• Dalrymple, G., , and Unsworth M. H. , 1978: Longwave radiation at the ground: III. A radiometer for the ‘representative angle.’. Quart. J. Roy. Meteor. Soc., 104 , 357362.

    • Search Google Scholar
    • Export Citation
  • Dines, W. H., , and Dines L. H. G. , 1927: Monthly mean values of radiation from various parts of the sky at Benon, Oxfordshire. Mem. Roy. Meteor. Soc., 2 , 18.

    • Search Google Scholar
    • Export Citation
  • Goody, R. M., , and Yung Y. L. , 1989: Atmospheric Radiation: Theoretical Basis. 2nd ed. Oxford University Press, 519 pp.

  • Miskolczi, F., , and Guzzi R. , 1993: Effect of nonuniform spectral dome transmittance on the accuracy of infrared radiation measurements using shielded pyrradiometers and pyrgeometers. Appl. Opt., 32 , 32573265.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Philipona, R., , Fröhlich C. , , and Betz Ch , 1995: Characterization of pyrgeometers and the accuracy of atmospheric long-wave radiation measurements. Appl. Opt., 34 , 15981605.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Philipona, R., and Coauthors, 2001: Atmospheric longwave irradiance uncertainty: Pyrgeometers compared to an absolute sky-scanning radiometer, atmospheric emitted radiance interferometer, and radiative transfer model calculations. J. Geophys. Res., 106 , 2812928141.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Robinson, G. D., 1947: Notes on the measurement and estimation of atmospheric radiation. Quart. J. Roy. Meteor. Soc., 73 , 127150.

  • Robinson, G. D., 1950: Notes on the measurement and estimation of atmospheric radiation—2. Quart. J. Roy. Meteor. Soc., 76 , 3751.

  • Unsworth, M. H., , and Monteith J. L. , 1975: Long-wave radiation at the ground I. Angular distribution of incoming radiation. Quart. J. Roy. Meteor. Soc., 101 , 1324.

    • Search Google Scholar
    • Export Citation
  • View in gallery

    Atmospheric longwave radiances at various zenith angles from Dines and Dines (1927). The radiance is multiplied by π to make it correspond to the irradiance provided the atmospheric radiance is isotropic. The plotted values are averages over 515 cloudless sky and 320 overcast sky measurements collected over 5 yr. The data listed in Table 1 of their paper were simply converted into SI units, although they were calculated with a somewhat small Stefan–Boltzmann constant. The gray dashed lines indicate hemispheric averages (irradiance) calculated from the observed radiance. The irradiances for the cloudless and overcast skies agree well with those values obtained by an assumption of homogeneous atmospheric radiance with values observed at the representative angle (52.5°).

  • View in gallery

    (a) Schematic diagram of the directional pyrgeometer for the representative angle. The thermistor is mounted on a sensor holder in good thermal contact with the thermopile sensor. (b) The radiation and rain shields on the directional pyrgeometer are made of vinyl chloride. The sensor is installed in a 16-mm-diameter aluminum pipe and placed at the center of the inner shield. The diameter of the inner shield is about 4 cm.

  • View in gallery

    Directional response of the prototype to a periodically shuttered target. The sensitivity is normalized by a value at the center of the view. The full-width at half-maximum is about 3.5°.

  • View in gallery

    Transmittance of the optical systems. The solid line indicates the transmittance of the total optical system (cut-on filter built into the thermopile sensor and the polyethylene lens) of the present directional pyrgeometer. The dotted line indicates that of the silicon window of CG3 redrawn from the CG3 manual.

  • View in gallery

    Spectral distribution of the net radiant flux, which is effectively measured by the infrared sensor. (a) When the sensor points to a cold blackbody in a laboratory, the electric signal of the sensor is proportional to the difference between the radiant flux from the reference blackbody (Fr) and the radiant emittance of the sensor (F0). The spectrum of the net radiant flux spread over the entire spectral range is indicated by the shaded area. (b) When the sensor points to the atmosphere, the radiant flux from the atmosphere (Fa) is close to the radiant emittance of the sensor (F0) except for the atmospheric window region, and therefore the net radiant flux exchange is constrained to the narrow atmospheric infrared window region.

  • View in gallery

    Time series of measured longwave downward irradiance from the atmosphere under fine weather conditions (7–13 Apr 2007). The gray line indicates the output of the present system and the black line indicates that of the CG3. Shown are the (top) uncorrected and (bottom) corrected values. The total cloud cover (N), the low cloud cover (n), and the cloud forms at low, mid-, and high levels (CL, CM, and CH) are listed at the top of the figure. They were observed at the Kyoto Meteorological Observatory, located about 3 km west of the observation site.

