Scattering of electromagnetic radiation by the permanent gases is well understood and the fact that they are well mixed, and that the vertical air mass is fairly uniform over large areas, and the scattering properties are simple means that the contribution of the permanent gases to the at-satellite signal is readily calculated. The situation with regard to the aerosols is more difficult. The aerosol concentration can be highly variable in space and time, with fluctuations on scales as short as a few kilometres (although usually longer). The scattering function is not known exactly, whereas it is for gases, and another factor is the strong forward component to the aerosol scattering, which makes accurate radiative transfer modelling difficult.

Experience with ATSR2 suggests that we can correct for aerosols more readily if the dataset is not simply a multispectral image, but consist of a directional set of such measurements. The multi-directional viewing afforded us by the CHRIS-PROBA system thus makes possible a data-driven atmospheric correction of the satellite images. By observing the surface through different air masses, the separate contributions from the surface and the atmosphere can be accounted for if we make a reasonable assumption about surface reflection properties. The ATSR2 work is described in (Godsalve 1996; Flowerdew and Haigh 1996; North, 1996; Mackay et al., 1998; North et al. 1999). Similar sentiments underlie the approach taken to atmospheric correction of data from the MISR instrument on Terra-1.

Compared to the ATSR2 instrument, CHRIS-PROBA offers additional look angles, more bands and a higher spatial resolution. Extending the ATSR2 algorithms to CHRIS-PROBA data should make the correction both more accurate and, because sub-pixel cloud and heterogeneous land surfaces will be less of a problem, more robust.

In addition to being needed to correct the satellite data to ground reflectance values, aerosol information is important in its own right. For example, it is believed that sulphate particles have had a significant cooling effect over Europe and China in recent decades, counteracting some of the local climate change that would have been expected from observed increases in atmospheric carbon. A particular source of uncertainty in modelling radiation transfer in the atmosphere is the interaction between aerosols and clouds. While some global climatologies exist for aerosols, these are based on limited numbers of observations and only recently has a network of observing stations been set up that tries to monitor aerosol properties over sensible periods and a widely distributed area. However, the existing distribution is heavily skewed towards the continental United States, and there is little coverage of Asia, the southern hemisphere or the open oceans. The only way to obtain estimates over un-instrumented areas is via measurements from space, and this is perhaps the only way that long term, global monitoring of aerosols can happen.

The aim of this science investigation is to develop further the methods developed in the earlier ATSR2 investigations to the more highly specified data from CHRIS. In that earlier investigation the technique was found to be quite robust to the aerosol optical depth, which means the method can be regarded as a retrieval method for aerosols in distinction to its being an atmospheric correction method.

Validation
The method will be tested over sites where a CIMEL sun photometer has been installed, running automatically, under the Aeronet programme. The CIMEL data directly measure the attenuation of the solar beam in a small number of wavelengths, from which aerosol extinction at those wavelengths may be calculated. In addition, measurements of the solar aureole allow estimates to be made of the scattering function of the aerosol. Further, for one of the sites detailed properties of the aerosol are measured at ground level. The sites are chosen to cover the main aerosol types, at least as they are categorised within the atmospheric radiative transfer codes LOWTRAN and 6S. These main types are: marine, continental/rural, desert, and urban. The aerosol types are essentially different mixtures of basic particles such as sulphates, soot, silicates etc. Different mixtures of these, in different humidities, give different particle size distributions and single scattering albedos to the aerosol ensemble. Th role of humidity is important, as the wavelength dependence of the aerosol extinction depends on this: the more humid the atmosphere, the larger the aerosol particles tend to be and the more non-selective, or grey, the scattering tends to be.

Marine sites
Four sites have been identified:

• Rame Head, Plymouth, UK. This is actually on the coast, and is run by the Plymouth Marine Laboratory. It is a SeaWifs calibration site.

• Venice. The Aeronet instrument is in a tower, 5km from the coast. The Venice Lagoon is itself a site of some considerable interest for coastal and sediment transport studies, and we have links to the Department of Environmental Science there, who are studying sediment transport and vegetation ecology within the lagoon. A series of images acquired here could form a basis for a coastal study data set.
• Hawaii. This was chosen because of the continual high humidity, which means the maritime aerosols here has slightly different properties to that encountered in temperate latitudes.
• McMurdo sound, Antarctica. The reasons for picking this site are:
  • Its southerly location.
  • The bright background: the method chosen makes assumptions about the surface that are better approximated the darker the surface.
  • The cleanness of the atmosphere. This and the previous point reflect our desire to fill the "parameter space" of aerosol and surface properties among the test sites.
  • The surface of snowpack is interesting in itself; the BRDF changes shape as the snowpack ages.

