PARENT SESSION
Symposium #25: Scaling information from plots or regions to landscapes: when do details matter?.

Organized by: DPC Peters and JE Herrick
Thursday, August 8. 9:00 AM to 11:30 AM. Crystal Ballroom, TCC.

Simple and weighted averaging approaches to scaling: When can spatial context be ignored?

PARTON, WILLIAM^*,1, LANE, DIANA², BESTELMEYER, BRANDON³, MITCHELL, KATHERINE³, KELLY, ROBIN¹, ¹ Colorado State University, Fort Collins, CO² University of Illinois-Chicago, Chicago, Ill³ USDA-ARS Jornada Experimental Range, Las Cruces, NM

ABSTRACT- Scaling from plots to landscapes, landscapes to regions, and regions to the globe based on simple or weighted averaging techniques can be accurate when applied to the appropriate problems. Simple averaging approaches work well when conditions are homogeneous spatially and temporally. For example, modelling of grassland and forest stands by averaging results from multiple small plots that have no horizontal transport across the plots is an appropriate use of simple averaging. The assumption of spatial homogeneity is often violated at the landscape and regional scale as a result of topographic and edaphic variability. Weighted averaging approaches can address landscape and regional variability by dividing the area up into a collection of homogeneous discrete patches. This approach has been used successfully for the regional and global Century modelling work where carbon and nitrogen fluxes are calculated as area weighted averages of model output from polygons with distinct combinations of climate and soil properties. Net primary production and soil carbon and nitrogen levels have been successfully extrapolated at the regional and global scale using area weighted averages of model results and observed soils and climate data bases. This approach would not work at the landscape level when horizontal fluxes of nutrients and water across the patches are significant and potentially impact the dynamics of the patch. In animal ecology weighted averaging approaches assume that limitations to patch perception, selection, or dispersal are unimportant over the scale of interest. This paper will use examples from biogeochemistry, and plant and animal ecology to demonstrate when simple or weighted averaging can be used to spatially extrapolate ecosystem process results.

KEY WORDS: modelling, plant production, nutrient cycling, scaling regional analysis