Assessment of CMAQ model outputs using remote sensing

Introduction: With the dramatically climate changing we are facing today atmospheric monitoring is of major importance. Several atmospheric monitoring instruments are used for measuring atmospheric composition, optical coefficients, PM2.5, aerosol optical depth, size distribution, PBL height and many other parameters. However an inexpensive method of determining these parameters is by use of models and one model that depicts the aerosol dynamics in the atmosphere is the Community Multi-scale Air Quality (CMAQ) model. Our paper is focused on converting CMAQ retrieval outputs into optical coefficients that can then be  comparing the lidar, AERONET and TEOM measurements performed at City College of the City University of New York . Differences between the full approach and parameterized methods such as the MALM formula used in AIR-NOW are observed and comparisons with AERONET show the full modeling is in general superior to the MALM formula [1] .

Results:  Several parameters have been used to validate the CMAQ outputs including backscatter coefficients, aerosol optical depth and PM2.5 concentrations.  The comparison of the aerosol optical depth  AOD values measured at 500nm by the CCNY sunphotometer with the CMAQ retrieved AOD indicate agreements 10% error on the first 2 days of analysis (January 21^st and 23^rd, 2009). To be noted that on these two days the PBL is stable with few scattered high altitude clouds as indicated in figure 2 a & c.  The larger difference in AOD on the other 2 days (December 8^th and November 21^st of 2008), shown in figure  1 below, can be explained by the low scattered clouds on day 3 and a slightly clearing PBL  towards the end of day 4 of analysis (which are not depicted by CMAQ). The MALM AOD also plotted for these days indicate the same trends but underestimate the sunphotometer measurements and the CMAQ model on all the days.

Figure 1. Sunphotometer, CMAQ and MALM AOD comparison over the 4 days of analysis

For more extensive validations we also took an insight into the Sunphotometer-CMAQ comparison versus wavelength. As can be observed in figure 2 we noted that there is a good agreement between these values with the exception of the AOD levels at lower wavelengths (340nm and 400nm).

Figure 2. Sunphotometer and CMAQ AOD comparison versus wavelength over the 4 days of analysis

Conclusions: In order to validate CMAQ vertical model data, we have  developed a suitable interface in which to use CMAQ outputs directly than some seasonal average based on climatology. We find in particular that integration of the CMAQ microphysicval outputs give reasonable results although direct comparisons with MALM's formula are often different with MALM in general underestimating the extinction cross-section at 500nm.  The need for the general model is further borne out with total column extinction comparisons. In particular, the AOD comparison indicates good agreement between sunphotometer measurements and the CMAQ model on days with PBL stable height unlike the days with dynamic PBL. On the other hand, we note that the TEOM PM2.5 trends are closely followed by the CMAQ model. The CMAQ backscatter coefficients are also in the lidar retrieval range. However the CMAQ model is not fully capturing the dynamics of the PBL on the days with large variability in the height and PBL aerosol load.

ACKNOWLEDGEMENTS

This work is supported by grants from NOAA #NA17AE1625 and NASA #NCC-1-03009.

References:

[1] Binkowski,  F.S. “Aerosols in models-3 CMAQ” , EPA/600/R-99/030, (2005).