Intercomparison of UV-B Spectroradiometers at Mauna Loa
During 1995-1996
Patrick J. Neale, Vernon R. Goodrich, and Douglass R. Hayes, Jr.
Smithsonian Environmental Research Center, Edgewater, Maryland 21037
Introduction
Our laboratory has been monitoring surface ultraviolet B spectral band (UV-B) irradiance at Mauna Loa Observatory, Hawaii (MLO), since the fall of 1984. The instrument (SERC-SR8) is similar to a radiometer in operation in Edgewater, Maryland, [Correll et al., 1992]. The instrument measures UV-B irradiance in a series of eight, 5-nm bandpass channels with center wavelengths between 290 and 325 nm, and records 1-minute averages. Operation was continuous during August 1984 to May 1996, except for an annual break of about 1 month when the instrument was returned to Maryland for calibration. Determinations of absolute responsivity were made by recording instrument output under a 1000 W quartz-halogen standard lamp traceable to the National Institute of Standards and Technology. These responsivity determinations revealed a steady linear decrease in overall instrument responsivity of about 6% per year (since 1987) in all channels (R2 = 0.90). However, even after adjustment of voltages for this drift, a trend of decreasing output of the instrument is apparent in long time series of clear sky irradiance at a fixed solar zenith angle (SZA) (data not shown). Thus it appears that on-site response decreased by a greater proportion than observed in the lamp calibrations conducted in Maryland. We are currently conducting experiments to determine the cause of these shifts in responsivity; the working assumption is that the malfunction is connected with high altitude operation since similar shifts have not been observed in the comparable long-term UV-B data set for Maryland [Correll et al., 1992].
While the above described situation precludes any estimates of long-term changes in absolute spectral irradiance of UV-B at MLO at the present time, we have used relative signal intensity in selected channels to provide information on changes in the solar UV-B spectrum. These analyses have included estimates of atmospheric optical depth in the UV-B at MLO [Neale et al., 1994] and evidence of increased incident solar UV-B during an episode of record low total column ozone at MLO during the winter of 1994-1995 [Hofmann et al., 1996; Neale et al., 1996] . In this latter report we presented UV irradiance measured in the 295, 300, and 305 nm channels, relative to the 325 nm channel, for clear days when the secant of the SZA was equal to 1.5 (i.e., an SZA of about 48°). The enhancement of short-wavelength UV-B observed by the SERC-SR8 was consistent with other evidence supporting higher global UV-B with decreased ozone. However, there was no direct verification that the absolute intensity of UV-B at these wavelengths actually increased.
During 1995 and 1996 another type of spectral radiometer was installed at MLO. This instrument (NIWA-SR) is based on a double monochromator and photomultiplier tube (PMT) as designed at the National Institute of Water and Atmospheric Research (NIWA) of New Zealand. The NIWA-SR made observations at a fixed series of SZAs, as well as frequent scans near midday. A description of the instrument and a report of a time series of MLO measurements for 45° SZA on clear mornings was presented by Bodhaine et al. [1997] . In this report we compare some of the results in this latter report to simultaneous measurements made with the Smithsonian instrument. The results show a consistent linear relationship between the outputs of the two instruments.
Results and Discussion
The instrument was calibrated at SERC and sent to MLO in June 1995 and remained in operation until May 1996. On May 4, 1995, transmission spectra were measured for all eight interference filters in the instrument. The center wavelengths (wavelength midpoint between the upper and lower wavelengths at which transmission is 50% of maximum) calculated from these spectra for the 295, 300, and 305 nm channels were 295.12, 300.7, and 304.9 nm.
Spectral scans as described in [Bodhaine et al., 1997] were kindly provided by CMDL. We used the 45° SZA observations for clear sky mornings for the period July 1995 to May 1996. For comparison, we used the 1-minute average from the SERC-SR8 for the time with the closest correspondence to the 45° SZA based on a computer ephemeris program. The voltages recorded at MLO were converted to provisional irradiances by application of responsivities determined under the standard lamp in May 1995. The SERC-SR8 irradiances were then compared with the readings of the NIWA-SR instrument at the wavelength closest to each channels nominal center wavelength.
Figure 1 shows correlation plots of 295, 300 and 305 nm irradiance, which are the wavelengths presented in the Hofmann et al. [1996] and Neale et al. [1996] reports. There is a close linear relationship between the readings of the two instruments with coefficient of determination (R2) varying between 0.87 and 0.92. The slope of the fitted relationship [E(SERC-SR8) = k E(NIWA-SR)] varies between channels (Figure 2). In the 295 channel the ratio is greater than unity, at 300 nm it is about 0.9, and for the remaining channels the ratio ranges between 0.5 to 0.6. These ratios were constant during July 1995 to March 1996 and then shifted to somewhat lower ratios (data not shown). Again, the cause for these shifts remains to be determined.