  • View in gallery

    Time series of the measured longwave downward irradiance from the atmosphere under various weather conditions (7–13 Apr 2007). The gray line indicates the output of the present system and the black line indicates that of the CG3. The radiant emittance of the CG3 sensor is also shown by the dotted line. The shaded zones indicate the periods of rain. During and after the periods of rain, the output of the CG3 shows the radiant emittance of the body rather than the downward longwave irradiance of the atmosphere.

  • View in gallery

    Hemispheric sky view and observed irradiances of a typical partly cloudy day (20 Jun 2007). (a) Snapshots, 10-min-average, and 1-h-average views of the sky. The photographs were taken every 30 s by a digital video camera looking down a curved mirror placed horizontally on the ground. They are stacked in the averaged views. The small circles in the images indicate the target positions of the directional pyrgeometer. (b) The irradiance observed by the present system (gray line) and CG3 (black line). In this figure, 10-min running averages were used. (c) Difference between the outputs (irradiances) of the present system and the CG3. The gray line shows the differences of 1-min-averaged data. The dotted and the solid black lines show 10-min and 1-h running-average data, respectively.

  • View in gallery

    Correlation of the atmospheric irradiance obtained by the present system and the reference CG3: (a) 10-min and (b) 1-h averages. The correlation coefficients (r) and the standard deviations (σ) are also shown.

All Time Past Year Past 30 Days
Abstract Views 0 0 0
Full Text Views 15 15 0
PDF Downloads 13 13 0

A Practical Pyrgeometer Using the Representative Angle

View More View Less
  • 1 Graduate School of Human and Environmental Studies, Kyoto University, Kyoto, Japan
© Get Permissions
Full access

Abstract

A simple directional pyrgeometer is tested and compared with a conventional standard pyrgeometer. The system presented in this article has a narrow directional response and points to the representative zenith angle of 52.5°. Because of its directional response, it can be used in a street canyon or in a forest provided that a small part of the sky is visible at the representative angle. The system can be assembled using inexpensive parts that are widely used in household appliances. As this system does not have a flat spectral sensitivity, a spectral correction method is also presented. The results show that the output of the new system agrees well with that from a conventional pyrgeometer (Kipp & Zonen CG3). The correlation coefficient is 0.995 and the standard deviation is 5.6 W m−2 for 1-h averaged values.

* Current affiliation: NS Solutions Corporation, Tokyo, Japan

+ Current affiliation: Denso Corporation, Aichi, Japan

# Current affiliation: Kyoto Municipal Horikawa Senior High School, Kyoto, Japan

Corresponding author address: Satoshi Sakai, Graduate School of Human and Environmental Studies, Kyoto University, Kyoto 606-8501, Japan. Email: sakai@gaia.h.kyoto-u.ac.jp

Abstract

A simple directional pyrgeometer is tested and compared with a conventional standard pyrgeometer. The system presented in this article has a narrow directional response and points to the representative zenith angle of 52.5°. Because of its directional response, it can be used in a street canyon or in a forest provided that a small part of the sky is visible at the representative angle. The system can be assembled using inexpensive parts that are widely used in household appliances. As this system does not have a flat spectral sensitivity, a spectral correction method is also presented. The results show that the output of the new system agrees well with that from a conventional pyrgeometer (Kipp & Zonen CG3). The correlation coefficient is 0.995 and the standard deviation is 5.6 W m−2 for 1-h averaged values.

* Current affiliation: NS Solutions Corporation, Tokyo, Japan

+ Current affiliation: Denso Corporation, Aichi, Japan

# Current affiliation: Kyoto Municipal Horikawa Senior High School, Kyoto, Japan

Corresponding author address: Satoshi Sakai, Graduate School of Human and Environmental Studies, Kyoto University, Kyoto 606-8501, Japan. Email: sakai@gaia.h.kyoto-u.ac.jp

1. Introduction

The earth’s longwave radiative flux is a crucial parameter in determining the energy balance at the surface, and is of great importance not only for the global climate but also for the local thermal environment. However, measuring longwave radiative fluxes is difficult compared with other parameters such as air temperature because it requires expensive equipment and an open sky for operation. In particular, the requirement of an open sky can obstruct the observation of longwave radiation in studies of boundary layer meteorology, which deal with a nonuniform land surface where buildings, trees, and structures on the ground block the sky view.

