A tunable infrared laser differential absorption spectrometer (TILDAS) was
used to remotely sense the nitric oxide (NO) emissions from 1,473 on-road v
ehicles. The real-world measurement precision of this instrument in the lim
it of low NO concentration is 5 ppm of the vehicle exhaust, which correspon
ds to a 3 sigma detection limit of 15 ppm. Our analysis of the distribution
of negative concentration measurements produced during this experiment sup
ports this claim, showing that the instrumental noise for this set of measu
rements was at most 8 ppm in the limit of low NO concentration. The high se
nsitivity of this instrument allowed us to measure the NO emissions of even
the cleanest vehicles. The measured vehicle fleet NO emissions closely fit
a gamma distribution with 10% of the fleet contributing about 50% of the t
otal fleet emissions. Newer vehicles had lower NO emissions than older ones
, but high NO emitters were found in every vehicle age cohort. On a vehicle
-by-vehicle basis, NO emissions correlated very weakly with vehicle velocit
y, acceleration, power per unit mass, carbon monoxide (CO) emissions, and h
ydrocarbon (HC) emissions. High NO emitting vehicles could not be identifie
d by remote sensing of CO of HC emissions and vice versa. When we compared
the NO emissions for 117 vehicles measured more than one time, about half o
f the high NO emitters were found to be very consistent, while the other ha
lf varied significantly.