Fig. 1. Comparison of spectral irradiance (mW cm-2 nm-1) at 45° SZA on clear sky mornings at MLO measured with the SERC-SR8 and NIWA-SR instruments. (a) 295 nm, (b) 300 nm, (c) 305 nm. Each point corresponds to one mornings observation. The line indicates a fitted linear regression with the intercept forced through zero.
Fig. 2. Slopes of the fitted linear regression of the SERC-SR8 irradiance on the NIWA-SR irradiance (irradiance ratio, dimensionless) for each of the eight filters in the SERC-SR8 instrument.
Overall, it appears that application of responsivities as measured under a standard lamp in Maryland to the SERC-SR8 voltages recorded at MLO, lead to estimates of irradiance about 55% of the irradiance recorded by the NIWA-SR. This is consistent with our conclusion that the on-site responsivity of the SERC-SR8 is less than that determined under a calibration lamp. However, at 295 nm, the SERC-SR8 irradiance actually appears to be higher than NIWA-SR irradiance. Also, the relationship between the two irradiances does not follow a straight proportion as well as seen for the other wavelengths (Figure 1a). This is probably due to the wide bandpass of the interference filters (nominally 5 nm, FWHM). If we performed the comparison at the effective center wavelength, i.e., the filter center wavelength adjusted for the greater proportion of longer-wavelength solar irradiance within the pass band [Correll et al., 1992], the proportionality between the SERC-SR8 and the NIWA-SR presumably would be lower and more linear as was the case for the longer wavelength irradiances.
Indeed, empirical determinations of effective center wavelengths could be made through a more extensive comparison between the SERC-SR8 and the NIWA-SR instruments. This is beyond the scope of the present report that had the objective of verifying the relative responsivity of the SERC-SR8. The close relationship between the two records during this intercomparison period motivates further analyses and continued engineering studies on the source of the calibration instability. In the near future it is hoped that the SERC-SR8 can be returned to MLO together with a new version of the Smithsonian UV-B radiometer, and the SERC-SR18 which has 18 interference filters with narrower (2 nm) bandwidth [Thompson et al., 1997] .
References
Bodhaine, B.A., E.G. Dutton, D.J. Hofmann, R.L. McKenzie, and P.V. Johnston, UV measurements at Mauna Loa: July 1995 to July 1996, J. Geophys. Res., 102 (D15), 19,265-19,273, 1997.
Correll, D. L., C. O. Clark, B. Goldberg, V. R. Goodrich, D. R. Hayes Jr., W. H. Klein and W. D. Schecher, Spectral Ultraviolet-B radiation fluxes at the earth's surface: long-term variations at 39°N, 77°W, J. Geophys. Res., 97, 7579-7591, 1992.
Hofmann, D.J., S.J. Oltmans, B.A. Bodhaine, G.L. Koenig, J.M. Harris, J.A. Lathrop, R.C. Schnell, J. Barnes, J. Chin, D. Kuniyuki, S. Ryan, R. Uchida, A. Yoshinaga, P.J. Neale, D.R. Hayes, R. Goodrich, W.D. Komhyr, R.D. Evans, B.J. Johnson, D.M. Quincy, and M. Clark, Record low ozone over Mauna Loa observatory, Hawaii during the winter of 1994-1995, Geophys. Res. Lett., 23 (12), 1533-1536, 1996.
Neale, P.J., D.L. Correll, V.R. Goodrich, and D.R. Hayes Jr., UV-B optical depths at Mauna Loa: Relative Contribution of Ozone and Aerosols, in Climate Monitoring and Diagnostics Laboratory No. 22 Summary Report, edited by J.T. Peterson and R.M. Rosson, pp. 132-134, NOAA Environmental Research Laboratories, Boulder, CO, 1994.
Neale, P.J., D.L. Correll, V.R. Goodrich, and D.R. Hayes Jr., Early morning UV-B during the 1994-1995 record low ozone at Mauna Loa, in Climate Monitoring and Diagnostics Laboratory No. 23 Summary Report 1994-1995, edited by D.J. Hofmann, J.T. Peterson, and R.M. Rosson, NOAA Environmental Research Laboratories, Boulder, CO, 1996.
Thompson, A., E.A. Early, J. Deluisi, P. Disterhoft, D. Wardle, J. Kerr, J. Rives, Y. Sun, T. Lucas, T. Mestechkina, and P.J. Neale, The 1994 North American interagency intercomparison of Ultraviolet monitoring spectroradiometers, Natl. Inst. Stds. Technol. J. Res., 102(3), 279-322, 1997.