A Meteo_LCD report
A preliminary analysis of the Safesun personal UV sensor
and a comparison with the Saitek and UV mod. 3D sensors
Francis Massen, Nico Harpes
ver. 1.0, May 1998
|Abstract: A first calibration study is made of the personal UVI sensor sold by Safesun, Australia. Comparing the readings of the Safesun sensor pointing to the sun with those of a stationary mounted UVB_Biometer with zenithal orientation shows that multiplying the Safesun UVI readings by 0.74 gives a reasonal good agreement.|
1. Methodology and problems of the study
2. The Safesun sensor
3. The time series of the UVI readings of the 3 portable sensors and the reference instrument
4. Calibration factors to apply to the portable sensors
5. Dose measured by the Safesun versus dose measured by the UVB Biometer ("correct" dose)
|1.Methodology and problems of the study|
The UV-Index is a unit for measuring solar effective UV irradiance. One UV index UVI corresponds to an effective UV irradiance of 25 mW/m2. IRPA (ICNIRP-1/95) recommends to take the mean of measurements made at 10 to 30 minutes interval around noon.
UVB measurements are normally done with sensors oriented to the zenith, i.e. the measured UVB radiation falls on a horizontal sensing element. The geometry and particularities of portable instruments often do not permit this method of operation: the only manner to obtain reproducible result with many handheld devices is to point them to the sun, which usually is identical or close to the orientation which gives a maximal reading.
Three portable sensors are used in this study: the Safesun, the Saitek Sunwatch II and the Solar Light UV Meter Model 3D (called hereafter UV_3D). Only the UV_3D can be used in a zenithal orientation, but in this study, the measurments with all 3 sensors were done by pointing them to the sun. The last 2 sensors have been studied in the report "A comparison of three portable UV sensors".
The primary reference is a precise UVB Biometer model 501A by Solar Light, operated at the Meteo_LCD station. This stationary instrument has a zenithal orientation. Its readings are assumed to correspond to the correct effective (IEC action spectrum) UVB irradiance. The relation-ship between the readings of the hand-held sensors and the UVB Biometer will be fitted to a linear function: in fact the study computes for each portable sensor a calibration multiplier which, when applied to their readings, gives a result that corresponds to the assumed correct measured value of the UVB Biometer.
All measurements with the portable sensors were done at Bettendorf (L-9357), at one of the authors (F. Massen) home. The distance to the meteorological station is about 3 km, the altitude practically the same. The measurements were made from a balcony facing South situated at 2nd floor level. This means, that diffuse radiation is blocked out from an angle of at least 2pi steradian. The problem of the diffuse radiation can not be solved adequately for these sensors: because of their geometry and operational practice, the Safesun and Saitek do measure an unknown quantity of diffuse radiation (unknown means that the spherical angle under which infalling radiation may reach the sensor is not known). As we found no practical answer to this problem, the study simply compares the readings of the portable sensors used at the balcony and pointing to the sun with one another, and with the readings given by the stationary UVB Biometer which measures both direct and diffuse radiation coming from an angle of 2pi steradian.
Most of the time three measurements were done at 30 minutes interval. The measuring period was from the 12th February 1998 to the 31th March 1998; measurement time was always situated between 11:00 and 14:00 UTC; in the vast majority of cases, the measurements started at 12:00 UTC.
The measurement period corresponds to low and moderate UVB levels. A follow-up study will extend the analysis to higher irradiances. The Safesun sensor measures UVI and UVB dose (in MED, 1 MED = 210 J/m2). A time-span of about 1.5 to 2 hours was usually taken for the dose measurements.
All computations and graphs are made using Statistica 5.0 from Statsoft; the linear fits are done using a Quasi-Newton/Simplex algorithm.
|2. The Safesun sensor|
All operational and technical details of this sensor can be found at Safesun's web-site (Australia). It should be noted that a sticker inside the instruments says that the manufacturer is Optix Tech Ltd., Europe.
