December 1986 H. Ikeda, T. Aoshima and Y. Miyake 987

Development of a New Sunshine-Duration Meter

By Hiroshi Ikeda, Takeshi Aoshima and Yukiharu Miyake

EKO Instruments Trading Co., Ltd. 21-8 Hatagaya 1-chome, Shibuya-ku, Tokyo 151, Japan (Manuscript received 24 July 1986, in revised form 16 October 1986)

Abstract

A new type of sunshine-duration meter,1) providing an optical system that consists of a mirror and a photodetector, was recently developed. The photodetector looks sky through a mirror rotating around an axis parallel to that of the 's. A uniform sensitivity, independent of the incidence angle of the sun, is performed by an optical mirror with two diffusible surfaces. Effect of diffuse sky- light has been remarkably well eliminated by an adoption of a pyroelectric detector.

1. Introduction they do not always give same results. In order Measurement of sunshine-duration is one of to establish an intercomparison method of meas- the oldest solar radiation measurements. Sun- uring results between different types of sun- shine-duration data are valuable for two main shine recorders, the World Meteorological Organi- purposes. First is one of the primary parameters zation recommended that all data should be for characterizing the climate of a given location. referenced to designated standard instruments. Second is valuable for estimating the total flux In 1962, the Commission for Instruments and of solar radiation at a place where no pyrano- Method of Observation of the World Mete- meteric measurement is available. Because of orological Organization (CIMO) adopted the the simplicity, convenience and relatively low Campbell-Stokes sunshine recorder as the stand- cost, a large number of stations are performing ard reference instrument. However, one problem measurements of sunshine-duration. And a in using the instrument was reported: it is dif- number of different types of sunshine-duration ficult to define a precise lower limit of direct meters have been developed over the last 140 radiant flux. In the extreme condition of a years (Coulson,1975). clear, dry atmosphere and a very dry card, the The duration of sunshine is defined as the threshold is as low as 70W/m2, while in the amount of time when the disk of the sun is not opposite extreme conditions it increases to as obscured by , but sunshine-duration meas- much as 280W/m2 (Coulson,1975). ured actually is when the intensity of direct In 1981 CIMO, considering the problem on solar radiation is large enough to activate the the standard reference, adopted a new threshold meter. Since the various sunshine recorders value for bright sunshine of 120W/m2 with an have different threshold values for activation, accuracy of *20%, and recommended a pyrhelio- meter as a reference instrument for sunshine- duration measurement (WMO CIMO-V-III, 1) This type of sunshine-duration meter described in 1981). this paper was patented by Japanese patent application No. 1111917, and British patent application No. This paper described a newly developed sun- 8324831. The sunshine-duration meter has been shine-duration meter which satisfies the recom- adopted officially as net work instruments by the mendation of CIMO 1981. Japan Meteorological Agency in 1986. 988 Journal of the Meteorological Society of Japan Vol. 64, No. 6

2. Construction ing on the solar declination. In order to provide The newly developed sunshine-duration meter a uniform electrical output in proportion to the has two main parts, an optical system and a base direct , the reflectance of the mirror has been made to be independent of the (Fig. 1). The optical system consists of a photo- detector, a mirror and a stepping motor driving incidence angles by using two diffusive surfaces the mirror. They are mounted on the body. A (see Fig. 2). cylindrical glass cover protects optical system The field of view of the sunshine-duration from the environment. The metal of chromium meter can be defined from the optical character- istics of the mirror. Fig. 3 illustrates the field is plated on the upper part of the glass sur- face to prevent rapid temperature increase of of view on the celestrial sphere. A dense dotted the photodetector induced by incoming solar area in Fig. 3 shows the field of view in a given radiation. A spirit level and screws are provided moment, and the dotted area shows the sweep- out area. The lengthes of arc BAC (or B'A'C') for adjusting the base horizontally and a vertical and arc AA' (or BB', CC') are defined by the post is provided for mounting the optical system characteristics of the mirror. When it is assumed on the base of the instrument. A circuit board for electric convertor is mounted inside the base that the field of view is occupied with the area in airtight. where the relative reflectance of the mirror rises The sunshine-duration meter is installed on a by above 50%, the length of BAC (*90*) and horizontal surface so that the stepping motor AA' (*8*) can be estimated approximately to of the optical system faces the south. The mirror 0.20 steradian field of view. rotated with the stepping motor reflects irradi- On the other hand, the sensitivity of the ance from a part of the sky to the photodetector, N.I.P. (Normal Incidence Pyrheliometer) was which generates output in proportional to the investigated depending on deviation angles of irradiance. When the sun enters the field of view, the sun from the center of the field of view. It a pulse of higher amplitude is generated. was shown that the field of view occupied by the area having the sensitivity above 50% is about The incidence angle of the direct to ±3* and corresponds to 0.009 steradian showing the mirror varies in the range of *23.5* depend- only 1/22 of that of the sunshine-duration meter. It is well known that scattering by cloud and haze has a remarkable effect on the skylight

