Analysis of fuzzy logic systems types 1 and 2 in identifying of IUU fishing and transshipment: a case study in Indonesia’s vulnerable waters

Abstract

The identification system for illegal, unreported, and unregulated (IUU) fishing and transshipment of incidents is one effort as early warning and it can be adopted by the Indonesian ministry of fisheries and marine affairs. The system has been developed using several methods. One of them is the artificial neural networks (ANN) have been used, but it is several weaknesses when AIS data is lost. The system able to identify in missing data in a short duration of time only. The development of a system for IUU identification proposed using a fuzzy logic system (FLS). The IUU identification system consists of subsystems: (i) missing data prediction, (ii) anomaly trajectory decision, (iii) IUU selection, and (iv) IUU decision. The performance of FLS is compared between fuzzy type 1 and type 2. The system tested using data of IUU occurrences waters in high frequency. Both SLF type 1 and type 2 are able to give result in very good accuracy for IUU fishing and transshipment identification.

Detecting communities at high-risk of IUU fishing: networks of shadow encounters in Area 81 of the Western Central Pacific

“…The automatic detection algorithm classifies vessels as interacting when they are positioned within 500 meters of each other and are at least 10 kilometers from a coast. Despite the potential for missing data (Masroeri et al, 2021;Kumar et al, 2022;Masroeri et al, 2022), GFW AIS data is arguably one of the most complete sources of information used at local and global scales to investigate potential IUU fishing and at-sea transshipments [e.g (Mazzarella et al, 2014;Miller et al, 2018;Purivigraipong, 2018;Welch et al, 2022)]. Step 1 involved identifying all encounters between 21 carriers and 141 fishing vessels observed to interact within the FAO Area 81 during the study period.…”

“…To extend this investigation, future research could evaluate potential change mechanisms, such as preferential attachment, more directly with dynamic multivariate stochastic actor-oriented modeling to add greater statistical rigor to prediction models [e.g (Kalish, 2020)]. Also, machine learning and other strategies can be used to impute illegal harvesting and missing transshipment encounters, i.e., detect unobserved events (Masroeri et al, 2021;Kumar et al, 2022;Masroeri et al, 2022). This would allow conservation researchers to test the robustness of findings under different scenarios of data completeness.…”