The global ocean is a major sink of anthropogenic carbon dioxide (CO2) emitted into the atmosphere. Machine learning has been actively used in the past decades to estimate the oceanic sink, but it is still a challenge to obtain an accurate estimate due to scarcely available CO2 measurements. One of the methods to deal with data scarcity was normalizing multiple years’ CO2 values to a reference year to increase the spatial coverage. The practice assumed a constant CO2 trend for the normalization. Here, we used three machine learning models to extract variable ocean CO2 trends on a decadal scale and proposed a method to use the extracted ocean CO2 trends to correct the decadal atmospheric CO2 trends for data normalization. The method minimizes assumptions of using the extracted ocean CO2 trends directly. Comparisons of our CO2 flux estimate with machine learning products included in Global Carbon Budget 2021 indicates that using the variable trends improved the bias resulted from using a constant trend and that the trends are a critical factor for machine learning methods. Our dataset includes monthly distributions of surface ocean CO2 concentration and air-sea flux in 1980-2020 with a spatial resolution of 1×1 degree.