Mixed tree-grass and shrub-grass vegetation associations are one of the most spatially extensive and widely distributed forms of terrestrial vegetation on earth. They constitute significant fractions of all continents, except Antarctica, in tropical, subtropical and temperate bioclimatic regions. While global tree-grass systems are diverse in their phylogeny, physiology and plant morphology, they share the key structural characteristic of woody plants distributed in the landscape at densities low enough to allow significant growth of herbaceous plants (mostly grasses) underneath and between them.
Despite the importance of tree-grass systems in earth system processes and human well-being, they are not well represented in our remote sensing and modeling capabilities. Ecosystems characterized by horizontally and vertically complex tree-grass mixtures are inherently difficult to measure with remote sensing and difficult to represent in ecosystem and earth system models. Here we leverage emerging data on slowly varying canopy structure (tree cover) to provide a key constraint in the estimation of rapidly (i.e. seasonally) varying tree and grass leaf area index (LAI) and fractional PAR interception (fPAR). This tree-grass separation in remote sensing data has not, to our knowledge, been attempted before at regional and continental scales using MODIS data.
This project capitalizes on prior Terrestrial Ecology sponsored research to develop and test a tree-grass separation methodology, validated using field data, and applied for Africa during the MODIS (2002-2012) era. We anticipate that the methodology thus developed will later be used to also partition VIIRS LAI and fPAR aggregates. Partitioned tree and grass biophysical parameters will be of direct use as a validation and data-assimilation stream for our savanna dynamic vegetation model (the Tree-Grass Vegetation Model, TGVM). The new tree-grass biophysical datasets will advance the testing and application of TGVM in both diagnostic and prognostic modes to explore how tree-grass systems are responding to human management and climate change.
As part of this research we plan to (i) refine methodologies for separation of tree and grass LAI and fPAR biophysical parameters using MODIS products for all Africa, (ii) validate our results using field measurements from our own and collaborator field sites across Africa, (iii) generate Africa-wide tree and grass biophysical parameters for 2002-2012, at 8-day and 1 km resolutions, (iv) analyze and synthesizing the new LAI and fPAR products for publication, and (v) use the tree and grass LAI and fPAR data with our tree-grass vegetation model (TGVM) to transform our understanding of, and ability to simulate, the future provision of ecosystem goods and services in tree-grass systems.
Over the past two decades NASA has made considerable investment in development and deployment of the MODIS platform and MODIS data products. This project seeks to leverage that investment using the LAI-fPAR product to diagnose and better understand the separate and distinct role of woody and herbaceous components in mixed tree-grass ecosystems.
LAI partitioning method involves use of allometric relationships we are developing describing how within tree (“in-canopy”) LAI varies for savanna trees across the rainfall gradient and independent estimates of slowly varying woody canopy cover. These data are used as a key constraint to infer seasonal woody LAI phenologies. Herbaceous LAI is then estimated by difference. Using available literature data, this approach has been used to make provisional estimates (beta version) of seasonally changing woody LAI and herbaceous LAI in sub-Saharan Africa for the MODIS time series 2003-2013, as shown in the animation above. Algorithms for the partitioning process will be improved and products validated based on fieldwork in Eastern Africa and our collaborators Southern and Western Africa. We are also availing the beta version of the partitioned LAI products (hyperlink) to the public for feedback on their usability.