Abstract
Abstrak Salah satu instrumen pengurangan risiko bencana longsor adalah peta kerawanan longsor yang dihasilkan dari pemodelan spasial. Alur kerja pemodelan spasial kerawanan longsor menggunakan model pembelajaran mesin belum terakomodasi dalam Standar Nasional Indonesia (SNI) yang berlaku saat ini (tahun 2024). Penelitian ini berusaha menjelaskan variasi langkah-langkah dalam alur kerja pembelajaran mesin dan menunjukkan perbedaannya dengan alur kerja model statistik. Model statistik regresi logistik dan model pembelajaran mesin random forest dipilih untuk menjelaskan perbedaan alur kerja pemodelan spasial kerawanan longsor. Formulasi alur kerja pemodelan spasial kerawanan longsor diterapkan untuk memetakan kerawanan longsor di Kabupaten Pacitan. Pada tanggal 27-29 November 2017, 743 longsor terjadi di Kabupaten Pacitan dipicu oleh hujan yang sangat lebat akibat Siklon Tropis Cempaka. Kabupaten Pacitan merupakan salah satu wilayah rawan longsor di Provinsi Jawa Timur. Alur kerja pemodelan spasial kerawanan longsor terbagi atas beberapa langkah yaitu penyiapan data, pra-pemrosesan data (pre-processing), melatih dan menyetel model, memvalidasi model, pemodelan spasial, dan uji akurasi. Hasil uji akurasi model RF dan LR yang diterapkan di Kabupaten Pacitan masing-masing sebesar 0,75 dan 0,73. Penelitian ini diharapkan dapat memberikan masukan dalam penyusunan SNI pemetaan kerawanan longsor di masa mendatang serta dapat digunakan sebagai acuan dalam pemetaan kerawanan longsor secara umum di Indonesia. Abstract One of the landslide risk reduction instruments is landslide susceptibility maps which can be produced by spatial modeling. The landslide susceptibility modeling based on machine learning workflows have not been accommodated in the current verison of Indonesian National Standard (SNI). This study seeks to explain the variation of machine learning workflows and show how they differ from statistical learning workflows. Logistic regression model and random forest machine learning models were selected to explain variations in landslide susceptibility modeling workflows. The modeling workflows were applied to map landslide susceptibility in Pacitan Regency. On 27-29 November 2017, 743 landslides occurred in Pacitan Regency triggered by very heavy rain due to Tropical Cyclone Cempaka. Pacitan Regency is one of the landslide-prone areas in East Java Province. The landslide susceptibility modeling workflow is divided into several steps, i.e. data preparation, data pre-processing, training and tuning the model, validating the model, spatial modeling, and accuracy testing. The accuracy test results of the RF and LR models applied in Pacitan Regency were 0.75 and 0.73 respectively. This research is expected to provide a benchmark for landslide susceptibility mapping in Indonesia.
Published Version
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callSimilar Papers
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