Currently available commercial pyrgeometers can measure the infrared radiation from the entire hemisphere for determining the irradiance at the ground; however, historically, directional radiometers such as a Linke–Feussner actinometer were used for the measurements of longwave radiation. Dines and Dines (1927) observed the atmospheric radiation from various directions of the sky and found that the atmospheric radiation from the entire hemisphere can be estimated by a single observation at a zenith angle of 52.5°, which is called the representative angle. The existence of such an angle provides an easy methodology for observations of atmospheric radiation, as studied by Robinson (1947, 1950), Unsworth and Monteith (1975), and Dalrymple and Unsworth (1978). However, this conceptually simple observational method has been disregarded after the introduction of the pyrgeometer with a silicon dome by Eppley Laboratories in 1976, which can measure the hemispheric infrared radiative flux.

Recent studies on the global climate demand accurate measurements of the atmospheric longwave radiative fluxes with errors less than 10 W m−2. Therefore, much effort has been put into improving the instrument quality to reduce observational errors (Philipona et al. 1995; Philipona et al. 2001). However, the atmospheric longwave radiative fluxes also have a great impact on the thermal response of the local land surface, where the radiation budget rapidly changes by 100 W m−2 in a very short period with the appearance of low-level clouds. For example, by measuring the atmospheric radiative flux change and the surface temperature, we can estimate the thermal inertia of the surface, which is an important parameter in numerical modeling. In such a case, the observational data even with an error of 10 W m−2 are very helpful.

The purpose of the present paper is to present and discuss a simple methodology for the observations of atmospheric longwave radiative flux at the earth’s surface. Special emphasis is put on ease of operation and low cost rather than high accuracy.

2. Representative angle

Dines and Dines (1927) observed the atmospheric radiation from various directions of the sky over a 5-yr period (Fig. 1). From 515 observations of longwave radiation under cloudless sky conditions and 320 observations under overcast sky conditions, they found a very close correspondence between the atmospheric radiation from the entire hemisphere and the radiation from a zenith angle of 52.5°. This correspondence was found not only in the values averaged over the total period, but also in the monthly averaged values. Longwave radiation under clear-sky conditions does not depend greatly on the position of the sun (Dines and Dines 1927).

The same results were obtained by Robinson (1947, 1950) with Dines and Linke–Feussner radiometers under strictly cloudless measurement conditions. Unsworth and Monteith (1975) and Dalrymple and Unsworth (1978) confirmed that the representative angle has a value of 52.5°, irrespective of the amount of water vapor in the atmosphere under both cloudless and overcast sky conditions. No significant variation in the flux density with the azimuth angle was observed (Unsworth and Monteith 1975). They also found a linear relationship between the atmospheric radiance L(Z) and ln(secZ), where Z is the zenith angle, and showed that this linear relationship is a necessary and sufficient condition for establishing the representative angle of 52.5°.

Robinson (1947) explained the existence of the representative angle based on the diffusivity approximation for a stratified atmosphere. In this approximation, the atmospheric irradiance is calculated from the atmospheric radiance at a single zenith angle, Z = sec−1 (1/r), where r is the diffusivity factor. The best value of the diffusivity factor r is 1.66, which was first proposed by Elsasser in 1942 (Goody and Yung 1989). Using this diffusivity factor, we obtain a zenith angle of 53.0°, which is very close to the 52.5° used in this study.

Note that the sensitivity of the atmospheric radiance on the zenith angle is not large. Figure 1 also shows the dependence of the measurements at the representative angle. The difference in the irradiances from measurements at 37.5° and 67.5° is 24.6 W m−2 for cloudless skies and 4.0 W m−2 for overcast skies. From these values, we can expect that the error in the calculated irradiance due to an error in the zenith angle is less than 1 W m−2 deg−1.

3. Instrumentation

The present instrument is shown schematically in Fig. 2a and a prototype is depicted in Fig. 2b. The device consists of a thermopile sensor with IR lens, a thermistor, and an operational amplifier. The thermopile sensor type is 15TP551N (SEMITEC Ishizuka Electronics) in a TO5 (metal can) package. A cut-on filter (5 μm) is built into the package to avoid errors induced by sunlight. It is sold mainly for noncontact temperature measurement applications. The sensor is installed in an aluminum holder with good thermal contact, and its temperature is measured using the thermistor (103JT050, SEMITEC Ishizuka Electronics) mounted on the holder. A commercially available Fresnel lens made of polyethylene is mounted in front of the holder. The directional response of the prototype to a periodically shuttered target is shown in Fig. 3. The full width at half maximum is about 3.5°.

