Common use of PRECISION Clause in Contracts

PRECISION. Collocated data was collected and analyzed for the period 4/11/01-6/30/01. To remove the loading-dependent artifact from the collocated instrument comparisons, the data was censored to retain only those 5-minute records with attenuation difference less than 5 units between the instruments. Furthermore, only those records with arithmetic mean concentration greater than ten times the MDL were retained, where the MDL value is assumed to be 0.050 μg/m3 and is based on three times the standard deviation of dynamic blank measurements. The results are summarized in Table 4-4. There is an improvement in the collocated precision when the data is censored to include only those records with similar attenuation for both instruments; with a four-fold worse precision when including all data. The difference between these two collocated precision values demonstrates the error introduced by the loading-dependent effect, and will typically vary by location and time of year. While it should be not be interpreted as a representative measurement error, it will be present in the data unless appropriate methodologies are used to compensate for the loading effect. For this reason, record-specific attenuation (ATN) values have been reported along with the mass concentration values for the data set submitted to the NARSTO database. Table 4-4. Collocated 5-minute PM2.5 Aethalometer BC measurements. Parameter Method Collocated N Absolute Precision (μg/m3) Mean Conc. (μg/m3) Relative Precision BC(a,b) ▇▇▇▇▇▇▇▇ ▇▇-16 15,338 0.313 1.232 25.4 % BC(a,c) ▇▇▇▇▇▇▇▇ ▇▇-16 1,042 0.249 1.353 18.4 %

PRECISION. Collocated data was collected and analyzed for the period 4/11/01-6/30/01. To remove the loading-dependent artifact from the collocated instrument comparisons, the data was censored to retain only those 5-minute records with attenuation difference less than 5 units between the instruments. Furthermore, only those records with arithmetic mean concentration greater than ten times the MDL were retained, where the MDL value is assumed to be 0.050 μg/m3 and is based on three times the standard deviation of dynamic blank measurements. The results are summarized in Table 4-4. There is an improvement in the collocated precision when the data is censored to include only those records with similar attenuation for both instruments; with a four-fold worse precision when including all data. The difference between these two collocated precision values demonstrates the error introduced by the loading-dependent effect, and will typically vary by location and time of year. While it should be not be interpreted as a representative measurement error, it will be present in the data unless appropriate methodologies are used to compensate for the loading effect. For this reason, record-specific attenuation (ATN) values have been reported along with the mass concentration values for the data set submitted to the NARSTO database. Table 4-4. Collocated 5-minute PM2.5 Aethalometer BC measurements. Parameter Method Collocated N Absolute Precision (μg/m3) Mean Conc. (μg/m3) Relative Precision BC(a,b) ▇▇▇▇▇▇▇▇ ▇▇-16 15,338 0.313 1.232 25.4 % BC(a,c) ▇▇▇▇▇▇▇▇ ▇▇-16 1,042 0.249 1.353 18.4 %