An inter-comparison of plot-scale, satellite and earth system model estimates of tropical net primary productivity (Invited)
Abstract
Tropical forests exchange more CO2 with the atmosphere than any other biome, making them a key control over Earth's climate. And yet, our ability to both measure and model the tropical carbon (C) cycle remains far from ideal, creating a substantial challenge for the development of Earth system models that couple the climate system with ecosystem dynamics. In part, this deficit arises from a lack of sufficient data combined with a biome that displays enormous biogeochemical heterogeneity. Here, we compare a new synthesis of plot-based measurements of tropical net primary productivity (NPP ) compared with two commonly used approaches to evaluating the tropical C cycle at large scales: NPP estimates derived from 1) the MODIS MOD-17 algorithm, and 2) the Community Land Model version 4.5. We also assess the major drivers of variance in NPP in each method, partly as a way to explore how well modeled and satellite-derived values compare to field-based measurements of NPP responses to environmental variables. At the largest scale, MODIS, CLM and a simple climate-based extrapolation of the plot-scale data compare reasonably well: multi-year averaged pan-tropical NPP values from the three approaches were 9.4, 10.8 and 9.5 PgC/yr, respectively. However, inter-comparisons at finer spatial and temporal scales reveal substantial differences among the three methods. For example, CLM predicts a steady increase in tropical NPP throughout the last decade or more, largely because of model assumptions surrounding the importance of CO2 fertilization, while MOD-17 produces a declining NPP trend. CLM also predicts significant N-limitation of lowland forest NPP, a finding that does not agree with most field-based evidence. MODIS estimates show little dependence on fPAR (fraction of absorbed photosynthetically active radiation), in part because the complex canopy architecture creates a radiative transfer environment that the MODIS sensor cannot resolve. Therefore, variation in MODIS-derived NPP across the tropics is almost entirely a function of differences in radiation load and climate variables. Regional patterns of NPP are also notably different among the three approaches. As well, assessments of the plot data in concert with information on soil type and major nutrient pools suggest varying degrees and identities of nutrient limitation of C exchange across the tropical biome, in ways that neither CLM or MODIS can readily capture. Taken as a whole, our analyses suggest the need for caution in using either current earth system models or MODIS data in evaluating tropical NPP responses to spatial or temporal variation in environmental drivers. At the same time, plot-scale data face their own major drawbacks. These include a frequent lack of information on all key components and drivers and their small spatial coverage relative to the scale at which many questions need to be addressed. That said, there is reason for optimism looking forward, as new field efforts in the tropics along with new developments in both airborne and satellite remote sensing hold real promise for producing data that can both better constrain the current tropical C cycle and help refine prognostic models like CLM.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2013
- Bibcode:
- 2013AGUFM.B12C..05T
- Keywords:
-
- 0428 BIOGEOSCIENCES Carbon cycling;
- 0439 BIOGEOSCIENCES Ecosystems;
- structure and dynamics