Basically a sky - sun photometer consists of a light sensor with an amplifier. The sensor is kept in a sealed metal cylinder (or sensor head) behind a quartz window. A tube for limiting the incident light reaching the sensor is attached in front of the window. By the use of a rotating wheel we can sequentially interpose different interferential filters to select a set of wavelengths for measuring incoming radiance. A programmable robot will aim the optical head at the different required sun and sky positions. This basic instrument can be completed with other features: a diode for keeping the alignment with the sun, solar panels for autonomy, power batteries, and wet sensors to park the instrument when it rains.
There are several commercially available sky – sun photometers in the international market. The two most well known models are the Cimel Electronique CE318 (produced in France) and the Prede POM (produced in Japan). Both instruments are devoted to atmospheric aerosol characterization and they offer very similar features.
Instruments and networks are very intricately related. These are the main differences between both network and instrument concepts:
Radiance sensor: The POM model uses only one sensor and tube for both sky and sun measurements. On the contrary, the CE318 uses two different sensors and tubes. New models of CE318 include now a single detector, but many of the active instruments still work with two sensors.
The single sensor solution allows to directly relate both sun and sky radiometric measurements. Moreover, temporal drifts would affect the sun/sky ratio if a two sensor solution is used. Therefore the single sensor case offers a better temporal stability solution.
Temperature effect: Due to sensor dependence on temperature in the NIR band, raw photometric measurements show a significant dependence on temperature. The POM model is equipped with a temperature controlled optical head. The CE318 standard model is not temperature controlled, but the measurements can still be corrected afterwards.
Unfortunatelly, the temperature correction cannot be completely cancelled as it is slightly dependent on the specific equipment. The temperature factor also shows a slight temporal drift, as highlighted by Campanelli et al. (2007).
Temporal resolution: CE318 radiometers measure with the AERONET standard schedule, broadly an almucantar scenario every 60 minutes and a direct sun measurement every 15 minutes. The POM radiometers measure an almucantar scenario every 10 minutes and a sun direct measurement 1 minute. CE318 can add up to 26 extra daily measurements. This is not possible however for those instruments federated in AERONET.
The CE318 frequency is set by AERONET as a compromise between atmospheric representativity and equipment deterioration. The POM resolution is better suited for studying rapid changes in the atmosphere.
Principal planes: The CE318 instrument measures solar principal planes in addition to solar almucantar planes. They are not routinely measured nor niverted by the POM model and Skyrad code, although they should be measured and inverted in the future.
Principal planes give a wider angular scattering ranges when the sun is high in the sky, so they are better suited than almucantar planes at around noon measurements.
Columnar water vapor retrieval: The AERONET code retrieves routinely the columnar water vapor (CWV) from the 940 nm direct sun measurements. This measurement is performed with the standard POM model, although it is still not processed in the standard packages (neither in ESR).
The CWV information is needed for atmospheric correction of remote sensing products. It will be included in the next versions of the ESR package (M. Campanelli, personal communication, 2008).
Columnar ozone retrieval: So far, neither CE318 nor POM models use the ultraviolet channels information for retrieving the columnar ozone content.
The retrieval of the columnar ozone should be included in next versions of the ESR package. The methodology of (J.P. Díaz et al., 2008) is under evaluation.
Algorithms availability: AERONET developed a well known inversion code (Demonstrat) that is directly appliable to CE318 data. This code is well documented in international literature (Holben et al., 1998). The code is applied to the instrument measurements only in a centralized server in United States, where the measurements are sent automatically. It is not public so standalone CE318 users can not use it, modify it or improve it.
The POM radiometer uses a different algorithm (Skyrad.pack). This code is also well documented (Nakajima et al., 1983; 1996). This is an open source code software, so it have to be used and improved by the user, even when they are not federated in any network.
Calibration: The CIMEL-AERONET calibration procedure is defined by a stringent protocol (Holben et al., 1998). Field instruments must be periodically sent to the network calibration facilities. The periodical return of the equipment allows the calibration and also provides an important opportunity to maintain the equipment, including changing filters, when required. It takes no less than a couple of months to have the instrument back, and the calibration status is checked every 6 to 12 months. Hence the highest quality processing is applied a posteriori, through the re-elaboration of the previous data. However, if the instrument is damaged during this period and a post – calibration cannot be performed, the data will never acquire the highest quality flag.
POM instruments do not need any lamp calibration, since direct and diffuse radiation measurements are performed with the same sensor and their more stable ratio is inverted. For direct sun measurements, an in situ decentralized calibration procedure (SKYIL: SKYRAD Improved Langley plot method) has been developed, that allows operators to monitor and evaluate the calibration status on a continuous basis (Campanelli et al., 2004) keeping the periodic check only for maintenance reasons and therefore reducing the data gaps incurred by the periodical shipments. It also reduces economic cost and unnecessary instrumental risks. Furthermore, the SKYIL method is very useful for diagnosing the condition of a sky radiometer, whose data analysis is very vulnerable to small errors in the measured data. Using SKYIL method, any eventual sudden change or even the natural temporal drift of the calibration constant, can be quickly spotted. Therefore the appropriate corrections can be applied much faster than in Aeronet case, starting short after the period in which the problem occurred.