Skip to main content
. Author manuscript; available in PMC: 2021 Feb 1.
Published in final edited form as: Atmos Environ (1994). 2019 Nov 14;222:117130. doi: 10.1016/j.atmosenv.2019.117130

Table A4. Cross-validation results by season.

Correlation coefficient between PM2.5 concentration predictions and observed PM2.5 concentration in μg/m3 stratified by season. The methods considered are: raw CMAQ output, ordinary least squares (“OLS”), inverse distance weighting (“IDW”), universal Kriging(“UK”), downscaler, random forests (“RF”), support vector regression (“SVR”) and Neural networks (“NN”). Methods use either CMAQ and/or other geographic covariates (“Covs”). The last line in the table provides the mean and standard deviation, in parenthesis, for monitored PM2.5 concentration during each season.

Method Winter Spring Summer Fall
CMAQ 0.57 0.60 0.57 0.48
OLS (CMAQ) 0.71 0.69 0.61 0.56
OLS (Covs) 0.80 0.75 0.70 0.69
OLS (CMAQ+Covs) 0.75 0.67 0.63 0.63
IDW 0.89 0.89 0.81 0.83
UK (CMAQ) 0.90 0.90 0.83 0.84
UK (Covs) 0.90 0.89 0.82 0.84
UK (CMAQ+Covs) 0.90 0.89 0.79 0.83
Downscaler (CMAQ) 0.90 0.90 0.82 0.84
RF (CMAQ + Covs) 0.80 0.73 0.69 0.70
SVM (CMAQ + Covs) 0.83 0.81 0.77 0;.75
NN (CMAQ + Covs) 0.84 0.81 0.71 0.76
PM2.5 10.59 (7.05) 9.47 (5.81) 10.82 (5.93) 9.01 (5.74)