AirInfoNow Logo Pima County Seal - link to
PDEQ home page EPA EMPACT Home Page
AirInfoNow in Spanish logo  
Search On-line Survey En Espanol  

Home    About Air Info Now     Activities     For Teachers     For Health Professionals     Ozone Mapping     Data     Frequently Asked Questions

U of A "Regression" View


EPA "Kriging" View




The U.S. Environmental Protection Agency (E.P.A.) has funded many mapping projects across the U.S. to help provide air pollution information in a more timely and effective format for the public. The ozone air pollution maps shown on these pages are similar to other ozone air pollution maps displayed for other metropolitan areas in the U.S. While the goal of providing air pollution maps has been the same across the U.S., the methods used to derive these maps are not always the same.

The methods used to derive the maps shown on these pages are different from the methods used for virtually all other web-based real-time mapping programs. The differences lie in the statistical approaches taken to produce estimates of ozone concentrations some distance from actual ozone monitor locations. Virtually all other ozone mapping programs in the U.S. utilize a geostatistical method known as "kriging". This method produces estimates at a given location on a map by considering a combined weighting of measurements taken at more than one monitor locations nearby. More specifically, the weighting scheme used on the nearby ozone measurements, and ultimately the estimate of ozone concentration, is a function of the actual distance(s) that the measurements are from the location where estimates are to be made. Therefore, a measurement from a relatively close monitor will receive a greater weighting towards the estimate than another monitor that is relatively farther away. Kriging is able to resolve this weighting scheme for many monitors, and produce an estimate that minimizes errors, as well as other statistical measurements. Using kriging to map ozone across a metropolitan area results in a smooth, but continuous surface of estimates for the entire area.

The maps on these web pages are not created using kriging, but rather use a different statistical approach. The methods used to create the maps on these pages utilize multiple factors that contribute to the spatial and temporal variation of ozone measurements at the several ozone monitors in the Tucson metropolitan area. Importantly, these factors are inherently geographic, and include such things as prevailing wind patterns, topography, and the spatial distribution of sources of air pollution across the metropolitan area (e.g. roadways). These factors, which vary across time and space, are used in what is called a regression analysis. From this regression analysis, a model is derived that is capable of providing hourly estimates of ozone at virtually any location in the Tucson metropolitan area.

While it can be seen that the two approaches taken here and elsewhere to mapping air pollution are different, the outcome is much the same. The public is able to access current, updated, and relevant ozone pollution information for the particular area that they live in, without necessarily having an ozone monitor in the vicinity. It is difficult to determine which method is better, but it is argued here that the regression approach allows for an increased 'flexibility' in developing the maps. Importantly, this flexibility may be necessary in areas where the ozone monitors are few and the terrain and layout of urban area(s) are not simple and relatively flat, as is the case in Tucson.

Home  divider  About Air Info Now  divider  Activities  divider  For Teachers  divider  For Health Professionals  divider  Ozone Mapping  divider  Data   divider  Frequently Asked Questions