1527 / 2024-09-27 20:35:22

Sensitivity analysis of the Carbon-Based Productivity Model (CbPM) in the Arabian Sea

primary productivity,sensitivity analysis,cbpm,Arabian Sea

The ocean's biological productivity plays a vital role in regulating the global carbon cycle and sustaining marine ecosystems. Monitoring phytoplankton primary production is critical for understanding aquatic ecosystems' response to climate change and for managing marine resources. The Carbon-based Productivity Model (CbPM), which incorporates both satellite-derived and model-based inputs such as Chlorophyll (Chl), particulate backscatter (bbp443), Photosynthetically Available Radiation (PAR), light attenuation (K490), Mixed Layer Depth (MLD), and nitrate concentration (ZNO3), to estimate Net Primary Productivity (NPP). A comprehensive sensitivity analysis of the CbPM model was conducted across the Arabian sea, employing Sobol sensitivity analysis, Random Forest (RF), and Gradient Boosting Machine (GBM) approach. Kd490 emerged as the most significant driver of NPP variability, contributing 36.2% to the overall model sensitivity. The interplay between Chl and Kd490 was significant, with regions of higher light attenuation showed more pronounced changes in productivity. Coastal regions exhibited heightened sensitivity to Chl and Kd490, whereas open ocean areas were more influenced by bbp443 and MLD. Machine learning models including RF and GBM further validated these findings, identifying PAR and MLD as critical variables, particularly in open ocean regions with stronger vertical mixing. The Generalized Additive Models (GAMs) provided additional insights into non-linear relationships between the parameters, emphasizing the role of light and nutrient availability in shaping NPP dynamics. Spatial analysis results indicated distinct regional variability in model sensitivity, particularly in the coastal upwelling zones and river-influenced regions. High sensitivity to Kd490 was observed in coastal areas with higher light attenuation, while bbp443 and MLD played significant roles in the open ocean. In the region, the coast of southwest India shown only strong variability in nutrient driven NPP compared to the other parts of Arabian sea. These findings suggest that the CbPM model is highly responsive to local environmental conditions, underscoring the need for regional-specific tuning when applying the model. While the current study utilized remote sensing and model-based inputs, the input data for model from the integration of autonomous platforms such as Argo float, and glider holds great potential for enhancing the understanding of sub surface variations. The results highlight the importance of considering both regional variability and parameter interactions and provides insights into the sensitivity of the model, when using the CbPM model, in dynamic environments as in tne Arabian Sea.