Abstract: A set of monthly mean global ocean(0~1 500 m)Argo data sets (1°× 1°) from 2004 to 2017 was constructed by using “Gradient-dependent Correlation Scale Method”. And based on the comparative test of the data set, we have initially applied the data set to the fisheries analysis of yellowfin tuna in the Central and Western Pacific Ocean. The results show that the temperature and salinity deviations between the Argo data set and the WOA13 historical data set were slightly larger in the upper surface layer of the ocean about 0.5 ℃ and 0.1 respectively, and the two deviations both decreases gradually with the increase of depth. The temperature deviations between the Argo data set and time series of the TAO buoy were less than 1 °C in 2004-2017, and the maximum salinity deviations were less than 0.5 while most of those in the sea areas were close to 0. In the Central and Western Pacific Ocean, the central fishing ground of yellowfin tuna mostly concentrated at the isotherm of the range of 28 ~ 29 ℃, and in the sea areas where temperature below 22 ℃, the catch per unit effort (CPUE) value was very small. In the central fishery area, the upper boundary depth of the thermocline was in the range of 20 m and 120 m, and the frequency of formation of the center fishing ground at each depth was generally normal distribution. When the upper boundary depth of the thermocline was 90 m, the possibility of forming the center fishing ground was the greatest. While further verifying the reliability of the data set, it is also shown that the data set constructed in our study has certain application value in hydrologic environment analysis and resource assessment.

Key words: gradient-dependent, OI, Argo, analysis of fishing ground