Seasonal Variations of Global, Reflected, and Diffuse Spectral UV Observations based on Brewer Spectrophotometers at Tsukuba, 2004 to 2012

Transcription

1 [Technical report] Seasonal Variations of Global, Reflected, and Diffuse Spectral UV Observations based on Brewer Spectrophotometers at Tsukuba, 2004 to 2012 Mahito ITO * Osamu IJIMA * Tetsuya SHIMAMURA * Itaru UESATO * Yoshiyuki NOTO * Yuji ESAKI * Noriaki OSHIKI * Abstract Since December 2003, monitoring observations of spectral ultraviolet (UV) radiation reflected from the ground surface (RFuv) have been carried out at the Aerological Observatory in Tsukuba with a modified Brewer spectrophotometer, BR#059. Similar monitoring of diffuse spectral UV (DFuv) from the sky has been carried out with a Brewer spectrophotometer, BR#058, with an automated shadow unit, since August 2005 (Ito: 2004, 2006, 2007). In this paper, the results of these measurements are compared with global spectral UV (GLuv) and solar radiation data, GLsolar, DFsolar, and RFsolar, from 2004 to The results are summarized below. (1) The monthly means of daily totals of GLuv, DFuv, and RFuv showed seasonal patterns different from GLsolar, DFsolar, and RFsolar, respectively. (2) The annual means of the daily total of DFuv have been increasing by small amounts from year to year since 2006, in contrast to the almost constant DFsolar. (3) The annual and monthly means of the daily total of RFuv are affected by UV absorption by vegetation and reflection from snow, but vegetation and snow had little effect on the means of RFsolar. (4) The monthly mean of daily UV diffusibility, RDFuv (DFuv/GLuv), averaged about 0.84 (84%) and was almost constant throughout the year, but the diffusibility of solar radiation, RDFsolar (DFsolar/GLsolar), averaged about 0.61 (61%) and showed seasonal variations. (5) The annual means of monthly means of daily UV reflectivity, RRFuv (RFuv/GLuv), averaged only about 0.02 (2%), compared to the average of about 0.21 (21%) for daily solar reflectivity, RRFsolar (RFsolar/GLsolar). The spectrum of RRFuv exhibited a small linear wavelength dependence, with higher reflectivity at longer wavelengths. (6) The contributions (percentages) of the daily totals of direct UV radiation, DRuv (GLuv DFuv), DFuv, and RFuv relative to the daily total of GLuv, averaged about 20%, 80%, and 2%, respectively. In contrast, their contributions to the daily totals of DRsolar, DFsolar, and RFsolar relative to the daily total of GLsolar were different and averaged about 50%, 50%, and 20%, respectively. The differences between the seasonality of the components of UV radiation, GLuv, DFuv, and RFuv, and the components of solar radiation, GLsolar, DFsolar, and RFsolar, were elucidated. Continuous observations with Brewer spectrophotometers are very important for, inter alia, ground truth calibration of satellite observations, clarification of aerosol distributions, and studies of vegetation growth. 1. Introduction A spectral ultraviolet (UV) radiation monitoring network using Brewer spectrophotometers has been maintained by the Japan Meteorological Agency (JMA) at Tsukuba since 1990; at Sapporo, Kagoshima, and Naha since 1991; and at Syowa in Antarctica since 1991 (Ito et al.: 1991). Many characteristics of global UV at the surface of the earth have been clarified by these observations. However, the potential effects of UV radiation on the human body under cloudy conditions is also a concern, because the UV radiation is reflected from the ground surface and scattered by aerosols and cloud particles in the sky (MOE: 2004). The Aerological Observatory therefore started spectral UV observations at Tsukuba of diffuse UV radiation (DFuv) and reflected UV radiation (RFuv) in addition to global UV radiation (GLuv). Observations of RFuv were begun with a modified Brewer MKII * Ozone and Radiation Division, Aerological Observatory, JMA. spectrophotometer in December 2003 (Ito: 2004, 2005a). These observations clarified several different results: (1) the RFuv evidenced a seasonal pattern different from the GLuv. The RFuv intensity was high in March to April and low in December to January, with a period of relatively high intensity from July to August, and (2) the UV reflection ratio, RRFuv (RFuv/GLuv), evidenced a different seasonal pattern because the RFuv was high from January to April and low from June to September (Ito: 2007). Observations of DFuv were started by using an automated shadow unit in August 2005, after experimental observations using manually controlled shadow units (Ito: 2005b, 2006). These observation revealed several results: (1) the DFuv evidenced a different seasonal pattern of high intensity from May to August and low intensity from December to January, and (2) the ratio of UV diffusibility, RDFuv (DFuv/GLuv), was high from June to July and January to February and low from March to September (Ito: 2007). Observations of these three UV components, GLuv, DFuv, and

