Resource information
Modeled carbon-dioxide (CO₂) emissions from an urban area are validated against direct eddy-covariance flux measurements. Detailed maps of modeled local carbon-dioxide emissions for a 4 km² residential neighborhood in Vancouver, BC, Canada are produced. Inputs to the emission model include urban object classifications (buildings, trees, land-cover) automatically derived from Light Detection and Ranging (LiDAR) and optical remote sensing in combination with census, assessment, traffic and measured radiation and climate data. Different sub-models for buildings, transportation, human respiration, soils and vegetation were aggregated. Annual and monthly CO₂ emissions were modeled on a spatial grid of 50 m for the entire study area. The study area overlaps with the source area of a micrometeorological flux tower for which continuous CO₂ flux data (net exchange) were available for a two-year period. The measured annual total was 6.71 kg C m⁻² yr⁻¹with significant seasonal differences (16.0 g C m⁻² day⁻¹ in Aug vs. 22.1 g C m⁻² day⁻¹ in Dec correlated with the demand for space heating) and weekday-weekend differences (25% lower emissions on weekends attributed to traffic volume differences). Model results were weighted using the long-term turbulent source areas of the tower. Annual total modeled (7.42 kg C m⁻² yr⁻¹) and measured emissions agreed within 11%, but show more substantial differences in wind sectors dominated by traffic emissions. Over the year, agreement was better in summer (5% overestimation by model) vs. winter (15% overestimation), which is partially attributed to climate differences unaccounted for in the building energy models. The study shows that direct CO₂ flux measurements based on the EC approach - if sites are carefully chosen - are a promising method to validate fine-scale emission inventories/models at the block or neighborhood scale and can inform further model improvements.