CARIBOU instrument, a Non Dispersive Infra-Red spectrometer developed at CEA will be installed on a high tower for the intensive campaign of spring 2005. The system is divided in two subsystems:

• The pumping compartment (figure VI.4.1-a left), which includes the pumps required to bring the air into the analyser, and a fridge for preliminary drying of the air.
• The analysis and control  compartment (figure VI.4.1-a right), which includes a Peltier-based cooling system to trap the remaining water vapour of the incoming air, a commercial LICOR-6252 analyser, pressure, flow and temperature regulators controlled by a PLC, and a PC used to configure and control the experiment. The rack containing the LICOR analyser and the hardware necessary to regulate the gas flow, the pressure and the temperature is shown open on figure VI.4.1-a middle.

An overview of the Caribou design can be found below. The lower part of the figure shows the analysis compartment, which will includes a Peltier cooler to achieve an adequate dryness of the incoming air (the Peltier system is at the moment under manufacturing and could be replaced by a commercial cryocooler if not available). The pressure of the reference and sample cells are regulated individually and absolutely around 1080 mbar, and the test results show a relative pressure difference between them lower than 0.1 mbar.

The flows of both the reference and sample gas are regulated at 20 cm³/min, which will allow very long lifetimes for the standard tanks (from half a year for the reference gas to several years for the other standards)

CARIBOU site requirements:

• The size of the control compartment is around 1m80 (height) X 70 cm X 70 cm.
• The air inlet will be located outside the tower, as high as possible.
• The air will be brought into the analyser via a DECABON 1300 tube (1.3 cm in diameter).
• The size of the pumping system is about 1m (height) x 60 cm X 70 cm
• There will be from 6 to 12 high pressure cylinders containing the various calibration gases.
• The total power supply requirement is lower than 1 kW (220 V, one phase)
• A phone line is needed to send the data.

The CARIBOU CO2 analyser is based on a commercial LICOR-6252 CO2 and H2O analyser, which uses the differential, Non Dispersive Infra Red (NDIR) technique. The CO2 spectrum includes a number of absorption features in the thermal Infra Red. The absorption band centered at 4,26 µm is chosen in the LICOR-6252 because this region of the spectrum is quite clear from other features of other atmospheric species.

figure VI.4.1-c depicts the operation principle of the LI-6252: The CO2 measurement is inferred by measuring the difference in absorption of the infrared radiation passing through two gas sampling cells. The reference cell is continuously flushed with a gas of known CO2 concentration, and the sample cell is flushed by the air sample or standard gas to be measured. The rotating chopping shutter allows blocking alternatively the radiation entering the reference and the sample cell, so the single IR PbSe detector alternatively measures the IR intensity coming from the reference and the sample cells. The analyser output is simply proportional to the difference in absorption between both cells. After adequate calibrations of the instrument by known standard gases, the analyser output is converted in CO2 absolute concentrations. The required level of precision for the measurement of the CO2 concentration in the atmosphere, is lower than 0.1 ppm for the CARIBOU instrument. This is far better than what can be achieved using only a commercial LI-6252 from the shelf. To increase the level of precision, the CARIBOU includes the regulation of the physical parameters of the reference and sample gas:

regulation of the LICOR cell pressure at 1080 mbar +/- 0.05 mbar

regulation of the LICOR cells at 35°C +/- 0.03 °C.

Finally, in order to minimise the running cost of the CO2 station, both reference and sample gas flows are regulated at 20 cm3/min.

The typical repetability of the CARIBOU instrument is approximately 0.02 ppm rms (Note: this number is preliminary because CARIBOU is still being characterised in the lab). See figure VI.4.1-d.

The white curve corresponds to the raw, uncalibrated measurements of the same standard gaz, during 3 days. The orange curve is obtained by subtracting the varying offset measured regularly by feeding the sample cell with the reference gas. The standard deviation after offset correction is of the order of 0.02 ppm