Spatial Distribution of Land Type in Regression Models of Pollutant Loading

Abstract

This paper proposes a method to improve landscape-pollution interaction regression models through the inclusion of a variable that describes the spatial distribution of a land type with respect to the pattern of runoff within a drainage catchment. The proposed index is used as an independent variable to enhance the strength, as quantified by R² values, of regression relationships between empirical observations of in-stream pollutant concentrations and land type by considering the spatial distribution of key land-type categories within the sample point's drainage area. We present an index that adds a new dimension of explanatory power when used in conjunction with a variable describing the proportion of the land type. We demonstrate the usefulness of this index by exploring the relationship between nitrate (− 3 NO) and land type within 40 drainage sub-catchments in the Ipswich River watershed, Massachusetts. Nutrient loads associated with non-point source pollution paths are related to land type within the upstream drainage catchments of sample sites. Past studies have focused on the quantity of particular land type within a sample point's drainage catchment. Quantifying the spatial distribution of key land-type categories in terms of location on a runoff surface can improve our understanding of the relationship between sampled − 3 NO concentrations and land type. Regressions that employ the proportion of residential and agricultural land type within catchments provide a fair fit (R² = 0.67). However, we find that a regression adding a variable that indicates the spatial distribution of residential land improves the overall relationship between instream − 3 NO measurements and associated land types (R² = 0.712). We test the sensitivity of the results with respect to variations in the surface definition in order to determine the conditions under which the spatial index variable is useful.

Geographically weighted regression to measure spatial variations in correlations between water pollution versus land use in a coastal watershed

“…Generally, built-up land and agricultural land have significant positive correlations with water pollution, which are associated with point or non-point source pollution (Johnson et al, 1997;Sliva and Williams, 2001;Fedorko et al, 2005;Mehaffey et al, 2005;Stutter et al, 2007;Tu et al, 2007;Bahar et al, 2008;Tran et al, 2010;Pratt and Chang, 2012;Yang, 2012). Woodland is significantly negatively correlated with nutrients, due to the general understanding that forests can absorb nutrients (Osborne and Kovacic, 1993;Novotny, 2002;Galbraith and Burns, 2007;Bahar et al, 2008;Lopez et al, 2008).…”

“…buffer versus entire watersheds) (Sliva and Williams, 2001;Uuemaa et al, 2005;Guo et al, 2010;Tran et al, 2010;Pratt and Chang, 2012) and watershed characteristics represented by various dominant land uses such as woodland, built-up, and mining (Mehaffey et al, 2005;Xiao and Ji, 2007;Bahar et al, 2008). The methods used in most of the abovementioned studies are conventional global statistical methods, such as Pearson correlation analysis (Tong and Chen, 2002;Galbraith and Burns, 2007;Bahar et al, 2008;Lee et al, 2009;Sun et al, 2011b) and multiple regression (Sliva and Williams, 2001;Fedorko et al, 2005;Huang et al, 2013a;Yang, 2012). These global statistics are commonly used to analyze the overall association for the entire study area, and may hide some local relationships, especially among watersheds that are dominated by different uses, such as urban, forest or agriculture (Tu and Xia, 2008;Tu, 2011).…”

Influence of carbon mapping and land change modelling on the prediction of carbon emissions from deforestation

“…Our method may be applied to any quantitative changes that can be represented spatially and whose value is a function of deforestation or other type of spatial disturbance. The method can be used to evaluate the influence of climate change scenarios on the viability of plant populations (Miles et al 2004) and/or fauna turnover (Peterson 2002), or to analyse the sensitivity of predicted nitrogen retention and deposition on the quantity and spatial distribution of climate, forest types and anthropogenic land changes (Pan et al 2004;Fedorko et al 2005).…”

Empirical relationships between land use/cover and estuarine condition in the Northeastern United States

“…The effect of observational scale has been described for terrestrial (Krummel et al 1987;Turner 1989) and aquatic ecosystems (Hall et al 1994;Legendre et al 1997). Future investigations of land use/cover and estuarine condition might benefit from the incorporation of spatial distribution of land-type categories into regression models (Fedorko et al 2005). Furthermore, our results show the need to improve our ecological data infrastructure (i.e., type, quantity, and quality) since future modeling will be quite restricted by not having overwhelming statistical signals.…”

“…• Development of connectivity or permeability landscape metrics at different scales and extents may be more appropriate to measure landscape structure than total land use area (Fedorko et al 2005). …”