Cloud Correction and its Impact on Air Quality Simulations Arastoo Pour Biazar 1 , Richard T. McNider 1 , Andrew White 1 , Bright Dornblaser 3 , Kevin Doty 1 , Maudood Khan 2 University of Alabama in Huntsville University Space Research Association (USRA)
Correcting for the radiative impact of clouds corrected 38 ppb under-prediction. (Pour-Biazar et al 2007)
W > 0
W < 0
Photolysis Adjustment (CMAQ)
Cloud top Determined from satellite IR temperature
BL OZONE CHEMISTRY
O3 + NO -----> NO2 + O2
NO2 + h (<420 nm) -----> O3 + NO
VOC + NOx + h-----> O3 + Nitrates (HNO3, PAN, RONO2)
Cloud albedo, surface albedo, and insolation are retrieved based on Gautier et al. (1980), Diak and Gautier (1983). From GOES visible channel centered at .65 µm.
Implementation in CMAQ
Interpolate in between.
Areas of disagreement between model and satellite observation
A contingency table can be constructed to explain agreement/disagreement with observation
Based on Agreement Index Model performance has improved.
The improvements are more pronounced at times that the model errors are larger
While RMSE for temperature is reduced, cold bias has increased and dry bias has decreased. This points to an inherent problem other than clouds in the model that is making the control simulation dry and cold.
Similar to 36-km simulation, for 12-km domain cloud assimilation improved Agreement Index. Using the lateral boundary condition from 36-km simulation with assimilation also improves the model performance.
For 12-km domain, unlike the 36-km, temperature shows a positive bias that for some days is improved by assimilation. RMSE and bias for mixing ratio are improved by using the lateral boundary condition from 36-km with assimilation or directly assimilating GOES observations.
CNTRL too opaque compared to satellite
NO2 photolysis rate
Large differences due to cloud errors
Difference in NO2 photolysis rates for selected days(CNTRL-SATCLD)
Difference in NO2 photolysis rates between control simulation and the simulation using observed clouds (CNTRL-SATCLD) for August 19, 21,22, and 29, 2006. Clouds in control simulation are more spread out and cover large areas (more opaque compared to observation).
Over-prediction of Clouds by CNTRL
Under-prediction of Clouds by CNTRL
Under prediction for higher ozone concentrations is slightly improved due to GOES cloud adjustment.
Night time over prediction is increased in some location while reduced in other locations, but generally it is slightly increased.
Largest Surface O3 Differences Due to Cloud Errors - August 2006 (SatCld-Cntrl)
CONCLUSIONS 2006 (SatCld-Cntrl)
ACKNOWLEDGMENT 2006 (SatCld-Cntrl)
The findings presented here were accomplished under partial support from NASA Science Mission Directorate Applied Sciences Program and the Texas Commission on Environmental Quality (TCEQ).
Note the results in this study do not necessarily reflect policy or science positions by the funding agencies.