2 Journal of the Aerological Observatory No Photo 1. Observation site and instruments used. A: Brewer spectrophotometers for GLuv (global UV radiation) observations. B: BR#058 for DFuv (diffuse UV radiation) observations. C: BR#059 for RFuv (reflected UV radiation) observations. D: Pyranometers and pyrheliometers for DFsolar (diffuse solar radiation) and DRsolar (direct solar radiation) observations. E: Pyranometers for RFsolar (reflected solar radiation) observations at the BSRN network site. Table 1 Contents of observation and data. Radiation Symbol Contents GL uv Global UV radiation J/m 2 RF uv Reflected UV radiation J/m 2 DF uv Diffuse UV radiation J/m 2 DR uv Direct UV radiation on the horizontal surface J/m 2 UV RRF uv Reflectivity of UV radiation [RRFuv=RFuv/GLuv] Radiation RDF uv Diffusibility of UV radaition [RDFuv=DFuv/GLuv] UVB Total UV irradiance at the wavelength Fig. 1 Data records and instruments used in this report. RFuv, were continued. The DFuv data were collected for about 8 years, from 2005 to 2012, and the RFuv data were collected for about 9 years, from 2004 to The basic GLuv data were collected for about 23 years, from 1990 to All of these data were sent to the data centre of the WMO/WOUDC. In this paper, the seasonal patterns and inter-annual changes of DFuv, RFuv, and GLuv have been compared and analyzed with respect to three components of solar radiation: global solar radiation (GLsolar), diffuse solar radiation (DFsolar), and reflected solar radiation (RFsolar). The symbols used in this paper are defined in Table 1. Please refer to Ito (2007) for detailed descriptions of the instruments and the observational methods, and to McElroy et al. (2008) and Kipp & Zonen (1996, 2008a, b) for all of the technical terms used in this paper. 2. Instrumental and observational data 2.1 Instruments The GLuv was observed with a Brewer spectrophotometer BR#052, a BR#173, and a BR#200; DFuv, with a BR#058 with from to nm J/m 2 CIE CIE UV irradiance by Commission Internationale de I'Eclairage J/m 2 GL solar Global solar radiation kj/m 2 RF solar Reflected solar radiation kj/m 2 Solar DF solar Diffuse solar radiation kj/m 2 Radiation DR solar Direct solar radiation on the horizontal surface kj/m 2 RRF solar Reflectivity of solar radiation [RRFsolar=RFsolar/GLsolar] RDF solar Diffusibility of solar radiation [RDFsolar=DFsolar/GLsolar] (DRuv=GLuv-DFuv, GLsolar=DFsolar+DRsolar) automated shadow units; and RFuv, with a modified BR#059. The GLuv and the DFuv were observed on the rooftop of a building, and RFuv was observed above a field of grass at the observation site. The size of the shadow ball and the length of the shadow frame of the shadow units for the DFuv observations made with a BR#058 were based on a standard protocol (WMO: 1996, Ohmura: 2002, BSRN: 2002). The distance of 180 cm between the diffuser in the UV dome and the surface of the ground for the RFuv observations with a BR#059 was based on standard protocol (WMO: 1996). The instruments and field are shown in Photo 1 (A to C). For the observations of solar radiation, GLsolar was defined to be the value of DFsolar + DRsolar on a horizontal surface detected by the pyranometer CM22 (CM21) and the pyrheliometer CH1-52 -

3 Seasonal Variations of Global, Reflected, and Diffuse Spectral UV Observations based on Brewer Spectrophotometers at Tsukuba, 2004 to 2012 (AOT/JMA: 1996). These observation sites were also on the rooftop. Pyranometer CM21 was also used for the observations of RFsolar on the observation field. The instruments and the field are shown in Photo 1 (D and E). Figure 1 provides a summary of the use of these instruments for data collection. 2.2 Observation and data The uv.rtn and the skc.rtn modified by time-control for the product of data every time were used for the observation schedules of GLuv, DFuv, and RFuv. The annual means, the monthly means, the daily totals of the integrated values of UVB and CIE in Table 1, and the spectral irradiances at wavelength intervals of 0.5 nm from 290 to 325 nm were produced by using those data. The spectrophotometer data were corrected to ensure a high degree of accuracy by using the results of regular external lamp tests. The errors associated with the process of narrowing the open sky area with the shadow units for BR#058, and the errors associated with changes in instrumental responsivity due to the reversal of BR#059, were included in the NIST lamp calibration errors (Ito: 2004, 2005b, 2007). The GLsolar, DFsolar, and RFsolar data were sampled every second and averaged every minute. The annual means, the monthly means, and the daily totals of these solar radiation data were generated by using those data as well as the above UV products. In the comparative analysis, the irradiance ratios of RDFuv (DFuv/GLuv), RRFuv (RFuv/GLuv), RDFsolar (DFsolar/GLsolar), and RRFsolar (RFsolar/GLsolar) were calculated with these data. In addition, monthly means of GLuv/GLsolar, DFuv/DFsolar, and RFuv/RFsolar were calculated from the daily means to investigate the ratio of UV to solar radiation. The results are consequently slightly different from the data generated by using the