A workshop was held at the H.J. Andrews Experimental Forest in May 1996 to work towards development of standardized methods to be used in the MODLERS project. This project brings together 14 long-term ecologi- cal research (LTER) network sites and NASA’s MODIS Land (MODLAND) Discipline Group for the purpose of locally validating Earth Observation System-era global data sets. Using several different methods that incorporate extensive ground data sets, ecosystem models, and remotely- sensed imagery, each LTER site is developing local maps of landcover class, leaf area index, and aboveground net primary productivity for a 100 km2 area at a grain size of 25 m. A nested, hierarchical ground-based sampling scheme will help establish error bounds on the variable estimates.
A number of different strategies are being used to spatially aggregate the fine-grain site maps to a coarse grain (1 km) so that they can be compared to coincident portions of global maps ofthe same three biosphere variables developed by the MODLAND Discipline Group. This coordinated, multi-site grain-size aggregation exercise presents an opportunity to grapple with one of the most vexing current problems in ecology: effects of scaling from a fine grain to a coarse grain on estimates of important biosphere variables. Participants are using several spatially- explicit, geostatistical methods to address this issue, with the intent of determining how best to maintain crucial information among grain sizes. Additionally, they are characterizing similarities and differences among the multiple sites and biomes and between the MODLAND maps and the site maps at each grain size, in terms of the three mapped biosphere variables. At the May workshop, participants decided to work toward development of a special issue of the journal Remote Sensing of Environment that will contain papers on methods to be used and early results from methods prototyping activity.
For more information visit our web site at http://atlantic.evsc.virginia.edu/~jhp7e/modlers/
