Modeling drivers of satellite-derived net primary productivity: inclusion of non-vegetated areas can be potential source of error

Abstract

ABSTRACT This study modelled the drivers of satellite-derived Net Primary Productivity (NPP). The NPP drivers used were human foot print (HFP), soil organic carbon (SOC), Total Nitrogen (TN), Total Phosphorus (TP), Potential evapotranspiration, (PET), and Aridity Index (AI). The results showed that inclusion of non-vegetated land cover types (LCT) in the modelling leads to establishment of wrong relationships between the NPP and its drivers, and wrong relationships among the drivers. Inclusion of the non-vegetated LCT revealed that 1) HFP had no relationship with NPP, CUE, and SOC; 2) PET had no relationship with all the three soil variables (SOC, TN, and TP), AI and the CUE; 3) PET and NPP were strongly and positively correlated; and 4) there was no relationship between the NPP and all the three soil variables and the AI. While exclusion of the non-vegetated LCT revealed that 1) HFP had negative relationship with NPP, CUE, and SOC; 2) PET had negative relationship with the CUE, AI and all the three soil variables; 3) PET and NPP were strongly negatively correlated; and 4) there was positive correlations between the NPP and all the three soil variables and AI. However, results of both the inclusion and exclusion of the non-vegetated LCT indicated that: 1) HFP did not correlate with TN, TP, PET, and AI; and 2) there was strong and positive correlation among all the three soils and AI. It is worth noting that exclusion of the non-vegetated LCT in the modelling established the most accurate results on the relationship between the NPP and its drivers; and on the relationships among the drivers. However, there is a need to replicate this study in different geographical regions to guarantee global generalization of the results.