The POLDER products are the optical thickness and the Angstrom exponent of the accumulation mode for a given refractive index (1.47 - 0.01 i). Such products are not directly measured by AERONET photometers, but they can be computed using the particle size distribution and the refractive index deduced from Dubovik's algorithm.

The POLDER algorithm is more effective when the polarized aerosol signal is important and when the ground contribution is rather low: the first case occurs with anthropogenic particles, which exhibit a predominant accumulation mode (rather biomass burning, urban or industrial pollution than dust). The second condition corresponds to observations rather over vegetation than over deserts.

An example is presented over the Mongu AERONET site (Austral Africa) where aerosols are produced from biomass burning. The comparisons are made considering the 7 months with different filters on the data (simultaneous measurements, cloudless atmosphere). On the left, the POLDER optical thicknesses are plotted versus the AERONET optical thicknesses of the small particles (radius r less than 0.6 µm). On the right, the Angstrom exponents are compared and they exhibit a POLDER overestimation.

Below, on the left, the synthesis of the optical thickness comparison is presented over sites with biomass burning aerosols located on Amazonia (Abracos Hill, Alta Floresta, Rio Branco), Austral Africa (Mongu) and South-East Asia (Omkoi).

Above, on the right, the same comparison is shown over regions where urban or industrial pollution dominates. In this case, the results are generally more scattered because aerosol amounts are weaker than in biomass burning events.

Comparisons over African desert sites (Agoufou, Banizoumbou) are very bad, with a large underestimation of the POLDER optical thickness (on the left). In the accumulation mode (r < 0.60 µm) all the particles are taking into account to compute the AERONET optical thickness; but as the mineral aerosol are non spherical, a lot of these particles (the biggest ones) do not polarize and do not contribute to the POLDER aerosol thickness. In fact POLDER only detect the smaller spherical particles: the comparison is strongly improved if we only consider particles with r < 0.35 µm to compute the AERONET optical thickness (on the right).

Same improvement is found for the Angstrom exponent successively computed with r < 0.60 µm (red dots) and r < 0.35 µm (blue dots).

Finally, we compare the optical thicknesses of the accumulation mode when small particles are detected with the photometers (AERONET Angström exponent of the total size distribution greater than 1.5): the result of the linear fit (above, on the left) on the small mode optical thickness is
This result shows the POLDER capability to measure the optical thickness of anthropogenic aerosols and more generally the optical thickness of the smallest aerosols (r < 0.35 µm).

Comparison with MODIS TERRA

Over land, the main problem is to correct the measurements from the spectral ground contribution. In the MODIS algorithm, the ground reflectances are deduced in the visible channels (670 and 470 nm) from empirical relations using the measured reflectances at 2.13µm (Kaufman). The algorithm uses four basic aerosol types (continental, dust, two anthropogenic models with strong or low absorption) and possible mixing between dust and anthropogenic aerosols. In this last case, the aerosol fine mode fraction gives some information on the dust content.
A fraction equal to one means that a model of small particles is chosen, which often occurs (see MODIS synthesis map) except close to desert areas. In this case, we can directly compare POLDER and MODIS optical thicknesses.

Aerosols fine mode Fraction
http://lake.nascom.nasa.gov/www/online_analysis/movas/monthly/index.shtml

Above, a first comparison is presented on the monthly syntheses of the small mode optical thickness at 550 nm. We only focus over land and mainly where small aerosol are detected by MODIS (red zones on the previous map). We can observe a good agreement over China and Central Africa where small particles dominate.

Then, we have compared daily optical thicknesses when both sensors observed the same pixel quite simultaneously, that happens with four successive orbits, every four days, as shown of the both pictures above.

Optical thicknesses are compared but we only considered small particle models (as detected by MODIS) and cloudless pixels. Two examples are presented over Africa and Amazonia.

It is also interesting to compare the aerosol optical thicknesses of the accumulation mode on each side of the seacoasts, where the results are derived from very different POLDER algorithms based on polarized measurements over land and on total radiances over ocean.

A good continuity of the small mode optical thickness is shown on June 2003 over three coastal regions: East Asia with large amount of small particles, Guinea Gulf with biomass burning aerosols and North-East America with pollution events.

This continuity is also shown on daily products concerning the small particle mode. The previous map presents the continuity of the Aerosol Index (A.I. = optical thickness times Angstrom exponent) over Mexico.