AOD product | WG2 - Algorithms | Forum

Hi all!

As you already know from the Azores and Edinburgh meetings, within the framework of COST Action 1207 and the WMO-CIMO Testbed for Aerosols and Water Vapor Remote Sensing Instruments (Izaña), at the RBCCE we have been working, in collaboration with other participants of EUBREWNET (Thomas Carlund and Henri Diémoz, to cite just two names), on the development of the AOD algorithm for the network.

We plan to start coding the AOD product in the server very soon. To make things easier, it would be better to have the different levels of the product defined in advance. So, to start somewhere, we have come up with the following:

* Level 0												
   Taken directly from the Brewer (IOS) program											
												
* Level 1												
   1) Ozone from the L1.5 product, with the standard Brewer Rayleigh correction replaced by the one produced by Bodhaine's coefficients
																				
   2) Corrections to the counts:											
      a) Same as in the ozone: Individual (not summaries!) raw counts with dark counts and dead time corrections, plus ozone L1.5 data filters (these counts do NOT include the standard Brewer corrections for temperature, Rayleigh, and filters)										
      b) AOD specific:	
         i) Temperature correction with absolute temperature coefficients (not available right now, use the relative ones from the ozone configuration)										
	 ii) Filter correction, with spectral attenuation coefficients for each filter										
	 iii) Earth-Sun distance correction																				

   3) AOD calculation (uses an ETC matrix, with one calibration constant for each wavelength and filter)										
      a) Rayleigh correction with the spectral Rayleigh coefficients from Bodhaine's prescription (to start, we will use the climatological pressure as in the ozone, but might change to a reanalysis value at a later date)
      b) Ozone correction with the spectral Ozone absorption coefficients										
												
* Level 1.5											
   1) To the AOD Level 1 product, add the AOD-specific data filters and corrections:											
      a) AOD data filter based on the standard deviation of each group of 5 observation (limit is 0.02, following Gröbner 2004)										
      b) Polarization correction (currently, Cede et al. but may change to Diemoz and Virgilio in the future)										
												
   2) Still to be developed:	
      a) Stray light correction										
      b) Standard lamp correction (can be used somehow to track changes in the AOD configuration?)										
      c) Other corrections and filters										
																			
* Level 2.0												
   1) Ozone from the L2 product																					
   2) AOD configurations validated against Brewer/PFR/AERONET reference

We would like to have your input -- do you agree with the general layout of the levels? do you miss some correction/filter? if you have experience with other AOD products,how does this compare to it?
and of course anything else you come up with.

We already got some suggestions from Thomas Carlund, which have been already included above, and Stelios Kazadzis, who points that:

I would put all level 1.5 corrections under level 1 since now level 1 AOD calculation is unusable and AOD has to be recalculated after the steps presented in level 1.5.
SO I would put
L1 output:  only corrected signals
L1.5: final AOD with preliminary calibration (including all corrections)  
L2: AOD with final calibration
The need of a new calibration could be identified some months later than the actual measurement. (see aeronet example below)
 
Another issue is the cloud flagging mentioned in level 1.5.1 a).
It has to be defined if the AOD cloud flagging principles will be the same as the ozone acceptance/rejection principles based on the standard deviation of the group of 5 observations.
Based on the fact that you need the ozone to derive AOD , AOD cloud flagging has to be the same or more strict (e.g. cirrus clouds cases).
 
Since you are mentioning aeronet data.
Level 1.5 data include all corrections and the cloud flagged final AOD product. So someone can use them more or less real time.
Level 2 data are calculated much later when the instrument is re-calibrated and they are the final data.
I think in the Brewer cases if someone wants to follow this would have to finalize everything under level 1.5
and then after x months later when you re-calibrate the instrument and determine new ETC’s, you go back and recalculate all AOD data again as level 2.
 
 
Concerning the ETC part:
“the ETC calibration is a matrix for filter and wavelength”
I would work with converting all brewer counts of all filters to nd 0 by having a conversion function Counts (wl,filter)= f(nd(wv,0)).
I suppose that’s how you have calculated ETCs for all filters.
Using ETCs calculated from filter conversions includes an additional step of the convolution of the ETC to the brewer slit at a specific wavelength and for different nd filiters.
But I think maybe this is a detail or you have calculated the ETCs for different nd filters with some other way.