Integration of satellite-based environmental data for Skipjack tuna fishing ground determination

Skipjack Tuna fishing ground area was studied using satellite remotely sensed environment and catch data. Skipjack Tuna tend to aggregate in ocean areas that exhibit specific environmental conditions with water quality parameters such as Sea Surface Temperature (SST) and dissolved oxygen. Weekly resolved remotely sensed SST, surface chlorophyll (Chl-a), and Sea Surface Height (SSH) in 2017 being used as the environmental parameters to determine Skipjack tuna fishing ground. Machine Learning method which is Decision Tree (DT) were constructed with the environmental parameters as model covariates to examine Skipjack Tuna fishing ground determination. As a comparison, Generalized Linear Model (GLM) also was applied. This study exhibits the ability of both model performances for the fishing ground determination. Both models appear to be appropriate for fishing ground determination, while the DT acquires a better performance than GLM. The mean accuracy and Area Under Curve (AUC) of DT was about 0.876 and 0.877, respectively. In other hand, GLM only able to acquire 0.6907 and 0.8397 for its mean accuracy and AUC, respectively.