Because the output of the sensor is typically less than 1 mV, the signal is amplified by 67 dB to eliminate noises on the transmission line. An operational amplifier (AD8571, Analog Devices) is used and operated at 5 V. It has a very low offset voltage (1 μV) and low-drift characteristics. Therefore, we can amplify the signal accurately without any correction method. The above-mentioned parts are assembled in an aluminum pipe with a 16-mm diameter and about a 50-mm length. It weighs only about 40 g. It is installed in a double cylindrical shield, as shown in Fig. 2b, and directed northward at the representative zenith angle to avoid heating by sunlight. This shield also contributes to protect the sensor from exposure to rain. The total weight including the shield is less than 150 g. With this installation, the temperature of the sensor is maintained almost equal to the atmospheric temperature. Because the operational amplifier consumes only 5 mW of electric power, the heat from the electric circuit does not affect the measurement.

The total transmittance of the present system (the polyethylene lens and the cut-on filter built into the thermopile package) is shown in Fig. 4, and it is obtained from the individual transmittance of the lens and the filter supplied by each manufacturer. The transmittance of the silicon window of the reference sensor (CG3) is also plotted. The narrow absorption lines of the present system of around 7 μm are due to the polyethylene lens, while the low transmittance at high wavelength is due to the cut-on filter. This nonuniform transmittance causes an error when we measure an object that emits infrared radiation, which is different from that of a blackbody as shown below.

4. Correction method for nonuniform transmittance

The spectral nonuniformity of the optical system affects the measurement of the infrared radiation as follows. When we calibrate the sensor with a reference blackbody, we associate the output of the sensor with the total radiant emittance of the blackbody having a broad spectral distribution. In practice, however, our sensor has a narrow sensitivity and does not cover the entire spectrum emitted by the blackbody calibration source. This nonuniformity does not affect the measurement of an object with a spectral emittance similar to that of the blackbody used for the calibration. However, this is not the case for an object with a different spectral emittance. The atmosphere is one such object, and therefore, we need some correction method to use a nonuniform spectral device for the measurement of atmospheric radiation. Here, we propose a correction method considering the spectral emittance of the atmosphere.