Fig. 1. Schematic view of the sunshine-duration meter. 1: stepping motor; 2: mirror; 3: photodetector; 4: glass cover; 5: body; 6: post; 7: base. Fig. 2. The mirror of the sunshine-duration meter. December 1986 H. Ikeda, T. Aoshima and Y. Miyake 989

Fig. 3. The field of view in the celestrial sphere.

intensity and that the effect is most pronounced in the region of the solar aureole. Therefore, it

is supposed that the measurement of direct solar Fig. 4. Time response of pyroelectric detector. radiation with the sunshine-duration meter having such wide field of view can be consider- ably influenced by the diffuse skylight. pyrheliometer, and a reference voltage for the However, the problem of the field of view comparator is adjusted with THRESHOLD has been considerably improved by adopting a ADJUST to 0.84 volts corresponding to 120 W/m2 designated by CIMO 1981. The two pyroelectric detector. As shown in Fig. 4, the revolution speeds of the mirror, 100 and 120 time response curve of the detector is a pattern of polarization. Its output is in proportion to r.p.h. are alternatively available depending on a data logger. The actuation of the sun switch the time rate of change of irradiance and is use- specified as 30 times per hour is realized by ful in eliminating the background of diffuse 120 r.p.h. and 2 binary counters, and 100 times skylight. actuation per hour is realized by 100 r.p.h. In The output of the pyroelectric detector is this case, the binary counters are removed. transmitted to the electric convertor producing two types of output: one is a sunshine switch 3. Characteristics and the other is an analog output. As shown The threshold value is affected by reflectance in Fig. 5, the output of the pyroelectric detector of the mirror, nonlinearity of the analog output, is converted to a positive pulses by means of thermal dependency of the sensitivity and condi- peak hold. Time constant of the peak hold is tions of the diffuse skylight. adjusted to match that of analogue recorder to Ideally, the reflectance of the mirror must be be connected. The analogue output is provided for the sensitivity adjustment of the sunshine- independent of the incidence angle of sunlight. duration meter. On the other hand, this pulses In an early investigation, some kinds of ideal is also transmitted to a comparator, and when specular mirror having a curvature was tried, however it was found difficulty to produce a the peak goes beyond some threshold value, the mirror with permissible accuracy. Furthermore comparator triggers a binary counter. When the it was found that the reflectance of a specular counter counts the pulses some given times mirror is considerably affected by a small crack, successively, one shot of pulse operates the small uneveness and small contamination on relay, and a switch of the relay works as a sun- its surface. As a result, it was concluded that shine switch. The sensitivity of the sunshine- the mirror having a diffusive surface (make of duration meter is adjusted with GAIN ADJUST aluminum) is most suitable from. the view point to 7 volts per kWm-2 in comparison with a of optical stability and also easy to process the 990 Journal of the Meteorological Society of Japan Vol. 64, No. 6

