Evaluating the use of digital terrain modelling for quantifying the spatial variability of 2,4-D sorption by soil within agricultural landscapes

Abstract

The most sensitive input parameter in many herbicide fate and transport models is the sorption-partitioning coefficient (Kd), a measure of herbicide sorption by soil. Spatial analyses of Kd are traditionally performed using geostatistics, but this approach requires intensive soil sampling and herbicide analysis. This study examined the use of digital terrain modelling as an alternative tool for quantifying the distribution of herbicide sorption within agricultural fields. Soil samples from a conventional-till (CT) and zero-till (ZT) field were analysed for soil organic carbon (SOC), soil pH, clay content, and the 2,4-D ([(2,4-dichlorophenoxy)acetic acid]) soil-water partitioning coefficient. Digital terrain models were used to calculate topographical variables (elevation, slope gradient, slope aspect, horizontal curvature, vertical curvature, mean curvature, specific catchment area, topographic index and stream power index) for each sampling point. Results indicated that topographic variables were adequate predictors of all soil properties in ZT (R2 ranging from 0.64 to 0.76), and of SOC in CT (R2= 0.65, P < 0.001). For CT, 2,4-D sorption by soil was very well predicted with soil properties alone (R2= 0.82, P < 0.001) and with soil properties in combination with topographic variables (R2= 0.85, P < 0.001), but was less predicted by topographic variables alone (R2= 0.50, P < 0.001). For ZT, the level of prediction of 2,4-D sorption by soil was weak with soil properties alone (R2= 0.53, P < 0.001) or when topographical variables alone (R2= 0.56, P < 0.001) were used, but a substantial increase in the level of prediction was achieved when both soil properties and topographic variables were used (R2= 0.73, P < 0.001). We conclude that digital terrain modelling, in combination with soil properties data, is an appropriate approach for predicting the spatial distribution of 2,4-D sorption within undulating-to-hummocky glacial till landscapes in western Canada. Key words: herbicide sorption, topography, zero-tillage, conventional-tillage, digital terrain modelling.