Pemetaan Luas Genangan Banjir Rob Berdasarkan Peramalan Muka Air Laut di Kabupaten Gresik Menggunakan Quantile Regression Neural Network

Rofananda, Shafira Husna (2025) Pemetaan Luas Genangan Banjir Rob Berdasarkan Peramalan Muka Air Laut di Kabupaten Gresik Menggunakan Quantile Regression Neural Network. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5006211001-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (3MB) | Request a copy

Abstract

Banjir rob yang sering terjadi di Kabupaten Gresik disebabkan oleh naiknya permukaan air laut dan berdampak pada kerusakan lingkungan, infrastruktur, serta kerugian ekonomi masyarakat. Oleh karena itu, diperlukan mitigasi yang tepat. Penelitian ini bertujuan untuk meramalkan tinggi muka air laut dan menghitung luas genangan banjir rob sebagai bagian dari upaya mitigasi bencana. Metode yang digunakan adalah Quantile Regression Neural Network (QRNN), yang mampu memodelkan distribusi data target, bukan hanya rata-ratanya, sehingga dapat menghasilkan prediksi yang lebih akurat untuk kondisi ekstrem. Data yang digunakan meliputi data tinggi muka air laut 1 Januari 2020 hingga 31 Desember 2024, Digital Elevation Model (DEM), dan batas wilayah dalam format shapefile. Tahapan penelitian mencakup pengumpulan dan pra-pemrosesan data, analisis pola musiman, pembangunan model QRNN, peramalan tinggi muka air laut, pemetaan genangan, dan perhitungan luas genangan. Hasil eksplorasi menunjukkan rata-rata tinggi muka air laut sebesar 59,98 cm, dengan nilai minimum -70 cm dan maksimum 150 cm. Distribusi data bersifat miring ke kiri (skewness negatif) dan lebih lebar dari distribusi normal (kurtosis negatif). Analisis periodogram menunjukkan pola semi-diurnal dengan periode dominan sekitar 12 dan 24 jam. Pengelompokan data berdasarkan kalender Hijriah memperlihatkan pola musiman bulanan yang lebih konsisten dibandingkan kalender Masehi. Pada kalender Hijriah, tampak adanya variasi yang berulang setiap dua dan empat minggu, serta pola ini cenderung sama antar tahun. Model QRNN terbaik menggunakan lag 1–24 dan 50 neuron di hidden layer. Model ini menghasilkan Median Absolute Error (MdAE) sebesar 2,41 cm untuk data training dan 2,59 cm untuk data testing. Nilai yang hampir sama menunjukkan bahwa model tidak mengalami overfitting. Hasil peramalan untuk Januari 2025 menunjukkan tinggi muka air laut maksimum diperkirakan mencapai 134,14 cm. Hasil pemetaan genangan menunjukkan bahwa desa Pangkahwetan di Kecamatan Ujungpangkah merupakan wilayah dengan genangan paling luas. Kecamatan yang terdampak genangan terbesar adalah Ujungpangkah, Manyar, dan Bungah, sedangkan genangan banjir rob terbesar terjadi di Ujungpangkah, Manyar, dan Kebomas. Penelitian ini merekomendasikan mitigasi berupa larangan pembangunan di zona rawan, pembangunan tanggul dan drainase, restorasi pesisir, serta sosialisasi rutin kepada masyarakat. Penelitian selanjutnya disarankan menambahkan variabel penanggalan Hijriah dan menggunakan data spasial dengan resolusi lebih tinggi agar hasil pemetaan lebih sesuai dengan kondisi sebenarnya.
========================================================================================================================
Tidal flooding (banjir rob) in Gresik Regency is caused by rising sea levels and has significant impacts, including environmental damage, infrastructure deterioration, and economic losses for local communities. Therefore, effective mitigation efforts supported by accurate predictions and flood mapping are necessary. This research applies the Quantile Regression Neural Network (QRNN) to predict sea level more accurately than traditional models. QRNN combines Neural Networks and Quantile Regression, allowing the model to capture the full distribution of the target variable rather than just its average. This research aims to forecast sea level and estimate the inundation area of tidal flooding in Gresik Regency as part of disaster mitigation efforts. The data used include tidal measurements from January 1, 2020, to December 31, 2024, a Digital Elevation Model (DEM), and administrative boundaries in shapefile format. The research stages consist of data collection and preprocessing, seasonal pattern analysis, QRNN model construction, sea level forecasting, flood mapping, and inundation area calculation. Exploratory analysis showed that sea level heights had an average of 59.98 cm, with a minimum of -70 cm and a maximum of 150 cm. The data distribution showed negative skewness and kurtosis, indicating most values were above the mean and the spread was wider than a normal distribution. Periodogram analysis revealed a semi-diurnal pattern with dominant periods of around 12 and 24 hours. Grouping the data using the Hijri calendar showed more consistent seasonal variations compared to the Gregorian calendar. Seasonal patterns tended to recur every two and four weeks and were relatively stable from year to year. The best-performing QRNN model used lags 1–24 and 50 neurons in the hidden layer, producing a Median Absolute Error (MdAE) of 2.41 cm for training data and 2.59 cm for testing data, indicating no overfitting. Forecasting for January 2025 predicted a maximum sea level of 134.14 cm, which was used as the basis for flood mapping in QGIS. The mapping results indicated that Pangkahwetan Village in Ujungpangkah District was the largest inundated area. Overall, the most affected sub-districts were Ujungpangkah, Manyar, and Bungah for general inundation, and Ujungpangkah, Manyar, and Kebomas for tidal flooding. This research recommends mitigation strategies including restrictions on development in flood-prone areas, the construction of sea walls and drainage systems, coastal restoration, and regular community outreach. Future research is suggested to use higher-resolution spatial data and add Hijri calendar variables to improve prediction accuracy and mapping detail.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Analisis Spasial, Banjir Rob, Kabupaten Gresik, QRNN, Tinggi Muka Air Laut, Gresik Regency, QRNN, Sea Level, Spatial Analysis, Tidal Floods
Subjects:	Q Science > QA Mathematics > QA336 Artificial Intelligence Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science)
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Actuaria > 94203-(S1) Undergraduate Thesis
Depositing User:	Shafira Husna Rofananda
Date Deposited:	24 Jul 2025 03:18
Last Modified:	24 Jul 2025 03:18
URI:	http://repository.its.ac.id/id/eprint/120932

Actions (login required)

View Item