monthly means of the daily totals. In the comparison with the amount of solar radiation, as a matter of convenience we interpreted the amount of UV observed spectrophotometrically (290 to 325 nm) to be part of the solar radiation, although the measurable wavelength range of pyranometers CM22 and CM21 was stated to be 305 to 2800 nm, and that of the pyrheliometer was 200 to 4000 nm. Those monthly and annual means are shown in Fig. 2 (a-1) to (e-5). The dark blue lines in the figure indicate the annual means. The normal 9-year averages of the monthly means (DF: the 8-years average) are shown Fig. 3 (a-1) to (e-5). The dark black lines in the figure indicate the normal 9-year averages of monthly means, and the thin lines indicate the monthly means every year. In the graphs of RF and RRF, the monthly means for 2007 without snow cover are shown as the heavy dotted lines. 3. GLuv, DFuv, RFuv, RDFuv, and RRFuv 3.1 GLuv and GLsolar The seasonal pattern and the inter-annual changes of GLuv were similar to those of GLsolar in Fig. 2 (a-1), (b-1), and (c-1). But a small difference between GLsolar and GLuv is apparent. The maxima of GLsolar were scattered among the months from April to August, whereas the maximum of GLuv was concentrated in the months of July and August. The minima of GLsolar were scattered within the months from October to December, whereas the minimum of GLuv was concentrated in December, the month when the solar zenith angle is a maximum. As shown in Fig. 3 (a-1), (b-1), and (c-1), the normal averages of the 9-year monthly means, the annual time series of GLsolar followed a bimodal pattern with peaks in May and August. In contrast, the annual time series of GLuv followed a unimodal pattern with a single peak in August. We estimate that this difference of maximum month (peak) reflects the fact that (1) the rainy season comes in June, when the solar zenith angle is a minimum at Tsukuba, and (2) the maximum total ozone occurs in April during the spring season, and the total ozone becomes a minimum from September to October during the autumn at Tsukuba. In contrast, the GLsolar irradiance is a minimum in December, when the intensity is about 50% of the maximum value in May and August, but the GLuv irradiance is a minimum in December, when the intensity is only about 20% of the maximum value in August. The inter-annual change of GLsolar shown by the dark blue line in Fig. 2 indicates that there was a local minimum in 2006 and that there has been a pattern of small increases since The GLuv also evidences a similar trend. 3.2 DFuv and DFsolar As shown in Fig. 2 (a-2), (b-2), and (c-2), the seasonal pattern and inter-annual changes of DFuv were almost the same as those of DFsolar. Although DFsolar reached a maximum during the period from June to July, DFuv reached a maximum during the period from July to August. This difference could reflect the fact that DFuv is affected by total ozone. In contrast, the minima of DFsolar were slightly scattered from December to January. However, the minimum of DFuv was concentrated in December, because total ozone increased from January. The annual time series of DFsolar shown in Fig. 3 (a-2) followed a unimodal pattern with a peak in June, the month when the solar zenith angle is lowest, and no effects of the rainy season on the seasonal pattern were apparent. The annual time series of DFuv followed a unimodal pattern with a peak in August, similar to the GLuv time series

4 Journal of the Aerological Observatory No Notes Instruments GLsolar: shaded pyranometer CM21(CM22) and pyrheliometer CH1. DFsolar: shaded pyranometer CM21(CM22.) RFsolar: pyranometer CM21. GLuv: BR#052 (#173, #200). DFuv: shaded BR#058. RFuv: modified BR#059. Middle blue lines in every figure: Annual means. Fig. 2 Monthly and annual means of daily total irradiance, daily diffusibility, and daily reflectivity at Tsukuba from 2004 to (a-1) GLsolar: Global solar radiation (kj/m 2 ). (a-2) DFsolar: Diffuse solar radiation (kj/m 2 ). (a-3) RFsolar: Reflected solar radiation (kj/m 2 ). (a-4) RDFsolar: Diffusibility of solar radiation (DFsolar/GLsolar). (a-5) RRFsolar: Reflectivity of solar radiation (RFsolar/GLsolar). (b-1) GLuv: Global UV (UVB) radiation (J/m 2 ). (b-2) DFuv: Diffuse UV (UVB) radiation (J/m 2 ). (b-3) RFuv: Reflected UV (UVB) radiation (J/m 2 ). (b-4) RDFuv: Diffusibility of UV (UVB) radiation (DFuv/GLuv). (b-5) RRFuv: Reflectivity of UV (UVB) radiation (RFuv/GLuv). (c-1) to (c-5) The same variables as in (b-1) to (b-5), but for UV (CIE) radiation. (d-1) The contribution of global UV (UVB) radiation to global solar radiation (GLuv/GLsolar). (d-2) The contribution of diffuse UV (UVB) radiation to diffuse solar radiation (DFuv/DFsolar). (d-3) The contribution of reflected UV (UVB) radiation to reflected solar radiation (RFuv/RFsolar). (e-1) to (e-3): The same variables as in (d-1) to (d-3), but for UV (CIE) radiation

