Geo-spatial Information Science (Nov 2024)

Optimal parameter schemes for global and regional gross primary productivity estimation: a comparative analysis

  • Yuyu Li,
  • Xiaobin Guan,
  • Shen Huanfeng

DOI
https://doi.org/10.1080/10095020.2024.2416658

Abstract

Read online

The spatiotemporal heterogeneity of the parameters greatly impacts the estimation of Gross Primary Productivity (GPP) by Light Use Efficiency (LUE) models. Vegetation type-specific parameters are commonly set at present; however, the various environmental heterogeneities at global and regional scales can also induce disparate optimal parameter schemes, which has not been further explored. Therefore, in this study, we comparatively explored the parameter differences and GPP estimation accuracy under different parameter schemes, including Vegetation type Parameters (VPs), Fixed Parameters (FPs), Vegetation type Monthly Parameters (VMPs), and Fixed Monthly Parameters (FMPs), at both global and regional scales. Two different strategies were applied to validate the ability of the temporal prediction and spatial expansion for the four parameter schemes. The results indicate that the VP scheme shows only a limited superiority over the FP scheme for GPP estimation at the global scale (ΔR2 = 0.01–0.02), which indicates that the VP scheme is not necessary for the global application of LUE models, due to the considerable spatiotemporal heterogeneity of vegetation. However, for regional applications, such as the Mediterranean region, the VP scheme is preferable to the FP scheme, with an improved ΔR2 of more than 0.05, which is because the vegetation within the same type is much more similar than at a global scale. Furthermore, it is found that the time-varying parameters (VMPs and FMPs) contribute little to the GPP simulation at both global and regional scales, which is possibly due to the limited amount of available data. Overall, the results of this study will not only offer guidance and optimal parameter schemes for the global and regional estimation of GPP, but also highlight the importance of considering spatiotemporal heterogeneity for parameters.

Keywords