Exploring Relationships Between Environmental Variables and Global Distribution of Particulate Inorganic Carbon versus Particulate Organic Carbon Using Random Forest Analysis

Understanding the distributional characteristics and sensitivities of global particulate inorganic carbon (PIC) and particulate organic carbon (POC) to environmental drivers is crucial for accurately modeling biogeochemical cycles in the ocean. PIC and POC are vital components of the ocean's carbon cycle, and their distributions and responses to factors such as temperature, nutrients, and pH offer valuable insights into marine ecosystem functioning and the effects of environmental changes. Previous studies have highlighted the significance of the export ratio of CaCO₃ to organic carbon in influencing atmospheric CO₂ levels and the global climate. However, there is still a critical knowledge gap regarding the extent to which Earth System Models (ESMs) capture the mechanisms governing the distribution and quantity of PIC versus POC in the ocean. Here we utilized Machine Learning techniques to reveal apparent relationships between physiological mechanisms and different types of phytoplankton in different regions. Our analysis of observations from MODIS suggests that different biogenic carbon species respond differently to environmental drivers. Initial results show that satellite-estimated PIC appears to be less sensitive to iron and more sensitive to light and mixed layer depth compared to satellite-estimated POC. Further investigation is needed to determine whether Earth System Models (ESMs) accurately capture these distinct changes for different carbon species to improve the modeling of marine biogeochemical cycles.