The radiation balance Q at the sensor surface is given by
i1520-0426-26-3-647-e1
where τ and ɛ are the transmittance and emissivity, respectively, with ɛ = 1 − τ of the optical system, F is the incoming radiant flux, B is the Planck blackbody function, Tp is the temperature of the optical system, and T0 is the temperature of the sensor surface. The emissivity of the sensor surface is assumed to be unity. When the optical system is placed very close to the sensor and the temperature of the total system is uniform, then Tp = T0, and hence Eq. (1) is reduced to
i1520-0426-26-3-647-e2
where ΔF = FB is the net radiant flux when the sensor surface is directly exposed to the incoming radiation without the optical system. In our approach, Eq. (2) shows that the transmittance of the optical system should be applied to the net flux rather than to the incoming flux only. Throughout the discussion in this section, we focus on the “net” radiant flux. Note that Miskolczi and Guzzi (1993) used the effective transmittance only for the incoming flux [Eq. (12) in their paper] to evaluate the effect of nonuniform transmittance. However, Eq. (2) indicates that the effective transmittance for the net radiant flux is more appropriate for evaluating the effects of nonuniformity.
Suppose we calibrate an ideal sensor with a uniform transmittance using a cold blackbody, yielding a sensitivity constant C0. Figure 5a shows the radiant flux at the surface of the sensor plate. Because the thermopile sensor detects the net flux, the shaded area in Fig. 5 is proportional to the output. The net radiant flux Qr and the output voltage of the sensor Vr are given by
i1520-0426-26-3-647-e3
i1520-0426-26-3-647-e4
where σ is the Stefan–Boltzmann constant, ΔFr is the net radiant flux, Fr is the radiant flux emitted from the reference object at a temperature Tr, and F0 is the radiant flux emitted from the sensor surface with temperature T0.
In practice, sensors do not have a flat spectral sensitivity. The nonuniform spectral sensitivity is mainly due to the transmittance of the optical system, that is, the cut-on filter, the window or the dome, the lens used in the system, etc. Assuming that the temperature of the optical system is identical to the temperature of the sensor surface, we can use Eq. (3) with the transmittance of the optical system τ (λ) to obtain the net flux Qr [similar to that in Eq. (2)]. Hence, for a real sensor, Eqs. (3) and (4) become
i1520-0426-26-3-647-e5
i1520-0426-26-3-647-e6
i1520-0426-26-3-647-e7
i1520-0426-26-3-647-e8
where αr is the effective transmittance of the optical system and the primes denote practical values corresponding to the ideal variables defined above. This effective transmittance αr is valid only when we are measuring blackbody materials. In practice, the sensitivity constant C′ is obtained rather than C0 by calibration with a blackbody.
The net radiant flux changes significantly when the sensor is pointed at the atmosphere, as shown in Fig. 5b. Roughly speaking, with the exception of the atmospheric window region (8–13 μm), an atmospheric layer adjacent to the surface absorbs longwave radiation emitted by the atmosphere above and the surface below, and emits thermal radiation corresponding to its own temperature. Therefore, only the net radiant flux emitted in the atmospheric window region effectively generates a sensor signal, provided the temperature of the sensor is equal to that of the atmosphere at the height of the sensor. The net radiant flux Qa and the output voltage of the sensor Va for an ideal sensor with a flat sensitivity is given as follows, similar to Eqs. (3) and (4):
i1520-0426-26-3-647-e9
i1520-0426-26-3-647-e10
Equations (5)(8) for the nonuniform sensitivity system are now
i1520-0426-26-3-647-e11
i1520-0426-26-3-647-e12
i1520-0426-26-3-647-e13
i1520-0426-26-3-647-e14
Equation (14) indicates that we can obtain the net radiant flux from the atmosphere Qa using the practical value V’a obtained by a system with a nonuniform sensitivity. For this procedure, necessary parameters are the sensitivity constant C′ obtained by laboratory calibration using a blackbody and the ratio of the effective transmittance αa/αr. The ratio can be calculated from the transmittance of the system (Fig. 4) and the simple spectral distributions of the net radiant fluxes, as shown in Fig. 5. Assuming that the temperature of the system is 290 K and that of the blackbody for the calibration is 270 K, the net radiant fluxes are expressed as follows:
i1520-0426-26-3-647-e15
i1520-0426-26-3-647-e16
where we assumed that the atmospheric radiation in the window region is equivalent to that of the blackbody at 240 K. The calculated ratio of the effective transmittance for the present system is 1.26. Hence, the voltage output of the thermopile sensor as calibrated using a cold blackbody should be divided by this factor when it is pointed at the sky to measure longwave radiance. This ratio for the CG3 is 1.10. These ratios are greater than 1 because both systems have relatively large transmittances in the atmospheric window region. Please note that these correction factors are not strongly sensitive to the choice of temperatures used in Eqs. (15) and (16). However, we want to point out that in our approach, the spectral transmittance of the system in the atmospheric window region plays a dominant role for the determination of the effective transmittance and hence the correction, while it has relatively small weight in the evaluation of the effects of nonuniformity by Miskolczi and Guzzi (1993).

5. Results

The atmospheric irradiance was measured using the present system and the reference equipment (CG3 mounted on a CNR1, Kipp & Zonen) at Kyoto, Japan, from March to August 2007. The equipment was set on the rooftop of a four-story building where no obstacle blocked the sky view. The temperature of the CG3 was measured using the same type of thermistor (103JT050) as the present system to avoid errors due to differences in the thermal characteristics of the sensors. The data were sampled every second and averaged for 60 samples, and they were recorded every minute.

Figure 6 shows an example of the time series under fine weather conditions (28–30 April 2007) when the sky was almost cloudless, and the measurements at the zenith of 52.5° were expected to represent the atmospheric irradiances, as reported in the previous studies (Dines and Dines 1927; Robinson 1947; Dalrymple and Unsworth 1978). The data were running averaged over 1 h. The cloud cover and the cloud forms observed every 6 h at Kyoto Meteorological Observatory, which is located about 3 km west of the test site, are also shown. The top panel in Fig. 6 shows the atmospheric downward longwave irradiance when the outputs of the sensors are simply added to the radiant emittance of the sensors. It is clear that the readings of the two sensors differ systematically by about 20 W m−2, although both sensors show the same reading for a blackbody in the laboratory. Using the correction factors described in the previous section, the corrected values are plotted in the bottom panel of Fig. 6. The corrected irradiances agree very well with each other, especially when the sky is completely cloudless. Such data under good weather conditions were selected to evaluate the performance of the correction method. For 60 samples of 1-h-averaged data under cloudless conditions, the difference between the measurements of these sensors (the directional sensor minus the CG3) was −1.9 ± 1.7 W m−2. This very good agreement indicates that the correction method described in the previous section works very well.