5 Seasonal Variations of Global, Reflected, and Diffuse Spectral UV Observations based on Brewer Spectrophotometers at Tsukuba, 2004 to 2012 Notes Thin lines: Monthly means in every year. Heavy lines: Average of the monthly means for the observation periods. Heavy dotted lines of green color in (a-3), (a-5), (b-3), (b-5), (c-3) and (c-5), shows the monthly means in the year without snow cover day, Fig. 3 Seasonal change of monthly means of daily total irradiance, daily diffusibility, and daily reflectivity at Tsukuba from 2004 to Caption shared with Fig. 2. Figures (a-1) to (c-5) show seasonal changes every year and the average of the changes, as in Fig

6 Journal of the Aerological Observatory No The annual averages of DFsolar were almost constant from year to year, but the inter-annual change of DFuv evidenced a small increasing pattern. The reason for this small increase is unclear. 3.3 RFuv and RFsolar As indicated in Fig. 2 (a-3), (b-3), and (c-3), the seasonal pattern and the inter-annual changes of RFuv followed almost the same pattern as RFsolar. But a small difference between RFsolar and RFuv was apparent. The maxima of RFsolar were scattered among two periods during the months of April to May and July to August, whereas the maxima of RFuv were widely scattered in the months from January to August and were affected by the degrees of snow cover. Except for days of snow cover, the maxima of RFuv were scattered in the months from April to August, very similar to the maxima of RFsolar. The minima of RFsolar were scattered within the months from October to December, whereas the minimum of RFuv was concentrated in December. The annual time series of RFsolar shown in Fig. 3 (a-3) followed a bimodal pattern with peaks in April and August. This seasonal pattern is similar to the seasonal pattern of GLsolar. In contrast, the annual time series of RFuv in Fig. 3 (b-3) and (c-3) followed a bimodal pattern, but the local maximum irradiance in April was larger than the local maximum irradiance in August. The RFuv irradiance during the period from January to March was not reduced because of snow cover. In 2007, when there was no snow cover, the RFuv irradiance in January was slightly larger compared to the irradiance in December and increased till April (the time when there was a local maximum). The inter-annual changes of RFsolar evidenced a local minimum in 2006, similar to the inter-annual changes of GLsolar, and have increased by a small amount since In contrast, the RFuv evidenced a local minimum from 2007 to 2009 because of snow cover and has increased since RDFuv and RDFsolar The RDFsolar irradiance ratio shown in Fig. 2 (a-4) evidences a maximum intensity of about 0.8 and intensity minima of about 0.4 scattered from December to February. In contrast, the values of RDFuv in Fig. 2 (b-4) and (c-4) evidence an almost constant and high value of about 0.83 during the entire period of observations. The maxima of RDFuv were widely scattered during the months from June to December with a peak value of about 0.88, and the minima were very widely scattered during the months from December to August, the lowest values being 0.63 to The annual time series of RDFsolar shown in Fig. 3 (a-4) followed a bimodal pattern characterized by peaks in June (about 0.78) and October (about 0.64). The minimum occurred in January (about 0.42). The difference between the maximum and minimum ratios indicates an intensity of about 0.36 (36%). The seasonal pattern of RDFuv shown in Fig. 3 (b-4) and (c-4) shows very small variations, without any remarkable maximum or minimum. The RDFuv irradiance ratio remained at a high intensity of throughout year. Very small local maxima occurred in June (about 0.84) and October (about 0.88), and very small minima occurred in April (about 0.81) and August (about 0.82). The difference between the maximum and minimum ratios indicates an intensity of about 0.07 (7%). This difference was discernible every year. The inter-annual changes of RDFsolar evidence a local maximum in 2006 (about 0.64). From 2006, the RDFsolar decreased by a small amount to about In contrast, the RDFuv evidenced a small local minimum from 2007 to 2008 (about 0.82). The RDFuv at other times was almost constant, the intensity being about RRFuv and RRFsolar The RRFsolar irradiance ratio shown in Fig. 2 (a-5) evidenced a maximum intensity of and intensity minima of about 0.18 that were scattered during the month of May. In contrast, the RRFuv shown in Fig. 2 (b-5) and (c-5) evidenced a wide variation of intensities, from to 1.10, the maxima being scattered within the time interval from December to April. The maximum intensities were affected by the number of consecutive days with snow cover. In 2007, when there was no snow cover, the maximum irradiance ratio corresponded to an intensity of about in March. In contrast, the ratio associated with the scattered minima from June to October corresponded to an intensity of about The annual time series of RRFsolar shown in Fig. 3 (a-5) was unimodal with a maximum in January (about 0.24) and a minimum in May (about 0.18). The difference between the maximum and minimum ratios indicates a very small intensity of about 0.06 (6%). This difference was discernible every year. In contrast, the annual time series of RRFsolar shown in Fig. 3 (b-5) and (c-5) was unimodal with a maximum in January (about 0.042) and a minimum in July (about 0.012). This difference between the maximum and minimum ratios indicates a small intensity of about 0.03 (3%). In 2007, the year without a single day of snow cover, the RRFuv irradiance ratio increased to a maximum in May (about 0.019) and decreased to a minimum in July (about 0.012). The difference between the maximum and minimum ratios indicates a very small intensity of about (0.7%). In 2007, the seasonal pattern of RRFuv was affected by the growing conditions of the grass, because the length of grass is shortest in May and longest from July to August

7 Seasonal Variations of Global, Reflected, and Diffuse Spectral UV Observations based on Brewer Spectrophotometers at Tsukuba, 2004 to 2012 The inter-annual changes of the RRFsolar irradiance ratio evidenced a pattern of very small increases from 2004 (about 0.18) to 2012 (about 0.21). In contrast, the RRFuv evidenced a small local maximum in 2006 (about 0.03), a small local minimum in 2007 (about 0.015), and a pattern of small increases since 2007, the intensity in 2012 being about GLuv/GLsolar, DFuv/DFsolar, and RFuv/RFsolar 4.1 GLuv/GLsolar As indicated in Fig. 2 (d-1) and (e-1), the maxima of the GLuv/GLsolar irradiance ratios were concentrated in the period from July to August, and the minima were concentrated in the period from December to January. In Fig. 3 (d-1) and (e-1), the annual time series of GLuv/GLsolar was unimodal, with a maximum in July (UVB: , CIE: ) and a minimum in January (UVB: , CIE: ). The differences between the maximum and minimum ratios indicates that the intensities were about (0.09%) for UVB and about (0.01%) for CIE. These differences were discernible every year. The inter-annual changes of the GLuv/GLsolar irradiance ratio evidenced a small minimum in 2005, but the ratio has been almost constant since DFuv/DFsolar As indicated in Fig. 2 (d-2) and (e-2), the DFuv/DFsolar irradiance ratios have been remarkably different in each year. The maximum has occurred in August, and the minima have been scattered from December to February. As indicated in Fig. 3 (d-2) and (e-2), the annual times series of DFuv/DFsolar has been triangular, with a maximum in August (UVB: , CIE: ) and a minimum in December (UVB: , CIE: ). The difference between the maximum and minimum ratios indicates intensities of about (0.07%) for UVB and about (0.01%) for CIE. A small local maximum occurs in May. However, the seasonal changes in every year evidence patterns different from the above average pattern for observations during the 8-year study period. The inter-annual changes of the DFuv/DFsolar irradiance ratio evidence very small minima in 2006 and 2008, and a small increase in intensities since The reason for the increasing intensities is the almost constant intensity of DFsolar and the small increase of DFuv discussed in section 3.2. occurred from December to January. The annual time series of RFuv/RFsolar was trapezoidal, as shown in Fig. 3 (d-3) and (e-3), with a maximum in April (UVB: , CIE: ) and a minimum in December (UVB: , CIE: ). The difference between the maximum and minimum ratios indicates that the intensities were about (0.006%) for UVB and about (0.0006%) for CIE. Small local maxima occurred in July. Inter-annual changes in the RFuv/RFsolar irradiance ratio evidenced local maxima in 2005, 2008, and Those local maxima were due to the small increase of RFuv caused by snow cover, as discussed in section Spectral irradiances and spectral ratios 5.1 Monthly spectral irradiances and monthly spectral ratios The GLuv, DFuv, and RFuv monthly spectral irradiances and the RDFuv and RRFuv monthly spectral ratios from 2004 to 2012 at Tsukuba are shown in Fig. 4 (1) to (3) and in Fig. 4. (4) and (5), respectively. Heavy lines in the figures show the averages for the observation period in 9 (DF: 8) years. The GLuv and DFuv monthly spectral irradiances in Fig. 4 (1) and Fig. 4 (2), respectively, were characterized by a maximum from June to August and a minimum in December. The DFuv irradiance intensity was about 80% of the GLuv irradiance intensity. In contrast, the RFuv monthly spectral irradiance in Fig. 4 (3) was characterized by a maximum from January to April and a minimum from October to December. The RFuv irradiance intensity was about 2% of the GLuv irradiance intensity. The monthly RDFuv spectral ratios shown in Fig. 4 (4) all fell within the range The wave-like lines in the figure may reflect small wavelength shifts in the BR#058 spectrophotometer used to make the DFuv observation. The monthly RRFuv spectral ratios shown in Fig. 4 (5) fell within the range , but the intensities changed in every month. In the figure, the maximum occurred during the snow cover season and the period when vegetation was depleted from January to March. The minimum occurred during the season when vegetation was mature from July to August. The long-wavelength spectral ratios were slightly higher than the short-wavelength spectral ratios. Extrapolation of the RRFuv spectral ratios produced estimates that agreed with the RRFsolar spectral ratios. This extrapolation is confirmed by the high value of RRFsolar (0.205) shown in following Table RFuv/RFsolar As indicated in Fig. 2 (d-3) and (e-3), the RFuv/RFsolar irradiance ratios were remarkably different in each year. The maxima were scattered from January to August, and the minimum 5.2 Spectral irradiances and spectral ratios at wavelengths of 305, 310, 315, 320, and 325 nm The seasonal changes of monthly spectral irradiances at wavelengths of 305, 310, 315, 320, and 325 nm in Figs. 5 (1) to (5)

