An example of fishery yield predictions from VMS-based navigational characteristics applied to double trawlers in China

Abstract

Vessel monitoring systems (VMSs) provide information on the spatial and temporal distribution of fishing vessels operating at sea and are mostly used for fisheries enforcement and vessel surveillance. VMS has potential shortcomings in fisheries resource assessment due to its inability to acquire information on catch yields. This study proposed a method for predicting fishing vessel yield based on VMS data. Speed, heading difference and track characteristics were used to determine the state of VMS data for three double trawlers (Zhepuyu 71319, 71528 and 71568) in 2021. Then, fishing effort and temporal and spatial features were calculated for each trip based on the VMS data of the dragging state. Finally, six machine learning (ML) models (k-Nearest Neighbor[KNN], Random Forest[RF], Support Vector Machine[SVM], eXtreme Gradient Boosting[XGBoost], Gradient Boosting Decision Tree[GBDT] and Ridge Regression[RR]) were used to predict the yield of the double trawlers per trip. Two of the three double trawlers were used in turn as the training set to construct the prediction model, and the other one was utilised as the test set to evaluate the model performance. The prediction results were evaluated using Root Mean Square Error (RMSE), Mean Absolute Error (MAE), Coefficient of determinant (R2) and model prediction error per trip. The prediction error of the total yield per year was calculated. Results showed that the six ML models exhibited high prediction performance and affirmed the accuracy of yield prediction, and the performance of the models varied with the dataset. XGBoost, GBDT and KNN had the best prediction performance when Zhepuyu 71319, 71528 and 71568 were used as the test set, respectively. Moreover, the total yield errors of all models were less than 10 %. This study selected multiple features to construct ML models to predict the yield of double trawlers based on the navigation characteristics of double trawlers and further explored the VMS data to improve the assessment methods of fishery resources, enabling the VMS data to play a greater role in aiding scientific decisions.