+ AERONET Home
 + SolRad-Net Home

   SolRad-Net (Solar Radiation Network) is an established network of ground-based sensors providing high-frequency solar flux measurements in quasi-realtime to the scientific community and various other end users.  This network was implemented as a companion to AERONET and its instrumentation are invariably collocated with AERONET sites.  The Brazilian core of the present network was developed within the scientific framework of the LBA-ECO component of the Large-Scale Biosphere-Atmosphere Experiment in Amazônia.  Historically, SolRad-Net has preferentially selected sites that routinely experience intervals of biomass-burning, such as Amazônia and Sub-Saharan Africa for its long-term monitoring. 

  All flux data are transmitted from the field on an hourly or half-hourly basis, and are accessible to the public shortly thereafter via our website as Level 1.0 data.  Data quality for all sites is evaluated within a week and qualifying data are designated as Level 1.5 and also made available for download. [More on data quality]

    Integration of these flux data with atmospheric aerosol data from Cimel sunphotometers provides a valuable tool for examining the effects of aerosols on flux attenuation.  Similarly, the year-round, multi-year flux data sets (including records of total solar (305 – 2800 nm), Photosynthetically Active Radiation (PAR; 400-700 nm) and UV plus visible: 305 - 695 nm irradiance) provide an opportunity to investigate the long-term influence of biomass burning aerosols on solar fluxes and the resulting effects on vegetation productivity and regional meteorological processes.

      More recently, there are also continuous flux records for Kanpur (India) and Nairobi (Kenya) and the Goddard Space Flight Center (Greenbelt, Maryland) as well as contributed data from collaborators in Sede Boker (Israel), Kishinev (Moldova) and Crete.

    As with AERONET, this network's instrumentation is functional with minimal power and infrastructure requirements and thus has been well-suited to venture into remote and underdeveloped regions and provide valuable, unprecedented data sets in climatically critical ecosystems.

    SolRad-Net, like its associated network, AERONET, maintains a fundamentally open data policy and encourages collaborative and multi-disciplinary integrative analyses.

   Each SolRad-Net site was intially equipped with two flux sensors--a Kipp and Zonen CM-21 pyranometer (305-2800 nm) for measuring the total solar spectrum and Skye Instruments SKE-510 PAR (photosynthetically-active radiation) Energy sensor (spectral range: 400-700 nm).  In 2001, the PAR sensors were replaced with a filtered CM-21 pyranometer.   The RG695 filter employed transmits light with a wavelength greater than 695nm and so the value determined by subtracting this measurement from the unfiltered pyranometer represents the UV + PAR irradiance (305-695nm).  The flux sensors record the instantaneous irradiance at 2-minute intervals.

    The Skye SKE-510 <http://www.skyeinstruments.com> uses a blue enhanced planar diffused silicon detector plus an interference filter and has a fairly even response from 400 to 700 nm. A special cosine correction head on the detector improves the sensor's response at low solar elevation angles. [info sheet]

    The Kipp & Zonen <http://www.kippzonen.com> CM-21 units are ISO 9060 Secondary Standard thermopile pyranometers featuring a receiving element housed beneath two concentric Schott K5 glass domes and a spectral range of 305 to 2800nm. [info sheet]

   The frequencies, channels and transmission windows are assigned by NOAA NESDIS for GOES, EUMETSAT for METEOSAT and GMS which are broadcast in the 401 to 402 MHz range. The satellite transmitter module used is a Vitel VX1004 which is commercially modified for use with the CE 318 A (The Vitel VX1004/2 is used by the PHOTON group but it is no longer in production). The antenna is conical approximately 40 cm in diameter and 40 cm long. The transmitter system is battery operated and charged by a 10 watt solar panel.

