The final step in a clean air plan is to predict future air quality to demonstrate that we can (if we can) meet the health standards by implementing the measures proposed in the plan. This is done by first projecting the emission inventory into the future, taking into account changes in population, housing, employment in specific business sectors, and vehicle miles traveled. These data are obtained from various sources and the resulting emissions are adjusted to account for regulations and control measures scheduled for implementation during the same time period. Additional adjustments are made to reflect large facilities that are expected to start up, modify, or shut down. The resulting inventory is an emission forecast, and is usually expressed in tons per day of particular pollutants for a given year. Additional steps may be required to determine how the forecasted quantities of air pollution will affect the overall air quality. One way to accomplish this is through computer modeling. A computer model simulates how pollutants disperse, react, and move in the air. The inputs to such a computer model are complex. They include weather patterns, terrain, and the chemical nature of air pollutants.