A Meteo_LCD report

A comparison of three portable UV sensors

Francis Massen, Nico Harpes, Patrick Breuskin

January 1998

file: uvcomp1.html  version 1.01a

Abstract:

This report compares 3 personal, portable UV(B) sensors:

• the Ultraviolet Meter Model 3D from Solar Light Co.
• the Sunwatch II UVI sensor from Saitek
• the UV Checker UC-120 from Casio

The report shows that the Solar Light and Saitek sensors are adequate devices, whereas the Casio sensor seems unsuitable for reliable personal UVB measurements.

1.1. UV Meter Model 3D (UVB_3D) from Solar Light Co.

This battery driven (9V block) handheld device (see photo) has two sensors: a UVB and a UVA sensor. For this report, only the UVB sensor was used. The readings of the LCD display give the effective UVB irradiance in MED/h (minimal erythemal dose per hour), with 1 MED being defined as an effective UVB energy of 210 J/m2 . The readings of the instrument are relatively independant of the orientation of the sensor(s). In this report, the UVB sensor  was always pointing straight upwards into the sky by holding the instrument in a vertical position.

1.2. The Sunwatch II from Saitek (see also the report "a first calibration..." )

This small battery driven sensor (see photo) can be worn like a wristwatch or clipped to a skirt or other piece of cloth. It measures the effective UVB dose and sounds an alarm, when a safety limit (which depends on the skin phototype (skinnumber) and/or sun protection factor (SPF) programmed into the device) is exceeded. A second small reading in the LCD display shows the UVI index. The documentation joined does not specifiy the definition of this UVI, except that an UVI > 7 "shows that the sun is very strong".

The UVB sensor is located under a small curved plexiglass (or similar) window. Readings vary remarkably with the orientation of that window in respect to the sun. In this report, an orientation which gave a maximum reading was always used. This corresponds more or less to the sensor element being placed perpendicular to the infalling sun rays.

1.3. The Casio UV Checker UC-120

This instrument (see photo) displays three readings: a series of symbols (stars), where the number of symbols should be proportional to the UV irradiance, a second unspecified number described as the "UV intensity" in the documentation, and a series of pictograms (a head  with or without a hat, and a sun umbrella) corresponding to the recommended sun protection.

The meaning of  "UV intensity" is not clearly specified.  The technical specifications state that the range of the measurable "UV intensity" is 0..124 W/m² !!! This would suggest that the instrument measures a combined and unweighted UVB+UVA irradiance...The sensor is located under a small circular window; the influence of the orientation to the sun is very strong. In this report, the instrument was always held to give a maximum reading, which was the case when the sensor was perpendicular to the infalling sunrays.

Under a flip-up cover, the instrument contains a small numeric keypad, which makes it usable as a calculator.

The UVB Biometer of the meteorological station of the Lycée Classique de Diekirch was used as a primary reference. This roof-mounted precision broadband instrument of the Robertson-Berger type has been recalibrated by the manufacturer in January 1997.

During 4 different days, the UVB_3D was mounted near the UV Biometer, and its output stored in a Onset Stowaway datalogger. Comparing the readings of both instruments gave a calibration equation, which allowed to use the UVB_3D as a secondary, portable reference.

Field measurements were done with the 3 portable instruments at three different locations: in a polder in the Netherlands (end May 97), at Diekirch and Bettendorf in Luxembourg (June-July 97), and at a location in the Provence, France (July-Aug. 97).

Here the precise coordinates, altitude and dates of these locations:

The solar irradiance in the Netherlands and at Diekirch/Bettendorf is quite similar, but it is much higher in the Provence. This means that our investigations cover two very dissimilar situations, which will be handled separately.

The evaluation of the Saitek and Casio instruments is always done against the secondary reference given by the adjusted UVB_3D readings.

The results are given in the form of an equation:

correct_value = multiplier * instrument_reading (+ offset)

This means that when the readings of the instrument are multiplied by the found multiplier, the result will correspond to the correct value. Correct means the value which would have been displayed by the UVB_Biometer if this instrument had been used.

As the definition of the MED and/or the UVI varies from instrument to instrument, or is not given by the manufacturer, we use the following definition (which is the standard now in most of Europe, and also that used by our meteorological station):

1 MED = 250 J/m²

1 UVI = 1/40 W/m² = 25 mW/m² effective UVB irradiance

1 MED/h = 69.4 mW/m² effective UVB irradiance

to get the UVI from the MED/h reading multiply by 25/9

to get the MED/h from the UVI reading multiply by 9/25

The readings of the UVB_3D are adjusted to that definition, by multiplying them by 210/250. The analysis is done using these adjusted readings as the definitive output of the instrument. So keep in mind that all values reporting to the UVB_3D are adjusted values!

