Figure 9: The coefficients of determination for each grid cell for two pollutants, based on the testing dataset. Left: SO 2. Right: PM 2 .5 . The black
points are outliers, which were removed to enable useful scaling. For SO 2 ,
these are points with values below .3; for PM 2 .5 , this is a single point with
value .005.
any of the inputs.
5
Discussion and Implications
The pollutants modeled best by this approach are the chemical pollutants
associated with passenger car emissions: CO, NO 2 , and the secondary pollutant 03. NO 2 and 03 in particular have few other urban sources, so their
presence is understandably well predicted by traffic patterns.
One explanation for the higher error in the PM 2 .5 models could be that
PM 2 .5 has a greater diversity of sources than the chemical pollutants, many
Figure 9: The coefficients of determination for each grid cell for two pollutants, based on the testing dataset. Left: SO 2. Right: PM 2 .5 . The blackpoints are outliers, which were removed to enable useful scaling. For SO 2 ,these are points with values below .3; for PM 2 .5 , this is a single point withvalue .005.any of the inputs.5Discussion and ImplicationsThe pollutants modeled best by this approach are the chemical pollutantsassociated with passenger car emissions: CO, NO 2 , and the secondary pollutant 03. NO 2 and 03 in particular have few other urban sources, so theirpresence is understandably well predicted by traffic patterns.One explanation for the higher error in the PM 2 .5 models could be thatPM 2 .5 has a greater diversity of sources than the chemical pollutants, many
การแปล กรุณารอสักครู่..
