Aerosol Models from the Aeronet Data Base. Application to Surface Reflectance Validation

Aerosols play a critical role in radiative transfer within the atmosphere and in climate change. As part of the validation of atmospheric correction of remote sensing data affected by the atmosphere, it is critical to utilize appropriate aerosol models as aerosols are a main source of error. Here, we define the aerosol model by recalculating the aerosol microphysical properties based on the optical thickness at 440 nm and the Ångström coefficient obtained from numerous AERONET sites. The associated uncertainties are up to 23%, except for the imaginary part of the refractive index (about 38%). Uncertainties of the retrieved aerosol microphysical properties were incorporated in the framework for validating surface reflectance derived from space-borne Earth observation sensors. It yields an overall uncertainty of approximately of 1 to 3% of the retrieved surface reflectance in the MODIS red spectral band, well below the specification used for atmospheric correction.


INTRODUCTION
The land surface reflectance is a fundamental climate data record at the basis of the derivation of other climate data records (Albedo, LAI/Fpar, Vegetation indices) and a key parameter in the understanding of the land-surface-climate processes. It is essential that a careful validation of its uncertainties is performed on a global and continuous basis. A possibility is to compare the surface reflectance product to reference reflectance determined from Top of atmosphere reflectance corrected using accurate radiative transfer code and very detailed measurements of the atmosphere obtained over the AERONET sites which allows to test for a large range of aerosol characteristics; formers being important inputs for atmospheric corrections [1] [2] [3] [4] [5]. However, the application of this method necessitates the definition of a very detailed protocol for the use of AERONET data especially as far as size distribution and absorption are concerned, so that validation methods or protocols could be compared [6]. We presented here the protocol with an exhaustive error budget [7].

PROTOCOL FOR GENERATING THE AEROSOL
MODELS FROM AERONET DATA.
The first part was to define a protocol to use the AERONET data. To correctly take into account the aerosol model, we used the aerosol microphysical properties provided by the AERONET network including size-distribution (%Cf, %Cc, rf, rc, σr, σc), complex refractive indices and sphericity. Over the 1139 available AERONET sites, we selected 851 sites with sufficient data (Figure 1). To be useful for validation, the aerosol model should be readily available anytime, which is rarely the case due to satellite overpassing time (no almucantar protocol measurement + partial cloud cover…). Following Dubovik et al., 2002 [8] approach, we used regressions for each of the microphysical parameters using as parameter τ440 and α (Angström coefficient).
Comparisons with the AERONET dataset indicate APU (Accuracy-Precision-Uncertainties) up to 30% less than while using directly Dubovik's 2002 approach for each parameter (with τ550 only).    To go further in detail, a study was performed to estimate the impact of the uncertainties of our aerosol models (represented by the 10 components given Table 1) on the surface reflectance to be used for validation. Table 2 shows results for the MODIS Red Channel and, as expected, the imaginary part of the refractive index (i.e. absorption) generated the highest part of the uncertainties followed by the radius of the fine mode.

REFERENCE FOR THE VALIDATION OF THE SURFACE REFLECTANCES
The second part of the study relies on the theoretical land surface reflectance retrieval. The global uncertainties are reported Figure 4 for MODIS band 1 (red channel) versus the atmospheric reflectance in the MODIS band 3 (blue channel). The new protocol allows us to define a reference (for surface reflectance validation) with an uncertainty in the MODIS red channel (as an example) always lower than 0.004 in term of surface reflectance for the 704 AERONET Sites (which is much lower than required specifications), Figure 5. For a mean loaded atmosphere, t550 less than 0.25, the maximum uncertainty is 0.0025 corresponding to a relative uncertainty (in the MODIS RED channel): U < 1% for rsurf > 0.10 1% < U <2% for 0.10 >rsurf > 0.04)