Fig. 5. Circuit diagram of the convertor, production. the lamp and the detector, the distance between As shown in Fig. 2, the mirror adopted in the mirror and the lamp, and the revolutions of this paper consists of two surfaces making angles the mirror are maintained without alteration. of 30* and 60* to the axis; and each surface is These discrepancies depending on incidence an- given some diffusivity by grinding with emery gles are plotted in Fig. 7. The remarkable scatter paper to a direction of traverse to the axis. Thus the reflectance of the mirror is given by a syn- thetic result of the reflectances of the two sur- faces. Two convenient methods for the mirror ex- amination have been provided: 1) a sun method in which the optical system of the sunshine- duration meter is installed on the post so that incidence angle of irradiance to the mirror can be controlled in a range of the -25 and +25* sun declinations and 2) a lamp method in which the optical system is placed on the post on an optical bench with a lamp of Halogen 250 watt. The distance between the mirror and the lamp is about 50cm. As shown in Fig. 6, significant discrepancies exist between the two methods which may be caused by a complicated inter- action between different type of light sources and by response time of the detector itself. It can be pointed out that the response time of the detector is on the same order as the time that the image of the light source passes through the detector. The reason for the discrepancy is due to the different shape of image projected on the detector and by the speed of that image as it Fig. 6. Examples of the mirror reflectance. pass through the detector. However it can be solid line: by the sun method; concluded that these systematic discrepancies broken line: by the lamp method; can be conserved as far as the specifications of point: measured point. December 1986 H. Ikeda, T. Aoshima and Y. Miyake 991

Fig. 8. Frequency distribution of temperature co- efficient of the pyroelectric detector sensitivity.

Fig. 7. Discrepancies of mirror reflectances between the sun and the lamp methods. may be caused by an irregularity belonging to each mirror. The averaged curve showing sys- tematic discrepancies is useful for performing the mirror production or the mirror examination with the lamp method. As shown in Fig. 6, the accuracy of the mirror can be easily improved to be less than *2 or *3% over a range of incidence angles *23.5*. In actually the mirrors are sup- plied under a specified accuracy *3% typically cally and *5% in maximum. The pyroelectric detector having two win- dows, a differential type (P2LD-F Horiba Japan) is .used. The pyroelectric detector without win- dow material has uniform sensitivity in a wide range of wavelength. In this study, quartz glass was selected as the window material, because Fig. 9. Comparison between output of sunshine- the measurement of the solar radiation must be duration meter with that of pyrheliometer in the accommodate the 0.3-3*m of wavelength. clearsky. Moreover, one of the window is evaporated by a metal of low emissivity to intercept irradiance, and* 98% of those belongs to the same by so that the characteristics of detector have been 0.18%/*. In other words, the sensitivity of the improved not only in the stability but in the sampling detectors changes by *3% or less in temperature dependency. The sensitivity of the typically and by *5% in maximum within tem- detector decreases almost linearlly with increas- perature range of *30*. ing temperature in a wide range. Frequency Fig. 9 shows the output of the sunshine dura- distribution of the temperature coefficients of tion meter as plotted in comparison with that of the detectors measured in a temperature range a pyrheliometer for the clear sky condition. -20*40* is shown in Fig. 8. It was found It is observed that the sensitivity of the sunshine that 80% of the sampling detectors belongs to duration meter is decreasing so gradually as in- the decrease of the sensitivity by less than 0.l%/ creasing of the direct solar irradiance. This seems 992, Journal of the Meteorological Society of Japan Vol. 64, No. 6 to be unavoidable because the effect of the that of pyrheliometer, the amount of virtual diffuse skylight contributes to the measurement increasing of the sensitivity that is approxi- of direct solar irradiance in each different man- mately 10% in extreme condition can be said ner. The sensitivity of the sunshine duration to be fairly small. meter is determined in comparison with pyrhelio- When the accuracy of the sunshine-duration meter. Therefore, when the determination is meter is assumed as the accuracy for the thresh- performed in a higher intensity such as 1kWm-2, old value determination, it is also equivalent to an exceeding of the sensitivity by aobut 8% will the accuracy for the measurement of the direct be given at lower level of irradiance; this corre- solar irradiance. An excess of the sensitivity will sponds to a lower value of threshold. For the give too low value of the threshold in the same purpose of accurate determination of the thresh- amount of per cent. old value, it is desired that it must be performed The accuracy of the threshold value deter- as close as possible to the threshold value. mination due to various factors is listed in Table Fig. 10 shows the comparison between the 1. This suggests that the accumulated error of sunshine-duration meter and the pyrheliometer -18*l0% under the extreme conditions might performed in a cloudy sky prevailing with thin exceed the permitted error of *20% defined by (cirrus and cirro-stratus). It was observed WMO (1981). Among of the listed errors, the that the almost data of the sunshine-duration systematic errors caused by the nonlinearlity meter appeared in higher value than that in the and the diffuse skylight are significant. As de- clear sky. The virtual increasing of the sunshine- scribed above, the systematic error caused by duration meter can be also considered as the diffuse skylight is unavoidable. Howeever, when result of the influence of the abounding diffuse the determination of the sensitivity is performed skylight in cloudy sky. However, when it is at the irradiance close to the specified thresh- considered that the field of view of the sunshine- old, the error of -8% arising from nonlinearlity duration meter is very wide in comparison with can be eliminated. As a result, the accumulation of errors under the extreme conditions is -10 10%, that is, it can satisfy the specified accu-* racy. 4. Conclusion According to WMO's recommendation de- fining the threshold value which is based on the direct solar irradiance and its accuracy for sun- shine-duration measurement, a new sunshine duration meter has been developed. To provide the electric output proportional to the direct