Sites for continental aerosols

• The Southern Great Plains site. This site is run by the Atmospheric Radiation Measurement (ARM) Programme and houses the Aerosol Observation System, a suite of instruments that take continuous measurements of aerosol properties at ground level. The list of measurements taken includes:
  • Single scattering albedo
  • Extinction coefficient
  • Backscatter coefficient
  • Angstrom coefficient
  • Number density of particles of sizes between 0.1 to 10µm

  Details of this facility may be found on the website: http://www.arm.gov/docs/instruments/
  static/aos.html

• Munich, Barrax, and the East Anglian site. These are not Aeronet sites, and imagery is being captured here under intensive ground observation for land surface studies. However, during each such experiment atmospheric measurements will be taken and it is proposed to use these data as further test sites for continental aerosols.

Sites for desert aerosols

• Solar Village, Saudi Arabia.
• Tinga Tingana, S. Australia. This second site is chosen because of its southern hemisphere location.

Sites for urban aerosols

• Mexico City. This appears to be the only urban site in the Aeronet network, and has some of the filthiest air in the world. It is important to confirm that the method will be able to retrieve aerosol properties when there is significant absorption by soot and smog.
• Southampton, UK. It is intended that this will form part of a small experiment determining the ability of CHRIS to monitor air quality over towns. Our partners here will be the University of Southampton, who will supply local radiometers and a CIMEL, and AEAT.

Other sites
It is intended to establish an aerosol facility and calibration site in northern Western Australia. The main reason for interest in this site is the strong seasonal changes in the aerosol, and the fact that biomass burning is an important aerosol source at some times of year. Mapping the smoke (as has been done for ATSR2) would be of definite interest.

Data requirements Ground and satellite data over these sites are required monthly over the length of the mission to give a full range of sun-surface -sensor geometries, humidity profiles and seasonal loading effects. A full year's data would be the ideal. A restricted swath will generally suffice with the full spectrum acquired except over Venice, where the whole swath is needed. The full pointing capability of the platform will be exercised in this experiment, and the maximum number of angles will be needed. It is particularly important that at least one measurement be taken at about 60 degrees (at ground).

The default for aerosol studies will be monthly acquisitions of one set of 5 images with along track ground incidence angles of 0°, +/-25° and +/-44° at any single across track angle (n.b. it is assumed that the across track viewing angles will vary month to month primarily to take advantage of cloud free viewing opportunities). The full spectral coverage of 62 bands will be acquired at a spatial resolution of 50m x 50m.

Site	Aerosol Type	Lat/ Long	Observing Schedule	Notes
Rame Head, Plymouth, UK	Marine	N 50 21 W 4 08 A 0	Monthly, throughout mission	NERC funded site; SeaWifs calibration site
Venice, Italy	Marine	N 45 18 E 12 30 A 10	Monthly, throughout mission	Acquisition should attempt always to image the lagoon
Lanai, Hawaii	Marine	N 20 49 W 156 59 A 80	Monthly, throughout mission	High humidity
Great Plains, Oklahoma	Continental	N 36 36 W 97 24 A 315	Monthly, throughout mission	CART site. Direct measurements of aerosols routinely taken
Barrax, Spain Munich East Anglia, UK	Continental Continental Continental		IOP IOP ref MJB	ex-EFEDA   MODIS calibration site
Solar Village, Saudi Arabia	Desert	N 24 54 E 46 24 A 650	Monthly, throughout mission
Tinga Tingana, Australia	Desert	S 28 58 E 139 59 A 38	Monthly, throughout mission
Mexico City	Urban	N 19 20 W 99 10 A 2268	Monthly, throughout mission	Some modification to the timing may be needed, given altitude
Southampton, UK	Urban	N 50 55 W 1 25 A 0-10	IOP	Non - Aeronet. Experiment to be set up with AEAT
McMurdo Sound, Antarctica	Maritime	S 77 37 E 162 52 A 75	Jan-Apr Only Monthly	May not be manned (expense). Bright background
Lake Argyle, NW Australia	Dry season burning		Monthly, throughout mission	May not be ready