Background and Context

The origins of BigFoot are traceable to 1992 when S. Running of MODLand contacted the Long-Term Ecological Research (LTER) community to assist with validation of anticipated MODIS products. The first proposal, funded by NASA's Terrestrial Ecology Program (TEP) in 1994, involved prototype scaling exercises using existing data at 14 (mostly LTER) sites. In 1996, the 20+ person team held an LTER-funded workshop to share results and stratagies for future direction. Workshop proceedings were published in a special issue of Remote Sensing of Environment in October 1999, edited by Cohen and Justice. In 1998, the largely amorphous project was rescoped, and with renewed TEP support became what we now call BigFoot. What defines the project is its focus on using remote sensing and ecosystem process models to scale ecological field measurements within a greater eddy flux tower footprint. Doing this across a range of biomes using standardized measurements and an innovative sampling design affords us a unique opportunity to examine ecosystem function at the local level in a global context.

Originally, BigFoot included four sites, with the first field season being in 1999 at two of these sites (see group 1 in field activity schedule below). Each site represents a distinct and globally important biome. The NOBS site (or Northern Old Black Spuce site from BOREAS) is a boreal needleleaf evergreen forest. HARV, (the Harvard Forest LTER site) is a temperate mixed forest. The agricultural cropland site, AGRO (a.k.a. Bondville) is primarily corn and soybeans. KONZ (the Konza Prairie LTER site) is a tallgrass prairie. The project focused on validation of MODIS land cover, LAI, and NPP products.

The overall goal of BigFoot is to continue with MODIS land cover, LAI, and NPP product validation at the four existing sites, but now we have expanded our focus to include validation of f_APAR and our biome coverage to include five new sites for a more globally comprehensive assessment of MODIS product validity (see field activity schedule above). The new sites include a desert grassland (SEVI, the Sevilleta LTER), an tundra near the Arctic coastal plain (TUND, near Barrow, AK), and a tropical broadleaf evergreen forest in the Amazon Basin (TAPA, the Tapajos Primary Forest site of LBA). In addition, we offer two sites at no additional cost to NASA (Group 3); these sites include a second temperate mixed forest (CHEQ, Chequamegon National Forest in Wisconsin), and a temperate needleleaf evergreen forest near Metolius, Oregon (METL). The Group 3 sites are fully funded by either the NASA EOS Validation Program (CHEQ to Gower) or the Environmental Protection Agency (METL to B. Law, Turner, Cohen et al.). Continued work at NOBS beyond 2000 is funded by the National Science Foundation to Gower. Another important goal of BigFoot is to explicitly assess if MODLand products can detect the effects of inter-annual variation in climate on vegetation cover, LAI, f_APAR, and NPP. Also, we examine how LUE varies within and among sites and vegetation types and propose to actively facilitate further formation and definition of GTOS in support of MODIS validation.

In addition to BigFoot, there are several other initiatives addressing the development of a global terrestrial monitoring and validation program. These include the Global Primary Production Data Initiative (GPPDI), the worldwide CO₂ flux network (FLUXNET), the MODLand Science Team and associated validation program, the Vegetation/Ecosystem Modeling and Analysis Project (VEMAP), and the Global Analysis, Integration, and Modeling (GAIM) NPP model intercomparison activity. These all recognize the need to consider several key elements in the formation of a global terrestrial monitoring system, including: in situ vegetation measurements, eddy flux tower measurements, vegetation surfaces derived from remote sensing, and biogeochemical process models, all used over time. However, these programs exist either as stand alone efforts (e.g., FLUXNET, MODIS-NPP modeling) or incorporate only one or two of the key elements of a comprehensive observation and validation program (e.g., GPPDI, VEMAP). In contrast, BigFoot integrates all of the aforementioned key elements at a spatial scale that serves as a rigorous validation of global sensor and terrestrial modeling products (as shown in the figure below).

Specifically, BigFoot:

• Uses field measurements and Landsat ETM+ data to parameterize eocsystem process models at a local scale (i.e., within the 5 x 5 km BigFoot footprint) so that we can understand the environmental and ecological controls on carbon exchange between terrestrial ecosystems and the atmosphere at that scale.

• Assesses how accurately process models capture the environmental controls on CO₂ and H₂O exchange between terrestrial ecosystems and the atmosphere.

• Statistically compares the land cover, LAI, f_APAR, and NPP surfaces we generate against field measurements to help establish the accuracy of those surfaces, so that they can be legitimately compared gainst co-located sections of globally-derived, generalized surfaces (such as those of MODLand).

• Compares BigFoot modeled GPP (gross primary production) against calculations of GPP from flux tower data–a strong first step towards effective integration of process models and flux measurements.

• Provides both field-based and map-based characterizations of flux tower footprints.