The Precision of Precipitation Chemistry Measurements in the Canadian Air and Precipitation Monitoring Network (CAPMoN)

Precision estimates are presented for precipitation chemistry and depth measurements made by the Canadian Air and Precipitation Monitoring Network (CAPMoN). The estimates were made for daily measurements of ion concentration and precipitation depth as well as for weekly, 28-day, seasonal and annual precipitation-weighted mean concentrations and depths. The data on which the estimates are based were collected from collocated samplers at five CAPMoN sites during the period 1985 to 1993. The data pairs from the collocated samplers were used to calculate the between-instrument error defined as 1/2 times the difference between the paired sample concentrations (or depths). For all of the ion concentrations and depths, the between-sampler errors were found not to be normally distributed, but the normality of the distributions improved with the length of the (volume-weighting) time period considered. A set of quantitative measures of overall network precision were derived in absolute (mg L^-1) and relative (%) units. These included the Modified Median Absolute Deviation (M.MAD), the P90% probability values and the Coefficient of Variation (CoV). The latter, defined as the percent ratio of the M.MAD to the median concentration (or depth), represents the relative precision at the center of the error and concentration (and depth) distributions. Based on the CoV values, the relative precision of the CAPMoN measurements was very high (better than 4%) for SO ₄ ^2- , NO ₃ ^- , pH, H⁺, NH ₄ ⁺ , sample depth and standard gauge depth, and not as high (between 10 and <35%) for Cl^-, Na⁺, Ca²⁺, Mg²⁺, and K⁺. The ions with the lowest median concentrations had the poorest relative precision since so many of the concentrations were at or near the analytical detection limit. Except for the sample and standard gauge depths, both the absolute and relative precision improved with the length of the precipitation-weighting period. Detailed statistical testing established that the precision of the daily measurements is dependent on a number of factors, the most dominant being sample depth and concentration, i.e., the absolute precision improves with increasing sample depth and decreasing concentration. The strength of these relationships diminished with the length of the precipitation-weighting period being considered. Laboratory-related sources of imprecision were found to account for less than 4% of the overall daily measurement imprecision for most species, while field-related sources of imprecision accounted for the balance. Specialized plots are shown which allow data users to estimate the absolute and relative precision at any concentration and depth value.