Statistical Assessment of the Impact of Local Sources on Regional Representativeness

Description

ARM deploys its Aerosol Observing Systems (AOS) in order to make in-situ measurements that are interpreted as being regionally representative. An ideal measurement site would not be impacted by local sources. However, practical siting constraints such as proximity to utilities, accessibility and housing, make complete isolation from local sources impossible. The presence of near-field emission sources distorts the relationship between the measurements reported and the domain of interest. Local influences can be flagged but only if they can be identified. Short temporal spikes resulting from traffic or power generation are visually obvious in a time record of signals but visual discrimination is subjective, leads to biases and cannot indentify smaller and/or persistent spikes that are not obvious. Developing algorithms to reliably flag local influences is complicated. Correlation of wind speed/direction with coincident spikes in chemical precursors is a more deterministic marker of local contamination. However, neither local winds nor synoptic winds can identify local wind shear or turbulence resulting from topology or structures. Assumptions that identify spikes from the ratios of measurements are risky. Large, local emissions from known sources are evident at the ENA AOS in the Azores and at the AMF2 AOS in AWARE. A statistical analysis of the amplitude, duration and frequency distributions of obvious large spikes is used to bound the influence of smaller amplitude, but likely more frequent, spikes which are not as obvious. The ENA AOS includes an Ultra-High Sensitivity Aerosol Spectrometer (UHSAS) which provides particle-by-particle sizing at 1 Hz. We statistically analyze these measurements in conjunction with high temporal resolution (~10-s or faster) measurements from the Particle Soot Absorbance Photometer (PSAP), extinction measurements from the Cavity Attenuated Phase Shift (CAPS), number concentration data from the Condensation Particle Counter (CPC) and high precision carbon monoxide measurements to show patterns in the local influences. We suggest the applicability of pulse height analyses of obvious plumes only to make an estimate of the distortion of the ‘background’ regional levels inferred from AOS measurements and not to provide a satisfactory correction for local emissions. Such estimates are crucial to make siting decisions, as a guide to modelers incorporating measurements in models and ultimately to the science.