degradation is eutrophication, which is closely associated with elevated concentrations of chlorophyll-a. This study aims to estimate chlorophyll-a concentrations in Laguna Lake using the Extreme Gradient Boosting (XGBoost) algorithm, a Decision Tree-based machine learning method. XGBoost was selected due to its capability to provide feature importance, allowing for the evaluation of each input variable’s significance in model performance. The methodological framework involved preprocessing Sentinel-3 satellite imagery to extract surface reflectance values, followed by training the XGBoost model using a 10-fold cross-validation approach. Model performance was assessed using Root Mean Square Error (RMSE) and the coefficient of determination (R²). The XGBoost model exhibited stable performance, achieving an Average RMSE of 1.8234 µg/L and an R² value of 0.2507. Comparative analysis with other algorithms, including Decision Tree and Decision Tree, demonstrated that XGBoost outperformed the alternatives in terms of predictive accuracy. Spatial mapping results revealed seasonally consistent distributions of chlorophyll-a, with higher concentrations observed during the rainy season. These findings underscore the potential of Decision Tree-based algorithms as viable alternatives to neural network approaches, which are more commonly employed in satellite-based environmental modeling.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Danau Laguna, Klorofil-a, Uji Akurasi, XGBoost, Laguna Lake, Chlorophyll-a, Predictive Accuracy, XGBoost
Subjects:	G Geography. Anthropology. Recreation > G Geography (General) > G70.5.I4 Remote sensing G Geography. Anthropology. Recreation > GC Oceanography > GC101.2 Seawater--Analysis T Technology > TD Environmental technology. Sanitary engineering > TD420 Water pollution T Technology > TD Environmental technology. Sanitary engineering > TD890 Global Environmental Monitoring System
Divisions:	Faculty of Civil Engineering and Planning > Geomatics Engineering > 29101-(S2) Master Thesis
Depositing User:	David Beta Putra
Date Deposited:	31 Jul 2025 04:21
Last Modified:	31 Jul 2025 04:21
URI:	http://repository.its.ac.id/id/eprint/124589

Actions (login required)

View Item