Uncertainty in scaling nutrient export coefficients

Abstract

Nutrient export coefficients are estimates of the total load or mass of nitrogen (N) or phosphorus (P) exported from a watershed standardized to unit area and unit time (e.g., kg ha-1 yr-1). They have been widely used as tools for environmental management of lakes, rivers and coastal waters because excessive nutrient enrichment (eutrophication) leads to several negative environmental effects (Carpenter et al. 1998). Yuan and Norton (2003) have shown, for example, that moderate increases in phosphorus concentrations in mid-Atlantic headwater streams shift benthic community structure toward a greater abundance of algae feeders, and others have shown that excessive nutrient input leads to "blooms" of nuisance algae (Paerl 1988, see Carpenter et al. 1998). Because of the potential negative environmental effects of excessive nutrient input, some programs are developing targeted goals of nutrient export by watershed (e.g., Linker et al. 1996). Nutrient export has been linked strongly to watershed land-cover composition (Beualac and Reckhow 1982, Frink 1991, Panuska and Lillie 1995). As forest is replaced by urban and agriculture, both nutrient export averages and variances tend to increase. Thus, the spatial heterogeneity of land cover in a watershed is an important influence on nutrient export. Watershed nutrient export coefficients are widely reported in the literature for watersheds ranging in size from 102 to 105 hectares (Dickerhoff Delwiche and Haith 1983, Lowrance et al. 1985, Clesceri et al. 1986, Jordan et al. 1997, Fisher et al. 1998), but have not been investigated for scale effects despite the wide range in watershed sizes for which they have been reported. Watersheds are analogous to grain (i.e., pixel size) in raster maps. Just as the size of a pixel can be changed, watersheds can also be resolved into few or many units for a given area. The watersheds used in the studies cited above could have been divided into two or more subunits with nutrient export coefficients estimated for each. Dividing watersheds into one or more subunits represents one approach to scaling. Scaling nutrient export introduces at least two issues related to error analysis and uncertainty (see Li and Wu, 2003, Chapter 3). First, moving nutrient loads across subwatersheds requires estimation of loss rates as they move from upstream to downstream. In-stream decay of nutrients varies as function of discharge and geographical location (Smith et al. 1997, Peterson et al. 2001). Second, the total load exported from the entire watershed is dependent on whether nutrient export among subwatersheds is assumed to be independent. Dependence exists if nutrient export patterns among subwatersheds are similar. Subwatersheds can be assumed to behave independently if their nutrient export values are not strongly similar. The purpose of this chapter is to examine the effects of these two sources of uncertainty on estimated nutrient export as a function of the number of subwatersheds resolved. © 2006 Springer.