Spatial Variability and Uncertainty of Water Use Impacts from U.S. Feed and Milk Production.

This paper addresses water use impacts of agriculture, developing a spatially explicit approach tracing the location of water use and water scarcity related to feed production, transport, and livestock, tracking uncertainties and illustrating the approach with a case study on dairy production in the United States. This approach was developed as a step to bring spatially variable production and impacts into a process-based life cycle assessment (LCA) context. As water resources and demands are spatially variable, it is critical to take into account the location of activities to properly understand the impacts of water use, accounting for each of the main feeds for milk production. At the crop production level, the example of corn grain shows that 59% of water stress associated with corn grain production in the United States is located in Nebraska, a state with moderate water stress and moderate corn production (11%). At the level of milk production, four watersheds account for 78% of the national water stress impact, as these areas have high milk production and relatively high water stress; it is the production of local silage and hay crops that drives water consumption in these areas. By considering uncertainty in both inventory data and impact characterization factors, we demonstrate that spatial variability may be larger than uncertainty, and that not systematically accounting for the two can lead to artificially high uncertainty. Using a nonspatial approach in a spatially variable setting can result in a significant underestimation or overestimation of water impacts. The approach demonstrated here could be applied to other spatially variable processes.

Laser profilometry and length-scale analysis of stone tools: second series experiment results.

Based on the need to develop a method to reliably and objectively document and discriminate the use-wear on archaeological stone tools, Stemp et al. (2009) tested whether the surface roughness measured on experimentally worn stone tools used on different contact materials could be discriminated. Results of these initial experiments indicated that discrimination was possible and also determined the scales over which this discrimination occurred. In this article, we report the results of additional experiments using the same method on a second set of tools to test its reliability and reproducibility. In these experiments, four flint flakes were intensively used for 20 min on either conch shell or dry deer antler. The surface roughness or texture of the stone tools was measured by generating 2D profiles using a UBM laser profilometer. Relative lengths (RLs) calculated from the profiles were used directly as characterization parameters and subsequently compared statistically at each scale using the F-test to establish a level of confidence for the differentiation at each scale represented in the measured profiles. The mean square ratios of measurement data were used to determine whether the variation in roughness was statistically significant and to what level of confidence. The scales at which there was a high level of confidence were the ones at which the tools were differentiable. The results of these experiments confirm our previous findings that RLs, over certain scale ranges, can discriminate the stone tool surface wear profiles produced by the different contact materials.