To demonstrate the responses of the sensors under various weather conditions, the time series of irradiances for 1 week is shown in Fig. 7. The data were averaged over 10 min and plotted every 10 min. They were corrected by the method described in the previous section. In general, the readings from the two sensors agree well, but some distinctions are clear. In the first half of the period, it rained twice. The rainy periods are indicated by shaded zones in the figure, where the readings from the two sensors differ considerably. This is because the rainwater stays on the silicon window of the CG3 and blocks the atmospheric radiation, and therefore, the CG3 measures the radiant emittance of the body rather than the downward component of the atmospheric longwave irradiance. This problem continues for several hours after the rain stops.

Other large differences are visible when low-level clouds are present, which partially cover the sky (9 and 11 April). Because the irradiance of the low-level clouds differs significantly by about 100 W m−2 from that of a clear sky, the instantaneous measurements by the directional sensor under partially cloudy conditions fluctuate in that range, although the amplitude shown in Fig. 7 is suppressed by the 10-min averaging. On the other hand, the CG3 reads out intermediate values within the range of fluctuating values from the directional sensor, because the atmospheric downward longwave irradiance is a spatial average of all radiances over the hemisphere. Hence, it can be expected that the irradiance is obtained by the time average of the observed values in consideration of the ergodic hypothesis.

Figure 8 shows details of a typical partly cloudy day (20 June 2007) when the surface wind speed observed at the Kyoto observatory was 2.0–2.8 m s−1 during the observation period between 0900 and 1500 LT. The hemispheric sky views were recorded using a digital video camera and a curved mirror, as shown in Fig. 8a. Some cumuli at low levels are clearly seen in the snapshots of the views throughout the day. These individual cumuli are smoothed out by 10-min averaging, but relatively large cloud patches are still visible. By averaging over 1 h, almost all clouds are smoothed out and the sky appears uniform. Although the sky views were taken in the visible spectral region, Fig. 8a gives a good idea how the averaging process works.

The atmospheric irradiances measured by the two sensors under these conditions are shown in Figs. 8b and 8c. Figure 8b shows 10-min averages of the atmospheric irradiance obtained with the CG3 and the present system. The present system shows large time variations similar to the data on 9 and 11 April in Fig. 7. Comparing these variations with the video recordings, it is clear that the time variations correspond to the appearance of the clouds from the viewpoint of the directional sensor.

Figure 8c shows the differences between the observed values obtained from the CG3 and the present system with different running-average intervals. The short-time variations seen in the 1-min-averaged data are smoothed out by 10-min averaging, but the effects of large patches of clouds remain in the 10-min-averaged data. These differences are considerably reduced by a 1-h-averaging process. This is consistent with the visible cloud observations shown in Fig. 8a. The differences in 1-h-averaged irradiances measured between the present system and the CG3 (Fig. 8c) are typically about 10 W m−2.

The time scale of the required averaging period is also estimated roughly as follows. Assuming that the visible low-level clouds lie within several kilometers from the observational site and they move away with a speed of several kilometers per second, the time scale to pass through the field of view for the clouds is estimated at about a half an hour. This time scale depends on various conditions, but it provides a rule of thumb to determine the averaging period.

Using the entire dataset composed of measurements from 23 March to 27 August 2007 and excluding the data affected by rain, the total performance of the directional sensor was evaluated. Figure 9 shows the atmospheric downward longwave irradiances obtained from the present system plotted against those from the CG3. For the 10-min-averaged data consisting of about 15 000 samples, the correlation coefficient is 0.989 and the standard deviation is 8.1 W m−2. The maximum absolute difference is 53 W m−2. For the 1-h-averaged data consisting of about 2500 samples, the correlation coefficient is 0.995 and the standard deviation is 5.6 W m−2. The maximum absolute difference is 30 W m−2. The slope of the best fit line is 1.02.

