Characterization of Uncertainty and Variability of Freshwater Consumption Impacts in Life Cycle Assessment

Abstract

Life cycle assessment (LCA) provides a standardized protocol for estimating a wide range of life cycle technology impacts. LCA is used to make comparisons between alternative technology systems and to identify opportunities for reducing environmental impacts at the local, regional, and global scales. Until recently freshwater as a resource has been neglected from LCA studies. However following record draughts and massive agricultural losses in recent years, the use of LCA, or life-cycle type approaches, have been suggested as a way to identify situations of high water use and possible alternative technologies to reduce water scarcity impacts. Several methods have been proposed in the literature for estimating freshwater consumption impacts--although varying in their overall approach and goal, the distinction between, and quantification of, freshwater use and consumption is common to all methods. Freshwater use refers to any freshwater for which some productive use has been made, and may or may not be available at its origin for further use by other water users. Freshwater consumption refers to freshwater use that is no longer available to other users, such as through evaporative losses. The importance of this distinction is that consumed freshwater can contribute to resource scarcity, unconsumed water that has been used can be recycled and does not contribute to scarcity. Also common to all methods in the literature is the lack of analysis of uncertainty and variability in the estimation of freshwater consumption. This work demonstrates the importance of uncertainty and variability characterization in Life Cycle Assessment, using the study of freshwater consumption for a crop production unit process as an example. A sensitivity analysis and uncertainty analysis is performed on the estimation of two classes of freshwater consumption impacts at the farm level spatial scale: green and blue water (abbreviated as GW and BW respectively). Green water refers to the amount of water consumed that originated as local rainfall, blue water refers to the amount of water consumed that was abstracted from surface or groundwater sources. Thus, in the context of crop production green water is a function of land use and blue water is a function of the amount of irrigation water applied. It is found that green water is most sensitive to precipitation and rate of crop water uptake, with estimates range from +/- 20% and +/-18% of the estimate respectively. Neglecting sparse environmental data such as wind speed and relative humidity can introduce uncertainties of up to 30% of the estimate. Uncertainty in blue water consumption is driven by the amount of irrigation water applied. For cases of under irrigation, uncertainty in blue water consumption is equal to uncertainty in the water application data and averages 18% of the estimate. For cases of over irrigation, the uncertainty in blue water consumption is equal to the uncertainty in green water consumption. For cases of irrigation application matching the crop water demand, the uncertainty compounds, and is equal to 40% of the estimate on average. Through a process known as atmospheric recycling, evaporated water can return to the terrestrial ecosystem as precipitation within days or weeks of having entered the atmosphere, reducing the impact of freshwater consumption. Variability associated with temporal and spatial scales and with boundary selection has a substantial impact on the magnitude of the freshwater consumption impacts, but the consideration of these control volume issues has not been considered in the literature in the context of freshwater consumption impacts in LCA. A bounding analysis is performed to determine the effect of atmospheric recycling uncertainty has on freshwater consumption impacts. Atmospheric recycling can reduce the impact of freshwater consumption from 0 to 80% of the farm level estimate, depending on the region, spatial scale, and temporal scale considered within the control volume. In addition there exists unquantified uncertainty associated with how changes in land cover will effect precipitation, irrigation response, and associated freshwater consumption impacts in the context of LCA. Energy required for irrigation is estimated, specifically for water withdrawal from surface and groundwater sources, and water application by pressure and gravity irrigation systems. It is shown that the energetic cost of water withdrawal and application is higher in regions experiencing freshwater scarcity. It is suggested that further research into the interdependence of water and energy production, known as the water-energy nexus, be considered as an alternative approach to currently proposed freshwater impact characterization methods. It is argued that, although uncertainty associated with energy use for irrigation can as high as 40% of the estimate, particularly for groundwater extraction in areas under water stress, it can be bounded as opposed to the unbounded uncertainty associated with atmospheric recycling. The following files are included with this dissertation as supporting information: * cumulative_plots.zip - contains the plotted results of the Monte Carlo estimates for green and blue freshwater consumption. * Supporting_information_1.xlsx - contains summary statistics for freshwater characterization factors and data and results for energy estimates for irrigation. * Supporting_information_2.xlsx - contains input data for green and blue water consumption characterization factors.