A comparison between phytoplankton community structures derived from a global 3D ecosystem model and satellite observation

Phytoplankton functional types (PFTs) derived from a global 3D marine ecosystem model, MEM-OU, are compared with those observed by satellite. The climatological field of the large phytoplankton (diatoms) and small phytoplankton (nano- and picoplankton) showed the model and satellite observation generally agrees in their spatial distribution but showed some biases in the equatorial oceans and higher latitudes. Grid-by-grid correlation derived from the time series of the model and satellite observation showed a significant dependence on latitude, and it was found that lower (higher) correlation coefficients were not necessarily found in the regions where higher (lower) bias was found. Spatial PCA analysis of monthly climatology of PFTs showed that a major spatial pattern of PFTs results from a steady state distribution superimposed by regional seasonal distribution. The model and the satellite observation both showed such seasonality, but we found that the model remarkably showed strong seasonality in the Southern Oceans and the equatorial oceans. The wavelet pattern comparison analysis was applied to quantify the spatial scale(s) at which the agreement in spatial distributions of PFTs can be obtained. The objective skill score determined using the pattern matching/mismatching analysis by the wavelet showed that the model has a skill to represent observed spatial patterns at horizontal scales over 1800 km (i.e. basin and global scale). The agreement is found even at the spatial scales down to 200 km (i.e. meso-scales), when the pigment biomass of the PFTs at the scales is relatively small. Our comparison enabled an objective and quantitative examination of the validity and weakness of the model by clarifying the spatial scale(s) to which the present PFT model has skill, and adds a value on conventional comparison procedure based on univariate statistics.