Standardizing catch per unit effort by machine learning techniques in longline fisheries: a case study of bigeye tuna in the Atlantic Ocean

Yang, Shenglong; Yang, Dongfang; Fan, Wei; Shi, Huiming

doi:10.1590/s2675-28242020068226

Cited by 5 publications

(3 citation statements)

References 28 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…It maps the independent variables to a high-dimensional feature space through a nonlinear mapping and finds an optimal classification surface in the high-dimensional feature space such that the error obtained from this optimal classification surface is minimized for all training samples (Zan et al, 2004). SVM has been used to describe fishery processes such as flow prediction (Asefa et al, 2006), hydroacoustic classification of fish populations (Bosch et al, 2013), CPUE normalization (Yang et al, 2020) and fish species classification (Morris et al, 2001).…”

Section: Support Vector Machinementioning

confidence: 99%

The use of spaceborne lidar to map Antarctic krill distributions and biomass in the Southern Ocean

Zhong,

Chen,

Zhang

et al. 2024

Front. Mar. Sci.

View full text Add to dashboard Cite

The Antarctic krill is a pivotal species in the Southern Ocean ecosystem, primarily due to its extraordinary nutritional content and plentiful resources. Studying the distribution of these resources and their environmental impact factors is crucial for the successful development of Antarctic krill fisheries. Traditional methodologies such as acoustic measurements, however, often face limitations in their capacity to provide a comprehensive and uninterrupted assessment. Moreover, the six-month duration of polar nights in polar regions presents significant challenges for traditional satellite observations. In this context, LiDAR, an active remote sensing observation method, offers a promising alternative. Known for their high resolution, flexibility, and efficiency, LiDAR systems can obtain detailed information on diurnal ocean parameters in polar regions on a vast scale and in a systematic way. Our study utilizes the spaceborne LiDAR system, CALIPSO, to successfully attain continuous Antarctic krill CPUE over the past decade, using various models such as the generalized linear model (GLM), artificial neural network (ANN), and support vector machine (SVM). A comparative analysis of the prediction results reveals that while both ANN and SVM models outperform the GLM, the SVM’s prediction capabilities are somewhat unstable. Our findings reveal CALIPSO’s potential in overcoming challenges associated with traditional satellite observations during polar winters. In addition, we found no obvious pattern of interannual variation in krill CPUE, with high values predominantly occurring from February to May. This suggests that krill is mainly located around the South Shetland Islands during January-April, before moving offshore towards South Georgia in May-June. A substantial krill aggregation community is found in the South Atlantic waters, indicating high potential for krill fishing. The optimum mix layer depth range for high krill CPUE is 270-390 m, with a chlorophyll concentration of approximately 0.1 mg m-3. The optimum sea surface temperature range is between -1.4-5.5°C, and the sea ice coverage range is approximately 0-0.1×106 km2. The predicted Antarctic krill bioresource has risen from 2.4×108 tons in 2011 to 2.8×108 tons in 2020. This increase in krill biomass aligns with the biomass of krill assessed by CCAMLR.

show abstract

Section: Support Vector Machinementioning

confidence: 99%

The use of spaceborne lidar to map Antarctic krill distributions and biomass in the Southern Ocean

Zhong,

Chen,

Zhang

et al. 2024

Front. Mar. Sci.

View full text Add to dashboard Cite

show abstract

“…Fishermen are people or individuals who work in exploiting potential resources from the sea and the fishing profession has the characteristics of work that is dangerous, dirty and difficult [7]. Geographically, Indonesia consists of 2/3 (two-thirds) of the seas of the entire territory of Indonesia which covers 6.32 million square kilometers (km²), with a total of 17,504 islands, and Indonesia has the largest Long Line in the world and after Canada, which is 99,093 km² , with the condition of the territory being 2/3 of the entire territory of Indonesia, most of Indonesia's borders with surrounding countries are in waters [8]…”

Section: Fishermanmentioning

confidence: 99%

Implementation Of Linear Regression Algorithm And Support Vector Regression In Building Prediction Models Fish Catches Of Fishermen In Ciparagejaya Village