6. Conclusions

A simple longwave radiometer with a narrow directional response function for the representative zenith angle is tested and compared to the CG3, a conventional pyrgeometer. Although the atmospheric radiance measured at a fixed zenith and azimuth angle varies significantly in time under partly cloudy conditions, 1-h-averaged values of the atmospheric downward longwave irradiance agree well with the measurements of the CG3. For the 5-month period of observation, the correlation coefficient is 0.995 and the standard deviation is 5.6 W m−2 for 1-h-averaged values. Since the present radiometer does not require an open sky, it facilitates measurements of the longwave radiation in many meteorological observations including boundary layer meteorology, which deals with a nonuniform land surface. The directional radiometer also has other advantages compared to the dome-type sensor; it is unaffected by rain and dome heating because it can be shielded from direct solar illumination.

A correction method for a sensor with a nonuniform sensitivity is also presented. The result shows very good agreement between the output from the present sensor and that from the reference when both values are corrected. This method enables the use of inexpensive parts that are widely used in many electric appliances. In fact, the present sensor is simpler and uses fewer parts than an ear-use clinical thermometer, which is sold at most drugstores.

Acknowledgments

The transmittances of the cut-on filter in the sensor and the Fresnel lens used in the present system were provided by SEMITEC Ishizuka Electronics Co. Ltd and Horiba Ltd. The observational data at Kyoto Observatory were supplied by the Japan Meteorological Agency. Professors H. Ishikawa and M. Horiguchi at DPRI Kyoto University are acknowledged for providing the reference equipment and data observed at the Ujigawa Open Laboratory. Professor O. Tsukamoto of Okayama University is also acknowledged for valuable comments. The comments by the anonymous reviewers were very helpful in improving the manuscript.

REFERENCES

  • Dalrymple, G., , and Unsworth M. H. , 1978: Longwave radiation at the ground: III. A radiometer for the ‘representative angle.’. Quart. J. Roy. Meteor. Soc., 104 , 357362.

    • Search Google Scholar
    • Export Citation
  • Dines, W. H., , and Dines L. H. G. , 1927: Monthly mean values of radiation from various parts of the sky at Benon, Oxfordshire. Mem. Roy. Meteor. Soc., 2 , 18.

    • Search Google Scholar
    • Export Citation
  • Goody, R. M., , and Yung Y. L. , 1989: Atmospheric Radiation: Theoretical Basis. 2nd ed. Oxford University Press, 519 pp.

  • Miskolczi, F., , and Guzzi R. , 1993: Effect of nonuniform spectral dome transmittance on the accuracy of infrared radiation measurements using shielded pyrradiometers and pyrgeometers. Appl. Opt., 32 , 32573265.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Philipona, R., , Fröhlich C. , , and Betz Ch , 1995: Characterization of pyrgeometers and the accuracy of atmospheric long-wave radiation measurements. Appl. Opt., 34 , 15981605.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Philipona, R., and Coauthors, 2001: Atmospheric longwave irradiance uncertainty: Pyrgeometers compared to an absolute sky-scanning radiometer, atmospheric emitted radiance interferometer, and radiative transfer model calculations. J. Geophys. Res., 106 , 2812928141.

    • Crossref
    • Search Google Scholar
    • Export Citation
  • Robinson, G. D., 1947: Notes on the measurement and estimation of atmospheric radiation. Quart. J. Roy. Meteor. Soc., 73 , 127150.

  • Robinson, G. D., 1950: Notes on the measurement and estimation of atmospheric radiation—2. Quart. J. Roy. Meteor. Soc., 76 , 3751.

  • Unsworth, M. H., , and Monteith J. L. , 1975: Long-wave radiation at the ground I. Angular distribution of incoming radiation. Quart. J. Roy. Meteor. Soc., 101 , 1324.

    • Search Google Scholar
    • Export Citation

Fig. 1.
Fig. 1.

Atmospheric longwave radiances at various zenith angles from Dines and Dines (1927). The radiance is multiplied by π to make it correspond to the irradiance provided the atmospheric radiance is isotropic. The plotted values are averages over 515 cloudless sky and 320 overcast sky measurements collected over 5 yr. The data listed in Table 1 of their paper were simply converted into SI units, although they were calculated with a somewhat small Stefan–Boltzmann constant. The gray dashed lines indicate hemispheric averages (irradiance) calculated from the observed radiance. The irradiances for the cloudless and overcast skies agree well with those values obtained by an assumption of homogeneous atmospheric radiance with values observed at the representative angle (52.5°).

Citation: Journal of Atmospheric and Oceanic Technology 26, 3; 10.1175/2008JTECHA1076.1

Fig. 2.
Fig. 2.

