Show simple item record

dc.contributor.advisorMushthofa
dc.contributor.authorWIRAGAMA, TAN BIMA
dc.date.accessioned2026-05-07T01:04:00Z
dc.date.available2026-05-07T01:04:00Z
dc.date.issued2026
dc.identifier.urihttp://repository.ipb.ac.id/handle/123456789/173047
dc.description.abstractDinamika oseanografi di WPP-NRI 716 menuntut teknologi deteksi yang presisi untuk mendukung pengelolaan sumber daya perikanan. Penelitian ini bertujuan menganalisis karakteristik fisik upwelling dan mengembangkan model deteksi berbasis deep learning. Metode yang diterapkan mengintegrasikan analisis time-series data suhu permukaan laut dan konsentrasi klorofil-a (2007-2017) dengan arsitektur hybrid CNN-BiLSTM yang dioptimasi otomatis menggunakan algoritma Heap-Based Optimizer (HBO). Hasil penelitian mengungkap adanya pola musiman upwelling yang memuncak pada Musim Barat (Desember-Januari) , serta jeda waktu respons biologis selama tiga minggu. Secara komputasional, integrasi HBO terbukti efektif mengatasi kendala penentuan parameter manual, menghasilkan performa yang cukup baik dengan akurasi 96,3% dan f1-score 92,2%. Model ini menunjukkan sensitivitas tinggi dengan kemampuan mendeteksi 98,4% kejadian upwelling intensitas high. Karakteristik model yang peka dan konservatif menjadikannya instrumen peringatan dini yang cukup andal untuk meminimalkan risiko kegagalan deteksi di lapangan.
dc.description.abstractOceanographic dynamics in WPP-NRI 716 require precise detection technology to support fisheries resource management. This study aims to analyze the physical characteristics of upwelling and develop a deep learning-based detection model. The applied method integrates time-series analysis of sea surface temperature and chlorophyll-a concentration data (2007-2017) with a hybrid CNN-BiLSTM architecture automatically optimized using the Heap-Based Optimizer (HBO) algorithm. The results reveal a strong negative correlation with seasonal patterns peaking in the West Monsoon, as well as a biological response lag time of three weeks. Computationally, HBO integration proves effective in overcoming manual parameter tuning constraints, yielding quite good performance with 96.3% accuracy and an f1-score of 0.922. The model demonstrates high sensitivity, with the capability to detect 98.4% of high-intensity upwelling events. The model's sensitive and conservative characteristics make it a reliable early warning instrument to minimize the risk of detection failure in the field.
dc.description.sponsorship
dc.language.isoid
dc.publisherIPB Universityid
dc.titleAnalisis dan Deteksi Upwelling di WPP-NRI 716 Menggunakan CNN-BiLSTM dengan Optimasi Heap-Based Optimizerid
dc.title.alternativeAnalysis and Detection of Upwelling in WPP-NRI 716 Using CNN-BiLSTM with Heap-Based Optimizer Optimization
dc.typeSkripsi
dc.subject.keywordCNN-BiLSTMid
dc.subject.keywordHeap-Based Optimizerid
dc.subject.keywordLag Timeid
dc.subject.keywordUpwellingid
dc.subject.keywordWPP-NRI 716id


Files in this item

Thumbnail
Thumbnail
Thumbnail

This item appears in the following Collection(s)

Show simple item record