Mahendra,

Mutoi Siregar,

Ahmad Baihaqi

et al. 2023

ECSIT

View full text Add to dashboard Cite

Fish catch is one of the indicators affecting the economic growth of coastal communities including the Ciparagejaya Village Community, fish catches recorded by the Fish Auction Place (TPI) vary every month, this is due to the unpredictable condition of the fish caught, for fishermen caught from sea fish are the main source of income, so a reference is needed to anticipate a decrease in fish catches in determining a strategy for sharing the results of savings that are deducted every day from fishermen's catches. The purpose of this study was to create a prediction model with the Linear Regression Algorithm and Support Vector Regression (SVR) from data recorded by TPI Ciparagejaya Village, the data consisting of 33 types of fish caught in 2021. The method used in this research is an analytical method using Linear Regression Algorithm and SVR. This research produces a Prediction Model which will be a reference in the process of calculating data accuracy values where in this study the Root Mean Squared Error (RMSE) method is used. Tests were carried out using Microsoft excel and python with the smallest RMSE value from Microsoft excel calculations of 0.577735, and from python calculations, the smallest RMSE value is 0.

show abstract

“…During the past few decades advances in SVM algorithms were developed to support the classification and regression of linear and non-linear data. Some authors have applied SVM technique to describe fisheries processes such as stream flow forecasting (Asefa et al, 2006), hydroacoustic classification of fish schools (Bosch et al, 2013) and CPUE standardization (Yang et al, 2020). The SVM performance highly relates to parameter selection and is proved perform well in seasonal flow volume predictions and CPUE standardization.…”

Section: Svmmentioning

confidence: 99%

CPUE retrieval from spaceborne lidar data: A case study in the Atlantic bigeye tuna fishing area and Antarctica fishing area

et al. 2022

View full text Add to dashboard Cite

The measurement of Catch Per Unit Effort (CPUE) supports the assessment of status and trends by managers. This proportion of total catch to the harvesting effort estimates the abundance of fishery resources. Marine environmental data obtained by satellite remote sensing are essential in fishing efficiency estimation or CPUE standardization. Currently, remote sensing chlorophyll data used for fisheries resource assessment are mainly from passive ocean color remote sensing. However, high-resolution data are not available at night or in high-latitude areas such as polar regions due to insufficient solar light, clouds, and other factors. In this paper, a CPUE inversion method based on spaceborne lidar data is proposed, which is still feasible for polar regions and at nighttime. First, Atlantic bigeye tuna CPUE was modeled using Cloud aerosol lidar and infrared pathfinder satellite observations (CALIPSO) lidar-retrieved chlorophyll data in combination with sea surface temperature data. The Generalized Linear Model (GLM), Artificial Neural Network (ANN) and Support Vector Machine Methods (SVM) were used for modeling, and the three methods were compared and validated. The results showed that the correlation between predicted CPUE and nominal CPUE was higher for the ANN method, with an R2 of 0.34, while the R2 was 0.08 and 0.22 for GLM and SVM, respectively. Then, chlorophyll data in the polar regions were derived using CALIPSO diurnal data, and an ANN was used for Antarctic krill. The inversion result performed well, and it showed that the R2 of the predicted CPUE to nominal CPUE was 0.92. Preliminary results suggest that (1) nighttime measurements can increase the understanding of the diurnal variability of the upper ocean; (2) CALIPSO measurements in polar regions fill the gap of passive measurements; and (3) comparison with field data shows that ANN-based lidar products perform well, and a neural network approach based on CALIPSO lidar data can be used to simulate CPUE inversions in polar regions.

show abstract

Standardizing catch per unit effort by machine learning techniques in longline fisheries: a case study of bigeye tuna in the Atlantic Ocean

Cited by 5 publications

References 28 publications

The use of spaceborne lidar to map Antarctic krill distributions and biomass in the Southern Ocean

The use of spaceborne lidar to map Antarctic krill distributions and biomass in the Southern Ocean

Implementation Of Linear Regression Algorithm And Support Vector Regression In Building Prediction Models Fish Catches Of Fishermen In Ciparagejaya Village

CPUE retrieval from spaceborne lidar data: A case study in the Atlantic bigeye tuna fishing area and Antarctica fishing area

Contact Info

Product

Resources

About