Mapping air quality: Spatial estimation of pollutant concentrations from point monitoring data.

摘要

New methods for mapping air pollutant concentrations are developed. Point monitoring data are integrated with statistical relationships, physical characteristics, and surrogate data to spatially estimate pollutant concentrations at unmonitored locations. The new spatial estimation methods are used to map quarterly average tropospheric ozone and particulate matter concentrations over the contiguous United States.; Statistical mapping is initially conducted using the traditional technique of inverse distance weighted interpolation. This method is supplemented with a cluster weight that accounts for monitor network spatial configuration and minimizes biases associated with monitor clusters. A second weight based on a monitor's temporal variance restricts the unreasonable spreading of local source influenced concentrations. Barriers are incorporated to account for the physical restrictions on vertical and horizontal air mass flow caused by mountain ranges and the meteorological scale height. Visibility observations and PM10 concentrations serve as surrogate data in the estimation of fine particle concentrations. The higher resolution spatial coverage of the surrogate networks, combined with their strong relationships to fine particle concentrations, aid the estimation between fine particle monitors. Cross-validation error analysis compares the new estimation methods with the traditional estimation techniques of inverse distance weighting and kriging. An uncertainty index is introduced to aid the understanding of estimated concentration reliability.; The new methods improve pollutant concentration estimates in sparsely monitored areas, while they perform equally well to other estimation methods in densely monitored areas. Estimating particulate matter concentrations with barriers decreases concentrations at high elevations and concentrations in valleys are restricted from spreading beyond their valley boundaries. Surrogate aided estimates of fine particle concentrations reflect concentrations where fine particles are monitored but inherit the spatial pattern of the surrogates at locations between fine particle monitors.