Calculating Aerosol Data Completeness

Procedures for calculating the aerosol data collection statistics.

This following was taken from the IMPROVE Newsletter Volume 9, No. 3.

Aerosol data collection statistics, reported in the IMPROVE Newsletters, are defined as the total number of successful PM 2.5 Teflon filter samples divided by the total number of possible sample periods. However, each sample period collects four independent filters (i.e., three fine particle samples and one total sample), which are subjected to several analytical methods, so this approach of reporting data completeness will tend to overestimate the completeness of the data with respect to the data needs of the regional haze index. IMPROVE is considering adopting several other completeness indicators to better communicate the completeness of the data.

Two possible completeness indicators that could be used are fraction of scheduled sample days with all major fine particle species and fraction of major sample days with both PM 10 and PM 2.5 mass plus five major fine particle species. The three lines in Figure 1 show these two plus the original method. The third line that shows the aerosol recovery rate for all six factors used to calculate light extinction indicate considerably poorer data recovery than the other methods, especially during the earlier years of IMPROVE. Since both PM 10 and PM 2.5 mass are needed to determine coarse mass concentration, the difference between the two new completeness indicators is the rate of invalid coarse mass sample periods. Note that the drop in completeness in 1998 was associated with a manufacturing change in the nylon filter material that caused 10% of the nylon filters to clog. The other filters were valid. Almost all cases were at eastern sites with elevated particulate loadings. If the extinction can be reconstructed without nitrate for these samples, the completeness fractions would increase by 10%.

While the goal of IMPROVE is to have as high a data recovery as possible, to support the data needs of the regional haze rule, it is more important than ever to quickly discover problems that result in missing data. There are a number of reasons that data are lost, including power outages, damaged filters, operator errors, flow rate errors, equipment failures, analysis errors, and sampling artifacts. Often these causes of missing data are randomly distributed throughout the network, but sometimes problems are seasonal, regional, or site-specific. The University of California-Davis will be closely tracking these causes of missing data. This will increase the emphasis on identifying and addressing problems as soon as possible.

Figure 1. Recovery and completeness of IMPROVE data. The upper line is the recovery rate for PM 2.5 Teflon samples only, rather than completeness. The second line is the completeness rate for all six fine particle species and the lower line is the completeness rate for all six fine particle species including coarse mass.