Identifikasi Potensi Bencana Banjir Menggunakan Data Satelit Gayaberat Gravity Recovery and Climate Experiment Follow On (Studi Kasus : Pulau Jawa)

Abida, Azza Nafis (2026) Identifikasi Potensi Bencana Banjir Menggunakan Data Satelit Gayaberat Gravity Recovery and Climate Experiment Follow On (Studi Kasus : Pulau Jawa). Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5016221021-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (8MB) | Request a copy

Abstract

Banjir merupakan bencana hidrometeorologi yang paling sering terjadi dan berdampak signifikan di Pulau Jawa. Tingginya curah hujan, urbanisasi cepat, serta penurunan muka tanah meningkatkan risiko banjir di wilayah tersebut. Pemantauan perubahan simpanan air terestrial (Terrestrial Water Storage/TWS) menjadi penting untuk memahami dinamika massa air yang berkaitan dengan kejadian banjir. Satelit GRACE-FO menyediakan data anomali TWS berskala global, namun resolusi spasialnya yang relatif kasar (0,5°) menjadi kendala untuk analisis detail pada skala regional. Penelitian ini bertujuan untuk meningkatkan resolusi spasial data GRACE-FO melalui teknik downscaling berbasis algoritma machine learning XGBoost dengan memanfaatkan variabel hidrologi dari GLDAS Noah pada resolusi 0,25°. Data hasil downscaling kemudian digunakan untuk menghitung Flood Potential Index (FPI) dalam mengidentifikasi potensi banjir di Pulau Jawa pada periode 2019–2025. Hasil evaluasi model menunjukkan performa yang sangat baik dengan nilai Correlation Coefficient (CC) sebesar 0,997 pada training set dan 0,959 pada testing set, nilai Nash–Sutcliffe Efficiency (NSE) sebesar 0,993 pada training set dan 0,916 pada testing set, serta nilai Normalized Root Mean Square Error (RMSE*) sebesar 0,083 pada training set dan 0,290 pada testing set. Dinamika temporal antara TWSA₀,₅ dan hasil prediksi TWSAR_0,25 menunjukkan akurasi tinggi dengan nilai CC sebesar 0,997, NSE sebesar 0,974, dan RMSE* sebesar 0,161. Validasi menggunakan data anomali Tinggi Muka Air (TMA) di Pos Duga Sungai Citanduy dan Bengawan Solo menunjukkan bahwa TWSA hasil downscaling resolusi 0,25° mampu mempertahankan dan meningkatkan korelasi dibandingkan resolusi asli 0,5°.Hasil analisis FPI menunjukkan kesesuaian dengan beberapa kejadian banjir di Pulau Jawa, seperti banjir Jabodetabek pada Maret 2025 dan Februari 2021, Kudus pada Januari 2023, Sidoarjo pada Januari 2025, serta Jombang pada Februari 2021. Penelitian ini menunjukkan bahwa kombinasi TWSA hasil downscaling dan data presipitasi GLDAS mampu memberikan representasi kondisi hidrologi yang lebih detail dalam mengidentifikasi potensi banjir di Pulau Jawa.
=====================================================================================================================================
Flood is one of the most frequent hydrometeorological disasters and has significant impacts on Java Island. High rainfall intensity, rapid urbanization, and land subsidence have increased flood risk in various regions. Monitoring changes in terrestrial water storage (TWS) is essential to understand water mass dynamics related to flood events. The GRACE-FO satellite provides global-scale Terrestrial Water Storage Anomaly (TWSA) data; however, its relatively coarse spatial resolution (0,5°) limits detailed regional-scale analysis. This study aims to improve the spatial resolution of GRACE-FO data through a downscaling technique based on the XGBoost machine learning algorithm using hydrological variables from GLDAS Noah at a spatial resolution of 0,25°. The downscaled data were subsequently used to calculate the Flood Potential Index (FPI) for identifying flood potential in Java Island during the 2019–2025 period. The model evaluation results indicate excellent performance, with Correlation Coefficient (CC) values of 0,997 for the training set and 0,959 for the testing set, Nash–Sutcliffe Efficiency (NSE) values of 0,993 for the training set and 0,916 for the testing set, and Root Normalized Mean Square Error (RMSE*) values of 0,083 for the training set and 0,290 for the testing set. The temporal dynamics between TWSA₀,₅ and the predicted TWSAR_0,25 also demonstrated high accuracy, with CC, NSE, and RMSE* values of 0,997, 0,974, and 0,161, respectively. Validation using Water Level Anomaly (WLA) data from the Citanduy and Bengawan Solo river gauge stations showed that the 0.25° downscaled TWSA improved the correlation at the Citanduy station while maintaining a comparable correlation to the original 0.5° resolution TWSA at the Bengawan Solo station. The FPI analysis results were consistent with several flood events in Java Island, including the Jabodetabek floods in March 2025 and February 2021, Kudus in January 2023, Sidoarjo in January 2025, and Jombang in February 2021. This study demonstrates that the combination of downscaled TWSA and GLDAS precipitation data can provide a more detailed representation of hydrological conditions for identifying flood potential in Java Island.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	GRACE-FO, Terrestrial Water Storage, downscaling, XGBoost, GLDAS Noah, Flood Potential Index (FPI), banjir, GRACE-FO, Terrestrial Water Storage, downscaling, XGBoost, GLDAS Noah, Flood Potential Index (FPI), flood
Subjects:	G Geography. Anthropology. Recreation > GB Physical geography > GB1399.2 Flood forecasting. G Geography. Anthropology. Recreation > GB Physical geography > GB1399.9 Floods T Technology > TD Environmental technology. Sanitary engineering > TD171.75 Climate change mitigation
Divisions:	Faculty of Civil, Planning, and Geo Engineering (CIVPLAN) > Geomatics Engineering > 29202-(S1) Undergraduate Thesis
Depositing User:	Azza Nafis Abida
Date Deposited:	21 Jul 2026 02:41
Last Modified:	21 Jul 2026 02:41
URI:	http://repository.its.ac.id/id/eprint/135904

Actions (login required)

View Item