Let us just insist here on several important points:
|3. The time series of the UVI readings of the 3 portable sensors and the reference instrument|
|Fig.1 shows all UVI readings (28 measurements) of the Safesun, Saitek, UV_3D and the stationary instrument. It can be seen, that the readings of the Safesun and Saitek are always well in excess to those of the UVB_Biometer, those of the UV_3D being much more close to the latter. The greatest difference is an excess UVI of about 1.5.|
At a first glance, one could attribute this excess readings to the fact that the portable sensors are pointed to the sun, and the reference Biometer is not (cosine factor!). Actually, this is not the case. In 1997, a comparison was made between the UVB Biometer and the UV_3D, both mounted vertically at the same location. A calibration factor of 0.843 was found for the UV_3D.
|When we apply this multiplier to the readings of the UV_3D given in Fig.1, we obtain "correct" values for the instrument pointing to the sun. Plotting the UVB Biometer data versus these corrected UV_3D data gives a multiplier of 1.07 (see Fig.2): this means, that if the UV_3D sensor had pointed to the zenith, with open surroundings, its readings would have been about 7% higher (not lower!). This increase is certainly due to the missing diffusive radiation, which could not reach the sensor pointing to the sun at the measurement place.|
Conclusion: the UVI values displayed by the Safesun and the Saitek are too high compared to the UVI values computed from the UV Biometer readings. As the Safesun follows quite closely the Diffey Erythema Sensitivity Spectrum according to the manufacturers scientific data sheet, (as does the Solar Light UV mod.3D), the probable cause of this behaviour might be the reduced field of view and/or less stringent inherent quality requirements of the Safesun, compared to the more professional sensor.
|4. Calibration factors to apply to the portable sensors|
and 5 give the plots of the assumed correct UVB
Biometer readings versus those of the three portable instruments (Safesun, Saitek Sunwatch
II and UV Mod.3D).
The goodness of the fits are never less than R=0.85.
The conclusions can be stated as follows:
|instrument||to obtain "correct" UVI multiply reading (or computed UVI) by ...|
|Saitek Sunwatch II||0.80|
|Solar Light UV_3D||0.90|
|5. Dose measured by the Safesun versus dose measured by the UVB Biometer ("correct" dose)|
The Safesun can display the effective dose (in MED, 1 MED = 210 J/m2) with a resolution of 0.1 MED. Assuming a correct working clock and computing unit, one should expect the same multiplier valid for obtaining the correct dose out of the reading displayed by the Safesun.
|Let us nevertheless plot the dose computed from the UVB Biometer readings against the Safesun data (Fig.6). The multiplier is 0.71, very close to the above found 0.74. The difference probably comes from the fact that in computing the dose from the UVB Biometers some interpolations had to be done, as this instruments gives a reading only every half hour, whereas the Safesun makes an update of its reading at every 0.1 MED increase.|
The Safesun seems, out of the 3 portable sensors, the one which deviates most from the "correct" values of the reference UVB Biometer. Nevertheless, it presents certain real assets compared to the Saitek Sunwatch II:
1. it is easier to operate (greater insensitivity in respect to its orientation)
2. the display is much more readable
3. it can measure the received dose, what the Saitek Sunwatch II can not do
Both sensors err on the safe side, which means they display readings that are too high. This is certainly preferable to giving readings that are too low.
|In our opinion, the Safesun could be bettered
in several points:
1. lower battery consumption or enable use of standard inexpensive batteries
2. disable accidental powering up
3. display UVI with a resolution of 0.1
4. add display of effective irradiance in MED/h
5. fix the delicate O-ring sealing the battery compartment to avoid loosing it
The very approximate prices of the three sensors are:
UV Mod. 3D (UVB and UVA) 1000 US$
Safesun 180 US$
Saitek Sunwatch II 40 US$
Please consult the manufacturer for up to date prices!
The high price of the UV Mod. 3D shows that this instrument is more a tool for the scientist and researcher, as a personal sensor for the casual user.
Attention: last minute information: (Juni 98)
Safesun has told us that the internal calibration factors used in the first release models of the Safesun have been updated: a correcting factor of approx. 0.86 should be applied to all raw Safesun readings. This gives a much better match with our reference sensor.
The opinions expressed in this report are those of the authors. This study has been made applying good scientific practice. The authors assume no legal liability or responsibility for the accuracy or correctness of any information given. The authors have no commercial or financial envolvement with any of the manufacturers cited in the report.
Back to Home-Page of Meteo_LCD