SUMMARY:
WG4 QA/QC procedures used in national networks
1. RESPONSES FROM:
Austria (Austrian UV Network – Univ. Innsbruck)
Switzerland (Meteoswiss)
Spain (INM)
Spain (INM – MAR project)
Spain (Univ. Extramadura)
France (Univ. Lille, one instrument)
Germany (DWD, 3 instruments)
Germany (ELDONET)
Slovakia (Hydrometeorological Institute standard GPI SAS)
Poland (Institute of Meteorology and Water Management, IMGW)
Cyprus
Norway (NRPA, multifilter radiometer network)
Hungary (HMS)
2. QUESTIONS AND ANSWERS
Question 1: Routine Quality Control of the network
Answers:
- "Routine comparisons with clear sky model
calculations"
- "Comparison of redundant measurements (e.g.,
direct+diffuse vs. global)"
- "Visual check and cleaning of instrument dome every
day/.../twice a week"
- "Performing 100 W lamp test twice a month"
- "Stability of sensor temperature and reliability of
measured UV-B irradiances is checked 3 times per day; desiccant
status"
- "UV Index comparison with nearby stations considering
synoptic situation"
- Comparison to spectral measurements
- "Time; Dome cleaning; Desiccant change"
- " Check daily variation"
- "Time evolution is visually checked; Detection of
anomalous data based on thresholds (periodically)"
- "Data are visually checked every day using relative total
global radiation *ratio to highest possible value) and total
ozone values."
Question 2: Calibration Procedures
Answers:
- "Determination of spectral response (annual); angular
response (bi-annual); calibration by comparison to
spectroradiometer (annual): All network instruments"
- "Comparison to reference broadband radiometer (every 6
months/annual): All instruments are brought together at home
site. "
- "Use of a traveling reference traceable to
spectroradiometer."
- "Comparison to spectroradiometer (bi-annual), but no
determination of spectral response/ angular response."
- "A quality assurance system ISO 9001:2000 is emplemented
so each calibration procedure is controlled by wok instructions."
- "Half clear day parallel measurements between network and
travelling reference radiometer result in a scale factor for the
network radiometer."
- "Two calibrations per year for each network radiometer,
mid-April, and July."
Question 3: Cosine Error correction
Answers:
- "As a consequence of the absolute calibration outdoors
relative to a spectroradiometer [without significant cosine
error], the average cosine error of the radiometer is therefore
taken into account (average over all weather conditions during
the calibration period)"
- "No cosine correction: calibration and comparison are
performed using central day measurements [solar noon];
Operational data is not cosine corrected"
- "Function of SZA and TOZ is obtained between erythemal
dose from spectroradiometer and RAW radiometer signal"
- "No information on cosine error, and no correction
applied"
- "YES supply conversion factors for different SZA"
Question 4: Erythemal dose rate calculation:
Answers:
- Calibration matrix as function of SZA and TOZ: 3 (number of
responses)
- Constant Calibration factor : 6
- Calibration factor as function of SZA: 1
Suggestions from draft WMO report (courtesy G. Seckmeyer)
"Instruments to measure solar UV radiation: Part 2: Broadband instruments..."
* "Examinations of broadband instruments [...] have revealed differences from the specifications offered by manufacturers [...]. Therefore, radiometric levels of performance of these instruments may not be assumed but must be verified by careful, periodic characterisations."
* "Characterisation of spectral response and cosine response of the radiometer should be completed at regular intervals"
* "Measurements of one broadband detector can be directly compared with results of other instruments only if all instruments have exactly the same spectral sensitivity. This is usually not the case, even for instruments of the same manufacturer."
*
Suggested calibration method:
- Measure the spectral response and the cosine response of the
broadband radiometer.- Measure the solar irradiance with a
colocated spectroradiometer, under clear skies.
- (Apply cosine corrections to both data sets.)
- Weight the measured solar spectrum with the spectral
sensitivity of the radiometer.
- Determine a scale factor to convert the radiometer signal [V],
to detector weighted irradiance [W/m^2]
- Use a radiative transfer model to convert from
detector-weighted irradiance to erythemal-weighted irradiance.
Result: Calibration factor matrix as a function of SZA and TOZ.
("In some cases, only values at small SZA (<40) are important. Then, a single calibration factor [...] may be adequate".)
* Comparison to reference (traveling) instrument:
-"[...] a level of differences between the field instruments smaller than 5% can be achieved. If significant changes in the diurnal patterns of the ratio of the traveling detector to the site detector are observed, then it can be assumed that the spectral sensitivity of the site detector has changed and a recalibration in the laboratory will be necessary". Note: Comparison is done on erythemal dose rates, not raw signals.
Examples for SZA, O3 variabilities: 200 – 500 DU, 75 bis 0 SZA
YES: 25% - 30%
Solarlight : 50%
K&Z : 50%