A comparison of three federal datasets for thermoelectric water withdrawals in the United States for 2010

Abstract

Historically, thermoelectric water withdrawal has been estimated by the Energy Information Administration (EIA) and the U.S. Geological Survey's (USGS) water-use compilations. Recently, the USGS developed models for estimating withdrawal at thermoelectric plants to provide estimates independent from plant operator-reported withdrawal data. This article compares three federal datasets of thermoelectric withdrawals for the United States in 2010: one based on the USGS water-use compilation, another based on EIA data, and the third based on USGS model-estimated data. The withdrawal data varied widely. Many plants had three different withdrawal values, and for approximately 54% of the plants the largest withdrawal value was twice the smallest, or larger. The causes of discrepancies among withdrawal estimates included definitional differences, definitional noise, and various nondefinitional causes. The uncertainty in national totals can be characterized by the range among the three datasets, from 5,640 m 3 /s (129 billion gallons per day [bgd]) to 6,954 m 3 /s (158 bgd), or by the aggregate difference between the smallest and largest values at each plant, from 4,014 m 3 /s (92 bgd) to 8,590 m 3 /s (196 bgd). When used to assess the accuracy of reported values, the USGS model estimates identify plants that need to be reviewed.