Sun photometer measurements of the direct (collimated) solar radiation provide information to calculate the columnar aerosol optical depth (AOD). AOD can be used to compute columnar water vapor (Precipitable Water) and estimate the aerosol size using the Angstrom parameter relationship. Two data versions (Versions 1 and 2) and three quality levels (Levels 1.0, 1.5, 2.0) exist for each product. While Levels 1.0 and 1.5 are provided in near realtime, the 12month or longer delay (due to final calibration and manual inspection) ensures that the highest quality data can be found in Version 2, Level 2.0 data products.
Spectral DeConvolution Algorithm (SDA) Retrievals  Fine and Coarse Mode AOD and Fine Mode Fraction
The aerosol optical depth processing includes the spectral deconvolution algorithm (SDA) described in O'Neill et al. (2003). This algorithm yields fine (submicron) and coarse (supermicron) aerosol optical depths at a standard wavelength of 500 nm (from which FMF*, the fraction of fine mode to total aerosol optical depth can be computed). The algorithm fundamentally depends on the assumption that the coarse mode Angstrom exponent and its derivative are close to zero. Its advantage lies in the fact that it produces useful indicators of aerosol size discrimination at the frequency of extinction measurements.
The key output products of the SDA are all computed at a reference wavelength of λ = 500 nm. These products include the AOD (τ_{a}, using the nomenclature of O'Neill et al., 2003), the spectral derivative of the AOD (α) and the spectral derivative of a (α'). These three "generic" products are employed in the SDA to compute the principal SDA output products of fine mode aerosol optical depth ( τ_{f}), the spectral derivative of the fine mode aerosol optical depth (α_{f}), the spectral derivative of α_{f} ( α'_{f}), the fine mode fraction ( η = τ_{f} /τ_{a}) and the coarse mode optical depth ( τ_{c} = τ_{a}  τ_{f} ). Note that spectral derivatives can be evaluated at 500 nm (in a differential calculus sense) because we fit the input AERONET AODs with a second order spectral polynomial.
The spectral derivatives α and α_{f} represent, respectively, (i) a generic Angstrom exponent containing information on the combined size contributions of fine and coarse mode aerosols (very strongly related to the typical multiwavelengthregression Angstrom exponent while not being spectrally diluted in terms of allowing the extraction of higher orders of spectral curvature information) and (ii) size information specific to the fine mode size distribution. The theoretical underpinning of the SDA technique is that the aerosol world is assumed to be bimodal; not always true, but it is a hypothesis of surprisingly universal applicability (when its not true it clearly produces an optically equivalent bimodal representation of the real aerosol world; not unlike the more primitive but indisputably relevant regression Angstrom exponent reduces the aerosol world to straight lines in log AOD versus log λ space).
O'Neill, N. T.,T. F., Eck, A. Smirnov, B. N.Holben, S. Thulasiraman, Spectral discrimination of coarse and fine mode optical depth, J. Geophys. Res., Vol.. 108, No. D17, 45594573, 10.1029/2002JD002975, 2003.
* Fine Mode Fraction; the algorithm divides aerosols into fine and coarse mode fractions in an optical sense. This parameter differs moderately from the SMF (SubMicron Fraction) which divides according to a simple microphysical (radius) cutoff.

In July 2005, a number of improvements were made to the direct sun measurement procedures and processing algorithm. Improvements include satellite retrievals of the total column O_{3} and NO_{2} concentrations, pressure using reanalysis data, improved absorption coefficients for trace gases, and updated algorithms. All of the Version 2 improvements can be found in the Version 2 AERONET Direct Sun Algorithm and Spectral Corrections/Components tables.
The prior algorithm and processing (Version 1) will be available through at least 2006.

AERONET Quality Levels
AERONET quality levels are provided in three levels: 1.0, 1.5 and 2.0. Level 1.0 data use the prefield deployment sun calibration. Level 1.5 data use Level 1.0 data and apply a cloudscreening procedure. Data are raised to Level 2.0 after applying the final postfield deployment sun calibration to Level 1.5 data and manually inspecting these data.
Data Quality
+ Calibration
+ Cloudscreening Summary and Full Description (PDF)
+ Quality Control Checklist for Manual Inspection (PDF)
SDA Retrieval Quality Levels
The SDA retrievals utilize the AERONET AOD (Level 1.0, 1.5, and 2.0) products. Level 1.0 SDA products do not apply any restrictions. However, Levels 1.5 and 2.0 SDA impose additional criteria depending on the number of wavelengths, spectral range and air mass dependence of AOD for each wavelength as well as logic to remove outliers.
Criteria for Level 1.5 SDA retrievals:
1. At least three wavelength combinations must include 440 and 870nm with either 490, 500 or 675nm.
2. The AOD for each channel must be greater than or equal to 0.02/m, where m is the optical air mass.
3. Outliers are removed according to the following criterion:
Abs(AOD_{500nm}AOD_{SDA500nm})>(0.02+AOD_{500nm}*0.005)
Criteria for Level 2.0 SDA retrievals:
1. At least four wavelengths must be included for input. The spectral range must be bounded by 380 and 870nm with at least two additional wavelengths between the bounds (e.g., 440, 500, 675nm). The three channel combination of 380, 500, and 870nm will also be accepted.
2. The AOD for each channel must be greater than or equal to 0.02/m, where m is the optical air mass.
3. Outliers are removed according to the following criterion:
Abs(AOD_{500nm}AOD_{SDA500nm})>(0.01+AOD_{500nm}*0.005) 