Fig. 3.1 shows the values of the UVB_Biometer of Meteo_LCD and those of the UVB_3D sensor (adjusted values to comply with 1 MED = 250 J/m2). It can readily be seen that the UVB_3D values are always too high, the excess being greater in high UVB conditions.

Fig. 3.2 gives the UVB_Biometer readings against the (adjusted) UVB_3D output: the conclusion is that the correct UVB irradiance in MED/h can be computed from the adjusted UVB_3D reading by applying the formula: correct_value = 0.83*(adjusted UVB_3D reading) + 0.08 This is the relationship we use to qualify the UVB_3D as a secondary reference. A fit forced through the origin gives: correct_value = 0.88 * (adjusted UVB_3D reading)

Fig. 3.3 shows the difference between the UVB_3D readings on the UVB_Biometer data and the actual air temperature. One should not conclude too hastily that this difference increases with air temperature, which could be due to a temperature sensitivity of the sensor. (Let us recall that the UVB_Biometer is thermally stabilized). Actually the last 2 days show that the magnitude of the difference does not follow the air temperature pattern.

The linear correlation between the difference of the readings and air temperature is  poor (r=0.47), and Fig.3.4 shows that there is no good linear relationship between the Delta_UV and air temperature.

Conclusion:

The relationship between the primary reference (UVB_Biometer) and the portable UVB_3D is a linear one; the deviations from this trend are small. There seems no obvious temperature dependance of the readings of the UVB_3D. Correct UVB results can be found by multiplying the adjusted UVB_3D readings by 0.88, or the raw readings by 0.74. The good linear relationship qualifies the UVB_3D as an acceptable portable secondary reference.

As stated above,  the mean UVB irradiances (given in MED/h) are quite different during the measurement periods:

The comparable UVB magnitude of the Netherland and Diekirch/Bettendorf series allows to combine these two periods into one single data series .

	Fig. 4.1 shows the correct UVI value against the Saitek UVI reading. The linear relation-ship is: Netherland/Diekirch/Bettendorf period (N=27): correct_UVI = 0.92 *(Saitek_UVI_reading) - 0.07
	Fig. 4.2   Provence period (N=26): correct_UVI = 1.07*(Saitek_UVI_reading) + 1.12

As a single multiplier is much more usable than the given affine relationships, we computed a linear regression forced through the origin; this yields (R = goodness of fit):

Conclusion:

The Saitek readings are best in moderate UVB conditions; the readings are about 25% too low in high UVB conditions, and their distribution may differ considerably from the calibration line.

In moderate UVI conditions (UVI<5) a best multiplier to compute the correct UVI from the displayed value is 0.9, whereas it is 1.25 in high UVI conditions (UVI>5). A best strategy to compute a "safe" UVI for all conditions would be to use 1.25 as a multiplier; "safe" means here that the computed UVI is never too small, but may be sometimes too high.

Let us first look for a relation-ship between the number of displayed star symbols and the unspecified number shown on the Casio LCD display:

Netherland/Diekirch/Bettendorf period (N=27):    # of stars = 0.097 *UV_reading)           (R = 0.97)

Provence period (N=26):                                         # of stars = 0.097*UV_reading)              (R = 0.97)

This means that the number of stars displayed is practically equal to 1/10th of the displayed numerical UV value.

Fig.   5.1 compares the measurements of the 3 instruments for all periods; obviously, the Casio readings deviate more from the other two readings during the high UV Provence period.

Fig. 5.2 shows the gives the correct UVI values against the displayed Casio numbers. The linear relation-ship is: Netherland/Diekirch/Bettendorf period (N=27): correct_UVI = 0.11 *(Casio_reading) - 1                 (R = 0.91)

Fig. 5.3: Provence period (N=26): correct_UVI = 0.13 *(Casio_reading) + 0.75          (R = 0.64)

The poor goodness of the last fit confirms the visual impression given by Fig. 5.1.

Let us, as done before, compute a linear regression forced through the origin for the 2 periods:

Conclusion:

The Casio UC-120 "UV intensity" readings deviate from linear calibration line in moderate UVB conditions in a tolerable manner, keeping into account the very low cost of the sensor. In high UVB conditions, the deviations are extreme.

In moderate UVI conditions (UVI<5) a best multiplier to compute the correct UVI from the displayed unspecified value is 0.09; the poor goodness of fit makes the sensor unsuitable in high UVI conditions (UVI>5).