Could there be clouds observed by the LIDAR? Could it be the noise of the LIDAR causing the higher "AOD"?

The method you describe is widely used to estimate the lidar ratio. You are right the predicted lidar ratio is too low. I can only assume that your lidar analysis is not correct.

Examples of lidar ratios for different aerosol types can be found here:

http://www.atmos-meas-tech.net/6/1397/2013/amt-6-1397-2013.pdf

and here

http://onlinelibrary.wiley.com/doi/10.1029/2006JD008292/abstract

You can try to apply AOD from AERONET to lidar inversion, when AOD from lidar has to have the same value. Then other paramers like lidar ratio and extinction coefficient at reference distance can be determinated in opposite way. Maybe this way you will find what is wrong. Ofcourse the measurement has to be done during clear sky.

no. lidar ratios as low as 15 sr have no physical meaning (20 sr are typical for maritime air masses). Mind also that AOD from aeronet depends on the quality level (1.5 level, may contain clouds). Cimel photometer may alos detect different air masses depending on the solar zenith angle! Surely, you have to reevaluate your lidar signals, depending on the atmos. background noise you have. Additionally, your lidar ratio is not constant along the vertical.

Our Lidar is a backscatter lidar, operating at 355nm, pointing at zenith. we usually operate from 10am to 5pm (local time). The main aerosol we observe are the urban pollutant, or some researcher termed it as south east asia aerosol, since I'm from Malaysia, where it is part of south east asia.

According to literature, lidar ratio for south east asia aerosol should be around 50 sr for 532nm lidar, and lidar ratio at 355nm should be slightly higher. So the first guess I take is 60 sr, but the Lidar's AOD is quite higher compared to AERONET's AOD.

For now the best quality data I can get from AERONET for my area is level 1.5 data, so i have to use that one.

For lidar data processing, I assume the lidar ratio to be constant with increasing height, and the range of signal taken into account to calculate the AOD is from 220m (just above the overlap height of our lidar) to 5km, and the reference height lies between this range. The region above the reference height is assume to be molecular sky. Is there any problem that might come from this part, causing the AOD calculated from lidar to be higher than that from AERONET?

Nope. My lidar operate at 355nm. So I take 340nm and 380nm AOD data from AERONET to compare.

One of the crittcal point in the AOD evaluation from lidar is the overlap region. How did you take it into account? When overlap region is not correctly taken into account, resulting AOD is typically too high. Lidar ratio of 15 sr is too low unless you have clouds or marine aerosols. For pollution 60sr but also hgiher values could be reasonable. See values reported in literature but consider also the natural variability of lidar ratio also for the same kind of aerosol (see for example papayannis et al jgr 2008) and more important the vertical variability. Automatic processing of backscatter ldiar as CALIPSO already uses different lidar ratio value for the various identified layer within the observed vertical column.

Assuming that the aerosols are horizontally homogeneous, you are right,

Aeronet AOD should be larger than lidar derived AOD.

Could you decribe with more details your calculations?

How did you set the reference value in the Klett retrievals?

Did you used the Rayleigh calibration method? The SNR was sufficient

at the height of the reference value? How long did you average over time

for improving the SNR?

What I mean is that 220m is the overlap height of my lidar. 220m and below i do not take it into account.

As AERONET provides the spectral AOD measurements, one can know the predominant aerosol type basing on Ansgtrom exponent with cluster analysis. Then basing on the aerosol type, you can find the possible range of lidar ratio that can be used to derive the extinction coefficient as a function of altitude and thus obtaining AOD from the lidar. In your case, AOD from lidar is higher implies that you may be including signal noise while applying Klett's method. Below are some references which can provide you with more information on aerosol type classification from AERONET measurements.

1. Giles, D. M., B. N. Holben, T. F. Eck, A. Sinyuk, A. Smirnov, I. Slutsker, R. R. Dickerson, A. M. Thompson, and J. S. Schafer (2012), An analysis of AERONET aerosol absorption properties and classifications representative of aerosol source regions, J. Geophys. Res., 117, D17203, doi:10.1029/2012JD018127.

2. P. B. Russell, R. W. Bergstrom, Y. Shinozuka, A. D. Clarke, P. F. DeCarlo, J. L. Jimenez, J. M.Livingston, J. Redemann, O. Dubovik, and A. Strawa (2010),

Absorption Angstrom Exponent in AERONET and related data as an indicator of aerosol composition, Atmos. Chem. Phys., 10, 1155–1169,

www.atmos-chem-phys.net/10/1155/2010/

AOD obtained by AERONET is a passive method using solar radiation as light source, and aerosol can interact with radiation by absorption and scattering, but Lidar isa n active method using the 355nm green LASER to interact with atmospheric aerosol and measure backscattering. With this different concept, AOD obtained by AERONET should be lower than Lidar, and more affected by clouds and black carbon which have strong absorption.

Actually, my lidar data processing is done with software that comes together with our lidar system. It is done using Inver Klett-Fernard method to calculate the backscatter coefficient. From there, together with the predicted lidar ratio we get the extinction coefficient and finally the AOD. From the software, we are asked to input the predicted lidar ratio (which are assume constant with height), the reference height, the backscatter ratio (the ratio of total backscatter to molecular backscatter) at reference height and the range of interest. The background noise subtraction had already done automatically by the software.

So for my case, i predict the lidar ratio to be 60 sr (lidar ratio for south east asia aerosol is around 51 sr for 532nm lidar, so it should be slightly higher for 355nm lidar, so I take 60 sr).

Next, the range of interest was taken from 220m up to 5km, where 220m is the overlap height of our lidar, so data below that height i do not take into account.

Then, i plot the natural logarithm of the range corrected signal. The height where the first 0 value come in will be taken as the reference height.

The backscatter ratio i assume to be 1, as i expect it will be molecular sky beyond the reference height.

5 km could be also too low as reference height where the aerosol contribution is assumed as zero. But this depends on the actual atmospheric situation. Forcing the aerosol cotnribution to 0 at those altitudes could lead to a too high backscatter at lower alttiudes. You could try to modify the reference value keeping still 60 sr, just to see if the situation improves.

Why don't you invert the lidar signal using the Klett algorithm by yourself?. In this way you could use the Aeronet AOD as a closure for the inversion, thus avoiding the climatologic value of the lidar ratio.

Aeronet measurements are columnar, so that if different aerosol types cohexist in the atmospheric column, the assuption of soemthign that can be a "mean along the column" lidar ratio could result in also larger error than of the climatological value. The backscatter profile can provide the indication abotu the presence or not of different layers.

OK. And what about an iterative process to:

a) use the AOD value as a closure for the extinction coefficient

b) obtain a lidar ratio profile at the same time

I'm running some tests and it's quite computationally expensive, in particular the lidar ratio profile. I'm using Matlab's fminsearch

You say that you mainly observe urban pollusion. Could it simply be that you are closer to the source of the said pollutants than the AERONET station, so that you could actually have a higher AOD at your location on the average?