Evaluating IMPROVE PM_2.5 element measurements

Figures & data

Figure 1. Timeline of elemental characterization methods employed at the University of California, Davis.

Table 1. Summary of IMPROVE monitoring locations included in this study.

Site Name	Site Code	Lat/Long	Samples Analyzed	Sampling Date Range
Great Smoky Mountains National Park	GRSM1	35.633 N, 83.942 W	940	June 3, 1995 to November 30, 2009
Point Reyes National Seashore	PORE1	38.122 N, 122.909 W	560	April 3, 2004 to November 30, 2009
Mount Rainier National Park	MORA1	46.758 N, 122.124 W	232	October 24, 2007 to November 30, 2009

Figure 2. Scatterplots of the XRF measurements by the two instruments for the twenty elements with at least 50 pairs measured above three times the reported MDLs. The black, dotted lines show the reported MDLs for each element, and the gray dashed lines show 1:1 agreement.

Figure 3. Scaled relative differences for Na, Mg, S, and to a lesser extent Cl and K, show strong relationships with the sum of elemental loadings for the CuMo XRF instruments. Al and Si are impacted by another issue, peak overlaps in the Cu-anode detector, which confounds the loading effect.

Table 2. Number of pairs where each value is greater than or equal to three times the MDL, Theil-Sen regression parameters, precision, and bias for elements with greater than 50 pairs meeting the MDL criterion. The elements P, Cr, Rb, and Zr were excluded by this criterion. Total number of pairs for each element was 1,732. The replicate analytical precision from 2021 and 2022 routine IMPROVE network samples are listed for comparison.

Element	Number of Pairs (> 3 x MDL)	Slope (E5/CuMo)	Intercept (ng cm^−2)	Precision (%)	Bias (%)	Replicate Precision (%)
Na	1236	0.57	499	34	−27	20
Mg	728	0.81	65.8	29	−6	18
Al	1119	0.57	62.7	45	−12	8
Si	1633	0.34	37.1	35	−92	4
S	1729	0.94	84.3	12	−10	2
Cl	482	0.90	−25.4	31	−30	9
K	1729	0.93	13.8	10	−4	2
Ca	1719	0.88	−9.23	14	−22	3
Ti	1405	0.95	0.45	13	−2	9
V	779	0.96	0.01	16	−2	19
Mn	691	1.08	6.49	17	47	11
Fe	1684	0.99	1.56	12	−1	3
Ni	325	0.87	0.41	13	−2	19
Cu	575	1.02	4.09	16	33	11
Zn	1595	1.04	−0.05	11	1	9
As	285	0.59	2.66	25	9	25
Se	912	0.98	0.74	14	5	17
Br	1419	0.78	−1.63	15	−43	14
Sr	686	0.68	2.16	24	10	17
Pb	760	0.99	6.46	17	25	18

Figure 4. Comparisons of concurrent measurements for related species reported in IMPROVE. (a,b) compare S and Cl by XRF to sulfate and chloride by IC, respectively. (c) compares Na by XRF to chloride by IC. (d) through (f) compare crustal elements measured by XRF. The gray dashed lines in (a,b) have a slope of one while (c) has a slope of 0.65, the molar ratio of sodium to chloride in NaCl. The gray dashed lines in (d) through (f) represent the bulk continental crust elemental ratios (Taylor and McLennan 1995).

Figure 5. Scatterplot highlighting the relationships observed in the sulfur-sulfate ratio with sample loading. The horizontal dashed line represents perfect agreement. The sample loading is estimated using the sum of all elements measured by the CuMo instruments.

Taylor, S.R., and S.M. McLennan. 1995. The geochemical evolution of the continental crust. Rev. Geophys. 33 (2):241–65. doi:10.1029/95RG00262.

 Web of Science ®Google Scholar

Data availability statement

All IMPROVE data is publicly available through the US EPA Air Quality System (https://www.epa.gov/aqs) as well as through the US Federal Land Manager Environmental Database (http://vista.cira.colostate.edu/Improve/improve-data/). Reanalysis data is available upon request.