Improvements to Lawn and Garden Equipment Emissions Estimates for Baltimore, Maryland

摘要

Lawn and garden equipment are a significant source of emissions of volatile organic compounds (VOCs) and other pollutants in suburban and urban areas. Emission estimates for this source category are typically prepared using default equipment populations and activity data contained in emissions models such as the U.S. Environ-mental Protection Agency's (EPA) NONROAD model or the California Air Resources Board's (CARB) OFFROAD model. Although such default data may represent na-tional or state averages, these data are unlikely to reflect regional or local differences in equipment usage patterns because of variations in climate, lot sizes, and other vari-ables. To assess potential errors in lawn and garden equip-ment emission estimates produced by the NONROAD model and to demonstrate methods that can be used by local planning agencies to improve those emission esti-mates, this study used bottom-up data collection tech-niques in the Baltimore metropolitan area to develop local equipment population, activity, and temporal data for lawn and garden equipment in the area. Results of this study show that emission estimates of VOCs, particulate matter (PM), carbon monoxide (CO), carbon dioxide (CO_2), and nitrogen oxides (NO_x) for the Baltimore area that are based on local data collected through surveys of residential and commercial lawn and garden equipment users are 24-56% lower than estimates produced using NONROAD default data, largely because of a difference in equipment populations for high-usage commercial appli-cations. Survey-derived emission estimates of PM and VOCs are 24 and 26% lower than NONROAD default estimates, respectively, whereas survey-derived emission estimates for CO, CO_2, and NO_x are more than 40% lower than NONROAD default estimates. In addition, study re-sults show that the temporal allocation factors applied to residential lawn and garden equipment in the NONROAD model underestimated weekend activity levels by 30% compared with survey-derived temporal profiles.