8 Journal of the Aerological Observatory No Fig. 4 Seasonal change of monthly GLuv, DFuv, and RFuv spectral irradiance, and their spectral ratios from 2004 to 2012 at Tsukuba. Heavy lines show the averages for the observation periods. (1) GLuv, (2) DFuv, and (3) RFuv: Global, diffuse, and reflected spectral UV per 5 nm from 290 to 325 nm in every year. (4) RDFuv and (5) RRFuv: Diffusibility and reflectivity relative to global spectral UV per 5 nm from 300 to 325 nm in every year. Fig. 5 Seasonal changes of GLuv, DFuv, and RFuv monthly irradiance and monthly RDFuv and RRFuv irradiance ratios, at wavelengths of 305, 310, 315, 320, and 325 nm, for the observation period from 2004 to 2012 at Tsukuba. (1) GLuv, (2) DFuv, and (3) RFuv: Global, diffuse, and reflected spectral irradiance. (4) RDFuv and (5) RRFuv: Diffusibility and reflectivity to global spectral irradiance. Dotted lines in Figs. (3) and (5) show the irradiances and irradiance ratios in 2007, the year without snow cover. Fig. 6 Relationships between UV (CIE) radiation and solar radiation, using averages of the monthly means for 9 (8) years from 2004 (2005) to 2012 at Tsukuba. (1): GLsolar versus GLuv, (2): DFsolar versus DFuv, (3): RFsolar versus RFuv, (4): RDFsolar versus RDFuv, and (5) RRFsolar versus RRFuv

9 Seasonal Variations of Global, Reflected, and Diffuse Spectral UV Observations based on Brewer Spectrophotometers at Tsukuba, 2004 to 2012 Fig. 7 Annual changes of direct, diffuse, and reflected irradiances, and their percentage contributions to global irradiance based on calculated results with averages of monthly means of daily total irradiances, for 9 (8) years from 2004 (2005) to 2012 at Tsukuba. (a-1), (b-1), (c-1), and (d-1): DR, DF, and RF irradiances. (a-2), (b-2), (c-2), and (d-2): DR, DF, and RF irradiance percentages. were calculated from the averaged monthly means of daily totals during the observation period of 9 (DF: 8) years. In Figs. (3) and (5), the seasonal changes in 2007, the year without snow cover, are shown by the dotted lines. The GLuv annual time series at 5 wavelengths shown in Fig. 5 (1) have triangular shapes with maxima in August. In contrast, the DFuvs at the same wavelengths in Fig. 5 (2) have trapezoidal shapes characterized by maxima from June to August. The RFuvs at the long-wavelengths (315, 320 and 325 nm) in Fig. 5 (3) have trapezoidal shapes characterized by maxima from January to April and minima from November to December. However in 2007, the year without snow cover, the RFuv pattern was bimodal with a main peak from May to April and a local maximum in August. The RDFuv in Fig. 5 (4) evidenced higher intensities at wavelengths of 320 and 325 nm. The reason is not clarified, because the characteristics are not agreed within the low value of RDFsolar (0.605) shown in following Table 2. The difference of the RRFuv between snow cover and no snow cover is clear in Fig. 5 (5). The ratios at long wavelengths are higher than the ratios at short wavelengths, especially from January to April. 6. The relationships between UV and solar radiation The relationships between the GLuv, DFuv, and RFuv UV (CIE) radiation and the GLsolar, DFsolar, and RFsolar solar radiation are revealed by examining the monthly means for the observation period in Figs. 6 (1) to (5). In these figures, the symbols for each month are color coded. The black thin line and the magnitude of R 2 indicate the approximate trend and the fraction of the variance accounted for by the regression line, respectively. The relationship between GLsolar and GLuv shown in Fig. 6 (1) is almost elliptical in shape, the associated R 2 being The relationship between DFsolar and DFuv shown in Fig. 6 (2) reveals more scatter at high irradiances, the associated R 2 being The relationship between RFsolar and RFuv shown in Fig. 6 (3) has been highly affected by snow cover, the associated R 2 being There is little recognizable relationship between RDFsolar and RDFuv apparent in Fig. 6 (4), the associated R 2 being The relationship between RRFsolar and RRFuv shown in Fig. 6 (5) has also been much affected by snow cover, the associated R 2 being As mentioned, DFsolar and DFuv are more highly correlated than GLsolar and GLuv. The higher correlation in the former case

