Integrasi Data Oseanografi, Logbook Penangkapan Ikan, dan Kecerdasan Buatan untuk Memprediksi Zona Potensi Penangkapan Ikan Decapterus spp. di Perairan Tropis Samudra Hindia Bagian Tenggara

Lupi Nugraheni, Prof.Dr.Ir. Djumanto, M.Sc.; Dr.rer.nat. Riza Yuliratno Setiawan, S.Kel.,M.Sc.

2026 | Tesis | S2 ILMU PERIKANAN

Ikan layang (Decapterus spp.) merupakan ikan pelagis kecil di perairan tropis yang distribusinya dipengaruhi oleh dinamika oseanografi. Penelitian ini bertujuan memprediksi zona potensi penangkapan ikan layang di perairan tropis Samudra Hindia bagian tenggara melalui analisis variabel oseanografi dan logbook penangkapan menggunakan pembelajaran mesin berbasis random forest regresi. Variabel oseanografi diperoleh dari Aqua MODIS, HYCOM + NCODA, Carbon-based Production Model-2 (CbPM2), dan Copernicus Marine Environment Monitoring Service (CMEMS). Seluruh variabel diselaraskan ke resolusi 0,04° serta diuji multikolinearitas, sedangkan data tangkapan ditransformasi menggunakan log(tangkapan+1) agar karakteristik distribusinya lebih normal. Parameter model random forest (ntree dan mtry) dioptimalkan, dengan pembagian data 80% pelatihan dan 20% pengujian. Hasil menunjukkan adanya zona potensi penangkapan ikan layang di dua lokasi utama penelitian, yakni perairan barat Sumatra (4°–6°LS; 98°–101°BT) dengan potensi tertinggi pada Mei–Agustus yang dipengaruhi oleh anomali permukaan laut 0,02–0,2 m dan klorofil-a 0,05–0,15 mg/m³, serta perairan selatan Jawa (9°–12°LS; 110°–114°BT) dengan puncak potensi Maret–Juli yang dipengaruhi oleh salinitas 32,9–34,36 psu, pH sekitar 8,04, oksigen terlarut 6,28–6,95 mg/L, dan klorofil-a hingga 0,30 mg/m³.

Scads (Decapterus spp.) are small pelagic fish in tropical waters whose distribution is strongly influenced by marine oceanographic dynamics. This study aims to predict potential fishing zones for scads in the southeastern tropical Indian Ocean by analyzing oceanographic variables and fishing logbook records using machine learning with a random forest regression approach. Oceanographic variables were obtained from Aqua MODIS, HYCOM + NCODA, the Carbon-based Production Model-2 (CbPM2), and the Copernicus Marine Environment Monitoring Service (CMEMS). All oceanographic variables were standardized to a 0,04° resolution and tested for multicollinearity, while catch data were transformed using log(catch+1) to approximate a normal distribution. The random forest model parameters (ntree and mtry) were optimized, with data partitioned into 80% training and 20% testing. The results identified potential scads fishing zones in two primary regions, the western waters of Sumatra (4°–6°S; 98°–101°E), with peak potential during May–August influenced by sea level anomalies of 0,02–0,2 m and chlorophyll-a concentrations of 0,05–0,15 mg/m³; and the southern waters of Java (9°–12°S; 110°–114°E), where the peak potential occurred from March to July, associated with salinity levels of 32,9–34,36 PSU, pH around 8,04, dissolved oxygen of 6,28–6,95 mg/L, and chlorophyll-a up to 0,30 mg/m³.

Kata Kunci : anomali permukaan laut, ikan layang, random forest, salinitas, zona potensi penangkapan ikan, sea level anomaly, scad, salinity, potential fishing zone