Identifikasi Area Terdampak Oil Seep Di Darat Dari Data Foto Udara Menggunakan Metode Convolutional Neural Networks

Alya, Nurul Fitri (2021) Identifikasi Area Terdampak Oil Seep Di Darat Dari Data Foto Udara Menggunakan Metode Convolutional Neural Networks. Undergraduate thesis, Institut Teknologi Sepuluh Nopember.

[thumbnail of 03311740000072-Project_Report.pdf] Text
03311740000072-Project_Report.pdf - Accepted Version
Restricted to Repository staff only until 1 October 2023.

Download (3MB) | Request a copy

Abstract

Rembesan minyak merupakan salah satu peristiwa yang merugikan lingkungan pada industri minyak dan gas. Hal ini dikarenakan senyawa kimia yang terkandung pada rembesan minyak dapat mengakibatkan penurunan kualitas lingkungan hidup. Rembesan minyak (oil seep) tidak hanya terjadi di wilayah perairan, tetapi juga di daratan, yang terserap oleh tanah. Kejadian ini dapat mengindikasikan adanya sistem perminyakan di bawah permukaan tanah.Dalam penelitian ini daerah terdampak rembesan minyak diidentifikasi menggunakan metode deep learning dengan Convolutional Neural Networks dimana mesin diharapkan meniru sistem kerja otak manusia dalam mengidentifikasi objek. Penelitian terdahulu yang dilakukan oleh Ozigiz., dkk. 2018, Csilik., dkk. 2018, and Cantorna., dkk. 2019 menghasilkan performa CNN yang baik dalam penggunaan CNN dalam mengidentifikasi tumpahan minyak maupun klasifikasi tutupan lahan. Data foto udara yang telah terorthorektifikasi dilakukan proses pelabelan objek yang akan dikaji, dalam penelitian ini yakni area terdampak oil seep. Training data tersebut menjadi data masukan pada tahap train deep learning model, dan akan dilakukan proses klasifikasi piksel untuk mendeteksi area terdampak oil seep. Hasil pengolahan berupa tingkat akurasi model mencapai 91,75% dan akurasi hasil pendeteksian berupa raster yang menampilkan area terdampak oil seep mencapai akurasi keseluruhan 90%. Hasil pendeteksian kemudian diubah menjadi poligon untuk dihitung luasan areanya, dan menghasilkan perhitungan area terdampak oil seep seluas ±3,87 hektar. Perbandingan dengan data maupun metode selain CNN diperlukan pada penelitian di masa depan untuk menguji performa CNN dalam pendeteksian area terdampak oil seep di darat.

=====================================================================================================

Oil seep is one of incident that can give negative impact to environment. This because of the chemical compounds contained in oil seep can lead to decrease in environmental quality. Oil seep does not only occur in water, but also on land which is absorbed by the soil. This incident could also indicate the presence of a subsurface petroleum system. In this study, areas affected by oil seep on land were detected using a deep learning method, Convolutional Neural Networks where the machine was expected to learn and imitate the working system of human brain in identifying objects of an image. Previous studies (Ozigiz., et al. 2018, Csilik., et al. 2018, and Cantorna., et al. 2019) proved that CNN performed well to detect oil spill and land cover classification. Aerial imagery is used in this study. The data then be labelled to give examples to the machine which one is the affected areas. Labelle object then becomes training data for deep learning and as the input data for training the deep learning model. The model will be carried out to detect areas affected by oil seep with pixel classification. The processing result in the form of a model with 91,75% in overall accuracy and a raster with 90% overall accuracy showing the area affected. The raster then exported to polygon to calculate the area, result in ±3,87 hectares. Data and methods comparation is needed for future studies to test the CNN performance to detect oil seep impact on land area.

Item Type: Thesis (Undergraduate)
Uncontrolled Keywords: Convolutional Neural Networks, Oil Seep, Deep Learning, Rembesan Minyak, Foto Udara, Fotogrametri
Subjects: T Technology > TA Engineering (General). Civil engineering (General) > TA1637 Image processing--Digital techniques
Divisions: Faculty of Civil, Planning, and Geo Engineering (CIVPLAN) > Geomatics Engineering > 29202-(S1) Undergraduate Thesis
Depositing User: Nurul Fitri Alya
Date Deposited: 18 Aug 2021 03:09
Last Modified: 24 Mar 2022 07:57
URI: http://repository.its.ac.id/id/eprint/87743

Actions (login required)

View Item View Item