10 Journal of the Aerological Observatory No is due to the fact that DFsolar and DFuv have not been affected by the rainy season at Tsukuba. There is almost no correlation between RDF and RRF. 7. UV (CIE): contributions of DRuv, DFuv, and RFuv to GLuv. The irradiance of direct UV radiation, DRuv on a horizontal surface, can be calculated from the difference between GLuv and DFuv. The seasonal changes of monthly means of daily total DRuv, DFuv, and RFuv are shown in Fig. 7. In panels (a-1), (b-1), (c-1), (d-1), and (e-1) are the solar radiation, the UV (UVB) radiation, the UV (CIE) radiation, the irradiance at a wavelength of 310 nm, and the irradiance at a wavelength of 320 nm, respectively. In the right-hand panel of each pair of figures, (a-2), (b-2), (c-2), (d-2), and (e-2), are shown the contributions (percentages) of DRuv, DFuv, and RFuv, respectively. The difference between the seasonal change of solar radiation and UV radiation are clearly apparent. The contribution of DRsolar to GLsolar was about 50%, except during the summer season. In contrast, the contribution of DRuv to GLuv was about 20% during all seasons. In contrast, the contribution of RFuv to GLuv evidences much smaller intensities than the contribution of RFsolar to GLsolar. 8. Conclusions Monitoring observations of spectral UV (RFuv) reflected from the ground surface have been carried out by using a modified Brewer spectrophotometer, BR#059, since December of 2003; similar monitoring of diffuse spectral UV (DFuv) in the sky has been carried out with a Brewer spectrophotometer, BR#058, with an automated shadow unit, since August of 2005, at Tsukuba. In this paper, the results of these measurements are compared with global spectral UV, GLuv, and solar radiation, GLsolar, DFsolar, and RFsolar, during the observation period of 9 (DF: 8) years from 2004 (2005) to These data are summarized in Table 2, and conclusions about seasonal changes in the data are summarized as follows. (1) GLuv, DFuv, and RFuv The annual time series of GLuv was unimodal, with a peak in August. In contrast, the time series of GLsolar was bimodal, with peaks in May and August. The inter-annual changes of GLuv evidenced a local minimum in 2006 and a pattern of small increases since GLsolar evidenced similar patterns. The annual time series of DFuv was unimodal, with a peak in August. In contrast, the annual time series of GLsolar was unimodal, with a peak in June. The inter-annual change of DFuv evidenced a pattern of small increases since 2006, but DFsolar Table 2 The averages of annual means of daily total irradiances and daily irradiance ratios from 2005 to 2012 at Tsukuba. (a): GLsolar, GLuv, DFsolar, DFuv, RFsolar and RFuv. (b): RDFsolar, RDFuv, RRFsolar, and RRFuv. (c): Irradiance ratios of "GLuv/GLsolar", "DFuv/DFsolar" and "RFuv/RFsolar". (a) GL DF RF Solar (KJ/m 2 ) UV (UVB) (J/m 2 ) UV (CIE) (J/m 2 ) (b) RDF: DF/GL RRF: RF/GL Ratio (Percent) Ratio (Percent) Solar (60.5) (20.5) UV (UVB) (84.3) (2.0) UV (CIE) (84.3) (2.0) (c) GLuv/GLsolar DFuv/DFsolar RFuv/RFsolar Ratio (Percent) Ratio (Percent) Ratio (Percent) UV (UVB) (0.1087) (0.1749) (0.0087) UV (CIE) (0.0115) (0.0184) (0.0009) was nearly constant from year to year. The annual time series of RFuv displayed a bimodal pattern, with peaks in April and August. This seasonal pattern was similar to the seasonal pattern of RFsolar, but the pattern was strongly affected by snow cover and vegetation. The RFuv has been increasing by small amounts from year to year since a local minimum in (2) RDFuv and RRFuv The average RDFuv ratio was high, about 0.84 (84%), during the observation period. In contrast, the average RDFsolar was about 0.61 (61%). The RDFuv showed little seasonality and no remarkable maximum or minimum periods. On the contrary, the annual time series of RDFsolar was bimodal, with peaks in June and October. The average of the RRFuv ratio was low, about 0.02 (2%), during the observation period. On the contrary, the average RRFsolar ratio was about 0.21 (21%). The RRFuv ratio was only about 1/10 of the RRFsolar ratio. The seasonal pattern of RRFuv was much affected by snow cover, but snow cover had little effect on the seasonal pattern of RRFsolar. The RRFuv has been increasing by small amounts from year to year since 2007 because of snow cover effects (snow cover days has been increasing). (3) GLuv/GLsolar, DFuv/DFsolar, and RFuv/RFsolar The GLuv (CIE) / GLsolar irradiance ratio indicates an intensity of about (0.0115%). The annual time series of the ratio is characterized by a unimodal pattern with a broad peak. The ratio has been nearly constant from year to year since

