Putri, Belvana Eka (2025) BathySAR-Net: Model Berbasis Artificial Neural Network Untuk Memprediksi Kedalaman Zona Intertidal Menggunakan Citra Synthetic Aperture Radar (SAR) Sentinel-1. Other thesis, Institut Teknologi Sepuluh Nopember.
![[thumbnail of 5016211106_Undergraduated_Thesis.pdf]](http://repository.its.ac.id/style/images/fileicons/text.png) |
Text
5016211106_Undergraduated_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (6MB) | Request a copy |
Abstract
Pemahaman terhadap kedalaman perairan pesisir memiliki peran penting dalam kegiatan pemetaan batimetri, navigasi, serta pemantauan dinamika lingkungan pantai. Namun, survei kedalaman secara konvensional sering terkendala oleh akses yang sulit, biaya operasional yang tinggi, dan keterbatasan waktu. Untuk menjawab tantangan tersebut, studi ini menerapkan model algoritma Artificial Neural Network (ANN) guna memprediksi kedalaman laut berdasarkan data citra radar Sentinel-1 SAR. ANN dipilih karena kemampuannya mengenali
hubungan kompleks antara sinyal radar dan variasi karakteristik perairan. Proses training dilakukan dengan mengombinasikan citra Sentinel-1 SAR dan data kedalaman dari data BATNAS. Model yang digunakan berbasis Multilayer Perceptron (MLP) dengan penerapan KFold Cross Validation untuk memastikan kestabilan dan akurasi hasil prediksi. Hasil evaluasi model arsitektur terbaik ANN menunjukkan nilai Root Mean Square Error (RMSE) sebesar 8,7060 meter dan koefisien determinasi (R²) sebesar 0,0838. Akurasi model bervariasi terhadap kedalaman, dengan performa terbaik pada rentang 10–15 meter dan cenderung meningkat hingga kedalaman maksimum. Sebaliknya, prediksi pada zona sangat dangkal (<5 meter) menunjukkan akurasi rendah, yang kemungkinan disebabkan oleh noise tinggi pada sinyal radar dan keterbatasan jumlah data latih. Hasil ini memberikan informasi bahwa pendekatan ANN berbasis citra radar satelit dapat menjadi alternatif yang cukup efektif dalam estimasi kedalaman, khususnya di wilayah yang sulit dijangkau oleh metode survei langsung. Pengembangan lebih lanjut dapat diarahkan pada eksplorasi fitur tambahan dan integrasi dengan algoritma machine learning lainnya untuk meningkatkan akurasi prediksi kedalaman.
=====================================================================================================================================
Understanding coastal water depth plays a crucial role in bathymetric mapping, navigation, and monitoring of coastal environmental dynamics. However, conventional depth surveys often face challenges such as limited accessibility, high operational costs, and time constraints. To address these issues, this study implements an Artificial Neural Network (ANN) algorithm to predict sea depth based on Sentinel-1 SAR radar imagery. ANN was selected for its ability to capture complex relationships between radar signals and the varying characteristics of water bodies. The training process involved combining Sentinel-1 SAR data with depth measurements from the BATNAS dataset. The model architecture is based on a Multilayer Perceptron (MLP) and employs K-Fold Cross Validation to ensure the stability and accuracy of predictions. Evaluation of the best-performing ANN architecture yielded a Root Mean Square Error (RMSE) of 8.7060 meters and a coefficient of determination (R²) of 0.0838. The model’s accuracy varied by depth, with the best performance observed in the 10–15 meters range and a tendency to improve with increasing depth. Conversely, predictions in very shallow zones (< 5 meter) showed lower accuracy, likely due to high radar signal noise and limited training data in those areas. These findings indicate that ANN-based approaches using satellite radar imagery can serve as a promising alternative for depth estimation, particularly in areas that are difficult to access through direct survey methods. Further development may include the exploration of additional features and integration with other machine learning algorithms to enhance the accuracy of bathymetric modeling.
| Item Type: | Thesis (Other) |
|---|---|
| Uncontrolled Keywords: | Multilayer Perceptron, Satellite Derived Bathymetry, Penginderaan Jauh, Kedalaman Pesisir, Life Below Water ======================================================================================================================== Multilayer Perceptron, Satellite Derived Bathymetry, Remote Sensing, Coastal Bathymetry, Life Below Water |
| Subjects: | G Geography. Anthropology. Recreation > G Geography (General) > G70.5.I4 Remote sensing |
| Divisions: | Faculty of Civil, Planning, and Geo Engineering (CIVPLAN) > Geomatics Engineering > 29202-(S1) Undergraduate Thesis |
| Depositing User: | Belvana Eka Putri |
| Date Deposited: | 25 Jul 2025 07:24 |
| Last Modified: | 25 Jul 2025 07:24 |
| URI: | http://repository.its.ac.id/id/eprint/121965 |
Actions (login required)
 |
View Item |