Table 1. The accuracies for threshold value determination depending on various factors.

Fig. 10. Comparison between output of sunshine- duration meter with that of pyrheliometer in the cloudy sky. The area enclosed by broken lines shows the permitted limit for the threshold deter- mination. December 1986 H. Ikeda, T. Aoshima and Y. Miyake 993

solar irradiance and independent of incidence Acknowledgements angle varied with solar declination, the mirror having two diffusive surfaces was introduced, The authors wish to thank Mr. S. Koinuma and it rotated with constant speed around earth and Mr. M. Anzai of section manager of Japan axis. Thus automatic operation was realized Meteorological Agency for their help in develop- through year without troublesome maintenance. ing this instrument and we also acknowledge Furthermore the pyroelectric detector whose valuable discussion and suggestion of Mr. T. output is proportional to the time rate change Yamauchi, Mr. K. Matubara, Mr. H. Shimura of irradiance is used as a sensor to eliminate and other staff of the same sections. undesirable effect of the diffuse skylight. This References adoption led also to eliminate error due to spectral variation of direct solar irradiance. As Coulson, K.L., 1975: Solar and Terrestrial Radia- a result when using as sunshine-duration meter tion. Academic Press, 215-234. , WMO CIMO-VIII, 1981: Radiation Measurement: error range as the threshold value is limited The definition of sunshine occurrence in terms of within -10*10% under extreme condition. a threshold of solar irradiance.

新 型 日照 計 の 開発

池 田 弘 ・青 島 武 ・三 宅 行 美 (英弘精機株式会社)

こ の た び 開発 し た新 型 日照 計 に つ い て 述 べ る 。 こ の 日照 計 の 光 学 系 は 反 射 鏡 と光 検 出 器 か ら成 る もの で,光 検 出 器 は,地 軸 に 平 行 な軸 上 を 回 転 す る 反 射 鏡 を と お し て 天 空 を 見 る よ う に な っ て い る。 鏡 面 に は,太 陽 の 入 射 角 に 無 関 係 な 一 様 な 感 度 を得 る た め に,あ る 散 乱 特 性 を与 え て あ る 。 な お 天 空 散 乱 光 の 影 響 は,焦 電 素 子 を光 検 出器 に 用 い る こ と に よ っ て 大 幅 に 減 少 させ る こ とが で き た 。