(a) Schematic diagram of the directional pyrgeometer for the representative angle. The thermistor is mounted on a sensor holder in good thermal contact with the thermopile sensor. (b) The radiation and rain shields on the directional pyrgeometer are made of vinyl chloride. The sensor is installed in a 16-mm-diameter aluminum pipe and placed at the center of the inner shield. The diameter of the inner shield is about 4 cm.

Citation: Journal of Atmospheric and Oceanic Technology 26, 3; 10.1175/2008JTECHA1076.1

Fig. 3.
Fig. 3.

Directional response of the prototype to a periodically shuttered target. The sensitivity is normalized by a value at the center of the view. The full-width at half-maximum is about 3.5°.

Citation: Journal of Atmospheric and Oceanic Technology 26, 3; 10.1175/2008JTECHA1076.1

Fig. 4.
Fig. 4.

Transmittance of the optical systems. The solid line indicates the transmittance of the total optical system (cut-on filter built into the thermopile sensor and the polyethylene lens) of the present directional pyrgeometer. The dotted line indicates that of the silicon window of CG3 redrawn from the CG3 manual.

Citation: Journal of Atmospheric and Oceanic Technology 26, 3; 10.1175/2008JTECHA1076.1

Fig. 5.
Fig. 5.

Spectral distribution of the net radiant flux, which is effectively measured by the infrared sensor. (a) When the sensor points to a cold blackbody in a laboratory, the electric signal of the sensor is proportional to the difference between the radiant flux from the reference blackbody (Fr) and the radiant emittance of the sensor (F0). The spectrum of the net radiant flux spread over the entire spectral range is indicated by the shaded area. (b) When the sensor points to the atmosphere, the radiant flux from the atmosphere (Fa) is close to the radiant emittance of the sensor (F0) except for the atmospheric window region, and therefore the net radiant flux exchange is constrained to the narrow atmospheric infrared window region.

Citation: Journal of Atmospheric and Oceanic Technology 26, 3; 10.1175/2008JTECHA1076.1

Fig. 6.
Fig. 6.

Time series of measured longwave downward irradiance from the atmosphere under fine weather conditions (7–13 Apr 2007). The gray line indicates the output of the present system and the black line indicates that of the CG3. Shown are the (top) uncorrected and (bottom) corrected values. The total cloud cover (N), the low cloud cover (n), and the cloud forms at low, mid-, and high levels (CL, CM, and CH) are listed at the top of the figure. They were observed at the Kyoto Meteorological Observatory, located about 3 km west of the observation site.

Citation: Journal of Atmospheric and Oceanic Technology 26, 3; 10.1175/2008JTECHA1076.1

Fig. 7.
Fig. 7.

Time series of the measured longwave downward irradiance from the atmosphere under various weather conditions (7–13 Apr 2007). The gray line indicates the output of the present system and the black line indicates that of the CG3. The radiant emittance of the CG3 sensor is also shown by the dotted line. The shaded zones indicate the periods of rain. During and after the periods of rain, the output of the CG3 shows the radiant emittance of the body rather than the downward longwave irradiance of the atmosphere.

Citation: Journal of Atmospheric and Oceanic Technology 26, 3; 10.1175/2008JTECHA1076.1

Fig. 8.
Fig. 8.

Hemispheric sky view and observed irradiances of a typical partly cloudy day (20 Jun 2007). (a) Snapshots, 10-min-average, and 1-h-average views of the sky. The photographs were taken every 30 s by a digital video camera looking down a curved mirror placed horizontally on the ground. They are stacked in the averaged views. The small circles in the images indicate the target positions of the directional pyrgeometer. (b) The irradiance observed by the present system (gray line) and CG3 (black line). In this figure, 10-min running averages were used. (c) Difference between the outputs (irradiances) of the present system and the CG3. The gray line shows the differences of 1-min-averaged data. The dotted and the solid black lines show 10-min and 1-h running-average data, respectively.

Citation: Journal of Atmospheric and Oceanic Technology 26, 3; 10.1175/2008JTECHA1076.1

Fig. 9.
Fig. 9.

Correlation of the atmospheric irradiance obtained by the present system and the reference CG3: (a) 10-min and (b) 1-h averages. The correlation coefficients (r) and the standard deviations (σ) are also shown.

Citation: Journal of Atmospheric and Oceanic Technology 26, 3; 10.1175/2008JTECHA1076.1

Save