PREDICTION OF FISHING BOAT DENSITY USING VIIRS IMAGERY

Title
PREDICTION OF FISHING BOAT DENSITY USING VIIRS IMAGERY
Author(s)
Jeon, Ho Kun; Cho, Hong Yeon
KIOST Author(s)
Cho, Hong Yeon(조홍연)
Alternative Author(s)
전호군; 조홍연
Publication Year
2022-05-17
Abstract
East Sea(Japan Sea), where the Kuroshio current and Liman current converged, is rich in nutrients and plankton, making it a good fishing ground. Monitoring capabilities for fishing vessels are required to prevent illegal fishing activities and conserve marine resources. This study suggests predicting fishing density distribution using pre-observed fishing vessels’ locations, marine biogeochemistry and seawater depth, and the Random Forest model. The locations of fishing vessels are generated from VIIRS DNB imagery through the speckle detection method. Ocean biochemistry forecast data are from Copernicus Marine Service, and depth data are from ETOPO1. The biochemistry and depth are in raster format, whereas ship locations are points. In order to meet data type as raster, vessels’ locations are transformed into fishing vessel density maps. Marine biochemical and water depth data are rescaled into the same grid size of the density maps. The three data types were rearranged and merged into a table type as the Random Forest dataset. The correlation of density to biochemistry and depth are examined, the importance of the variables to predict density are checked, and the optimum hyperparameter is set in advance. The data of the target prediction date is chosen for validation, while the training dataset is consisted of several days five days before the target date. The trained model using the training dataset predicts the density of the target date. The daily prediction performance is recorded after running the model.
URI
https://sciwatch.kiost.ac.kr/handle/2020.kiost/42486
Bibliographic Citation
ISRS 2022 (International Symposium on Remote Sensing 2022), pp.227 - 230, 2022
Publisher
ISRS
Type
Conference
Language
English
Files in This Item:
There are no files associated with this item.

qrcode

Items in ScienceWatch@KIOST are protected by copyright, with all rights reserved, unless otherwise indicated.

Browse