National Aeronautics and Space Administration

Large scale forest productivity estimates are of increasing interest as more demands are made on forest resources. In principle three different methods are currently available: (i) forest growth data from forest research plots and/or forest inventory sampling points based on repeated tree observations, (ii) flux tower observations which record the gas exchange of the plant atmosphere interactions for a given vegetation type, and (iii) remotely sensed data for a continuous cover of net primary productivity estimates. In this paper we focus on the conceptual challenge in comparing “space based” moderate resolution imaging spectroradiometer (MODIS) satellite driven net primary production (NPP) vs. terrestrial "ground based" productivity estimates using forest increment data from 151 long term research plots with a well documented management history. The Austrian biomass functions are applied to derive ground based NPP estimates based on repeated tree observations from the plots. In addition we use BIOME-BGC as a diagnostic tool for exploring conceptual constraints among the two methods. The results of the study can be summarized as follows: (i) MODIS satellite driven annual NPP estimates provide a continuous productivity estimate across Austria and no significant differences between different daily climate input data sets were evident. (ii) MODIS NPP predictions provide forest productivity estimates of fully stocked forests with a complete crown cover. This is confirmed by the results of spin-up runs of the BIOME-BGC model. (iii) Terrestrial driven NPP predictions using the Austrian biomass functions compared well with MODIS driven estimates after addressing stand density effects of the forest plot data. The influence of stand density were known to be an integral component in reconciling "space based" satellite vs. "ground based" derived forest productivity estimates. After addressing stand density, computed forest productivity estimates compared well with MODIS-based estimates. This suggests that combining both methods will enhance our ability to generate forest productivity assessments across large forest areas.