    When possible, calibration of the PAR sensors was accomplished by using in-situ comparisons to a radiative transfer model calculations of flux on selected optimal days.  The 6S [Vermote et al, 1997] model we used for this purpose is based on the successive order of scattering method.  Field days used were of minimal aerosol-loading (AOT500nm < 0.1) under cloud-free conditions. Such aerosol levels are sufficiently low to render an exact knowledge of the absorption properties unimportant for computing clear-sky insolation to an accuracy of >1%.  The manufacturer (Skye Instruments) states an initial accuracy for its factory calibration of 5% (generally < 3%).   However, due to changes in the stability of the silicon detector (estimated by Skye as ~ 2% per year) and more notably, diminished transmittance of the bandpass interference filter upon prolonged exposure, we observed significant degradation of sensitivity in these sensors of approximately 6-8% per year.  This necessitated that we employ in-situ calibrations when possible to ensure the best possible determination of PAR calibration.  The utility of this technique varied greatly depending on local climatology (a cloudless sky with low AOT is required) and thus our estimation of PAR flux accuracy varies with site.   At some locations, such as Alta Floresta, Brazil, we benefited from an environment that presented numerous optimal calibrations dates and were able to characterize the trend in calibration drift with high confidence [In-situ calibration plot].  Locations with few, or no suitable calibration conditions (such as persistently cloudy locations) are necessarily calibrated with the manufacturer's calibration only, and thus are not considered suitable for promotion to the highest quality level (2.0).  More on data quality designations can be found in the Data Quality Assessment section below.

    Due to additional complicating factors, such as water vapor influence, thermal equilibrium issues, and also, because of the greater degree of accuracy (2%) provided by the manufacturer and generally much greater stability of this type of radiometer, the factory calibrations were used for the pyranometers.   These pyranometers have all been observed to retain consistent calibration coefficients (calibration drift less than 1% per year) between re-calibration sessions.

  All flux data are initially screened to remove any intervals where operational problems are known or suspected to have compromised the data.  The most common examples of such problems include amplifier malfunctions, a contaminated or dirty pyranometer dome, a non-level sensor, etc.  All of the sensors utilized output a milliVolt signal that is amplified prior to being recorded by the VITEL data logger.   By a large margin, the most common cause of data loss has been temporary or permanent failure of the amplifier due to the rigors of year-round monitoring in hot and often humid climates.  Data that have been cleared as free of any operational problems are designated as Level 1.5.  The raw, unscreened data are Level 1.0 by default and may contain observations that have been compromised for any of the (occasionally subtle) reasons described above.  The Level 1.0 data are made available in the interest of presenting a comprehensive dataset and to provide full transparency of our methods, however any use of these data is strongly discouraged.

    While the criteria for the Level 1.5 qualification are the same for all data, the highest quality data designation differs in meaning for the two sensor types.  For the pyranometers (filtered and unfiltered), due to the their superior stability and greater calibration accuracy, any Level 1.5 data which is from an interval that includes a pre- and post-calibration determination from Kipp & Zonen will be qualified as Level 2.0.   Data lacking a post-deployment calibration will necessarily remain Level 1.5, however we have found even these data to be of useful accuracy as subsequent calibrations have typically shown minimal sensor drift (< 1%).

    Given the observed exposure-dependent calibration drifts in the Skye PAR sensors, we have found in-situ calibrations [see Calibration] to yield the best results for our data set.  The quality of this method will obviously increase with the availability of suitable field calibration days, and consequently the accuracy stated for a given PAR data set will vary with the quantity and quality of suitable calibration dates.   The Level 2.0 designation for PAR sensors is appropriate in situations where a pre- and post-deployment in-situ calibration was possible for a given interval of field data.  Such data represent the optimal data quality attainable for our existing PAR data, although the absolute accuracy will generally be notably less than that associated with Level 2.0 pyranometer data.  At times, PAR data which can not be raised beyond Level 1.5 could be manifesting moderate to large calibration drift which we are unable to characterize adequately, and any use of these data should be made with this understanding.

Note:

The remote nature of most of our sites preclude the use of ventilators for the pyranometers since these require an AC power source, and our instrumentation is wholly reliant on solar power from photovoltaic cells.  Such ventilators are often employed in the interest of minimizing the phenomenon known as zero offset associated with undesirable thermal gradients in the sensor and also to prevent condensation on the dome.  Zero offset effects occur in pyranometers when the high emissivity of the outer dome leads to stong radiative cooling under clear sky conditions.  The cooling of the outer dome creates a temperature gradient opposite in sign to that experienced during the normal solar heating of the thermopile, which can be observed as an apparent negative flux during night-time conditions.  While the influence of this effect on our pyranometer data can not be properly quantified, it is nevertheless the case that for total flux measurement the pyranometer calibration process will inherently account for any offset as it is incorporated into the derived calibration coefficient.  We therefore don't feel that the absence of ventilation for our pyranometers adversely affects the quality of the data set.  The potential for the introduction of significant errors due to the zero offset is much greater for observations utilizing shaded pyranometers to measure diffuse flux.