11 Seasonal Variations of Global, Reflected, and Diffuse Spectral UV Observations based on Brewer Spectrophotometers at Tsukuba, 2004 to 2012 Table 3 Annual means of monthly means of DF, DR, and RF irradiance percentage against to GL (100%) from 2005 to 2012 at Tsukuba. GL DF DR RF Solar Radiation UV (UVB) UV (CIE) nm nm nm (percent) The DFuv (CIE) / DFsolar irradiance ratio indicates an intensity of about (0.0184%). The annual time series of the ratio is triangular in shape. The ratio has been increasing by small amounts from year to year since The RFuv (CIE) / RFsolar irradiance ratio indicates a low intensity of about (0.0009%). The annual time series of the ratio is trapezoidal in shape. The ratio has been changing by small amounts from year to year since (4) The relations between UV radiation and solar radiation The DFsolar is more highly correlated with DFuv than is GLsolar with GLuv. The reason for the higher correlation in the former case reflects the fact that DFsolar and DFuv are not affected by the rainy season at Tsukuba. (5) The UV (CIE) contributions of three components, DRuv, DFuv, and RFuv to GLuv. The average contributions (percentages) of the daily totals of three components, DRuv (i.e., GLuv DFuv), DFuv, and RFuv to GLuv are shown in Table 3. The contributions of DRuv, DFuv, and RFuv compared to GLuv were calculated to be about 20%, 80%, and 2%, respectively. For comparison, the contributions of DRsolar, DFsolar, and RFsolar compared to GLsolar were about 50%, 50%, and 20%, respectively. We were able to elucidate the differences of seasonal variations between the UV radiation, GLuv, DFuv, and RFuv, and solar radiation, GLsolar, DFsolar, and RFsolar. In particular, RFuv is affected by the reflective effect of snow cover and the absorptive effects of vegetation. Continuous observations with Brewer spectrophotometers are very important for, inter alia, ground truth calibration of satellite observations, clarification of aerosol distribution, and studies of vegetation growth. Acknowledgments We would like to thank Dr. D. I. Wardle, Dr. V. Savastiouk, Mr. T. Grajnar, and Mr. M. Brohart of the Experimental Studies Division of the Meteorological Service of Canada; Mr. I. B.A.H. Dieterrink of Kipp & Zonen Inc. in the Netherlands; and the members of the Brewer Workshop. We also thank the staff members of the Aerological Observatory and Ozone Layer Monitoring Office, JMA. References AOT/JMA(1996) : "Radiation Data in Tsukuba", Aerological Observatory, JMA, 196pp. BSRN (2002) : BSRN 7th Report. WCRP/BSRN, 53pp. Ito, M (2004) : Reflected spectral UVB observation on the ground surface using modified Brewer spectrophotometer. Jour. of Aerological Observatory, 64, Ito, M (2005a) : Diffuse spectral UVB routine observation using Brewer spectrophotometer and simple shadow unit. Jour. of Aerological Observatory, 65, Ito, M (2005b) : Reflected spectral UVB routine observation on the ground surface using modified Brewer spectrophotometer at Tsukuba in Jour. of Aerological Observatory, 65, Ito, M. (2006) : Diffuse spectral UVB observation using Brewer spectrophotometer and new automated shadow unit. Jour. of the Aerological Observatory, 66, Ito, M. (2007) : Reflected and diffuse spectral UV observations using Brewer spectrophotometers at Tsukuba. Jour. of Aerological Observatory, 67, Ito, T., Ueno, T., Kajihara, R., Shitamichi, M., Uekubo, T., Ito, M. and M. Kobayashi (1991) : Development of monitoring technique of ultraviolet irradiance on the ground - an assessment of UV-B increase due to ozone depletion based on spectral observations -. Jour. of Meteorological Research, 43, Kipp & Zonen (1996) : Brewer MKIII Spectrophotometer Final Test Record, BR#174. Kipp & Zonen Inc., 75pp. Kipp & Zonen (2008a) : Brewer MKIII Spectrophotometer Operators Manual. Kipp & Zonen Inc., 132pp. Kipp & Zonen (2008b) : Brewer MKIII Spectrophotometer Service Manual. Kipp & Zonen Inc., 125pp. McElroy, C.T., V. Savastiouk and T. Grajnar (2008) : Standard operating procedures manual for the Brewer Spectrophotometer, Ver. D.01. Environment Canada, 138pp. MOE (2004) : "Shigaisen-Hoken-Shido-Manual". Ministry of the Environment, 38pp. Ohmura, A. (2002):Examination of shadow mechanism for diffuse sky irradiance measurement for the BSRN use. Report for the seventh BSRN science and review workshop, WCRP Informal Report No.18/2002, Annex 3 to Diffuse Geometry WG Report, 4pp. WMO (1996):Guide to meteorological instruments and methods of observation. WMO, No.188, I I

12 Journal of the Aerological Observatory No つくばにおけるブリューワー分光光度計を使用した波長別全天 散乱 地面反射紫外線量の季節変化 2004~2012 年 伊藤真人 * 居島修 * 島村哲也 * 上里至 * 能登美之 * 江崎雄治 * * 押木徳明要旨ブリューワー分光光度計を改造して地面反射波長別紫外線量 (RFuv) の観測を 2003 年 12 月に, またブリューワー分光光度計用自動遮蔽装置を開発し散乱波長別紫外線量 (DFuv) の観測を 2005 年 8 月より開始した. 本稿では, 現在までに蓄積された約 9 年間 (DFuv は約 8 年間 ) のデータにより, 両成分の年変化の特性について全天波長別紫外線量 (GLuv) や全天 散乱 地面反射日射量 (GLsolar,DFsolar,RFsolar) との比較を行った. これらは, 以下のようにまとめられる. (1) GLuv,DFuv,RFuv の年変化は,GLsolar,DFsolar,RFsolar とはそれぞれ異なった特徴を持つ.(2) DFuv の経年変化は,DFsolar がほぼ一定で変化が認められないのに対し,2006 年以降, 若干の増加傾向を示す.(3) RFuv の年変化や経年変化は, 積雪や植物の生育状況に大きく左右される.RFsolar では, これらの影響は少ない.(4) RDFuv は, 年変化を持たず約 0.84 (84%) の年間を通してほぼ一定で推移する. これに対し,RDFsolar は年変化を有し, 年平均約 0.61(61%) を示す.(5) RRFuv は 0.02(2%) を示し,RRFsolar が 0.21 (21%) を示すのに対し低い値をとる. また,RRFuv は微弱な波長依存性を持ち, 長波長側で若干大きくなる.(6) 紫外線量の各成分の割合は,GLuv を 100% とすると DRuv が約 20%,DFuv が約 80%,RFuv が約 2% となり, 日射 Solar の場合の約 50%, 約 50%, 約 20% と大きく異なる. 以上にように,GLuv,DFuv,RFuv の年変化と経年変化の特徴を把握することができた. 今後, これらのデータが, 衛星観測の地上較生, エアロゾル挙動の研究, 植物生育の研究等々の分野において利用されることを期待したい. * 気象庁高層気象台観測第三課