We observed the 1-h resolution fine (particle diameter less than 2.5 μm) and coarse (particle diameter between 2.5–10 μm) mode NO₃^- and single particle (0.5–10 μm) number concentrations during the long-lasting yellow sand episode that occurred from May–June, 2014, in Fukuoka (Uno et al., 2016), and showed that both the fine and coarse NO₃^- concentrations strongly correlated with the coarse mode particles PM_c (particle diameter between 2.5–10 μm) which is dominated by the yellow sand. The ratio of the concentration of the fine and coarse NO₃^- is 1 : 3, and the coarse NO₃^-/PM_c ranges from 5–8% with R=0.87. This indicated that the NO₃^- was generated from the HNO₃ gas by the heterogeneous reaction on the surface of the yellow sand particle (forming the dust-nitrate) and transported as Ca(NO₃)₂. A chemical transport model including the dust acid uptake processes reproduced both the observed dust and fine/coarse NO₃^- time variations. A source-receptor analysis for the origin of the observed dust-nitrate in Fukuoka showed that the 70–80% of the dust-nitrate originated from the NO_x emission over the wide regions from Beijing to Shanghai, China, and the contribution of the Japanese NO_x emission was less than 11%. Both observations and a numerical analysis showed that the coarse mode dust-nitrate fraction is larger than the fine-mode dust-nitrate. A chemical transport model analysis also indicated that the fine-mode dust-nitrate is an important fraction of the observed fine NO₃^- and showed that the dust-nitrate is also important for PM_2.5 air pollution.

Hourly PM_2.5 mass concentration data during blank test of the automated continuous monitors provided by 21 municipal governments was analyzed. The arithmetic mean concentrations were well maintained within the range of ±5 μg/m³, though the duration time (number of data) differed from 15 to 55 h. The standard deviation tended to differ by type of the monitors, seemingly due to the difference in the measurement principle. Some results showed remarkably large standard deviations, which might be caused by certain instrumental problems. Based on these results, a procedure to evaluate accuracy and precision of the blank test data was proposed, in which the confidence intervals (CI) for mean and standard deviation were evaluated. Practically, the results were rejected if the CI for the population mean (μ) did not cover 0 μg/m³, or the CI for population standard deviation (δ) did not cover initial CI, that was determined from a long-term blank test conducted prior to the routine blank test. The test case study conducted with the confidence level of 95% successfully rejected the results with lowered data or increased dispersion caused by instrumental problems.