Geostatistical interpolation of ozone exposure for southeastern forests

To assess the impact of ozone on forests, it is necessary to quantify ozone exposure on regional scales. Since ozone monitoring stations are widely scattered and mostly concentrated in urban and suburban areas, spatial interpolation is necessary to estimate ozone exposure at non-monitored forest sites. However, ozone precursor emissions, wind transport of precursors and ozone, and weather conditions conducive to the formation of ozone exhibit considerable spatial variability. These factors are not considered in simple interpolation between monitored sites. We defined an index of ozone formation potential, which incorporated spatially distributed NO{sub x} emissions, wind directional frequencies, distance downwind from NO{sub x} sources, and a measure of solar radiation which drives the photochemical production of ozone. This index was computed for each 2Ox2O km grid cell within the range of loblolly pine (Pinus taeda) in the southeastern U.S. Monitored ozone exposure was quantified by SUM06, the monthly sum of hourly concentrations {>=}0.06 ppm. SUM06 was interpolated using cokriging with the ozone formation potential index as a covariate. For comparison, interpolations without the covariate were also done with inverse distance, inverse distance{sup 2}, and kriging methods. The incorporation of this additional information on ozone formation potential slightly increased predictive capability formore » ozone exposure, with precision and accuracy increasing in the order INVD{sup 2} < INVD < kriging < cokriging.« less