Examination of Optical Processes in the Atmosphere during Upper-Air Soundings
1. Introduction
Remote sensing is the main source of information for characterizing clouds, in particular, radar and satellite methods. The centimeter band typically associated with meteorological radars is capable of detecting only clouds where precipitation is present (Doviak and Zrnić 2006). Millimeter-wave radars (Görsdorf et al. 2015) can detect clouds without precipitation, but they are typically only employed for research purposes. Satellite remote sensing can detect essentially all type of clouds (WMO 2000). Based on the decoding of optical images at different wavelengths, satellite methods make possible the detection of clouds as well as the determination of their cloud-top height (CTH). However, these methods are subject to errors in both CTH measurement and cloud detection (Mokhov and Schlesinger 1994). Also, polar regions are particularly problematic in determining the presence of clouds through satellite visible and infrared imagery, given that optical characteristics of snow and ice are similar, and CTH in these regions is low (Esau and Chernokulsky 2015). As a result, the reliability of remote sensing methods decreases. In this case, it is critical to have an independent source for obtaining cloud data.
One possibility for extending cloud detection is through the employment of cloud-detection sensors carried on radiosondes. Given that the radiosonde is a disposable device, a fundamental requirement for a cloud-sensing device is low cost. Since the financing of a national hydrometeorological network is carried out from the budget of a given state, most countries cannot afford to significantly increase the cost of these meteorological observations. This economic limitation strongly motives the creation of our cloud-sensing device, which we elaborate upon in the present study.
A cloudy environment differs from a cloudless one by the presence of water droplets and/or ice crystals. This presence of clouds changes the optical properties of the atmospheric medium. Typically, cloud detection sensors are based on an assessment of the optical properties of the medium. The study of optical processes in the atmosphere has been carried out previously using radiosondes (Asano et al. 2004; Kondratiev et al. 1964; Kostyanoy 1975). To ensure high-quality measurements, a system for stabilizing the inclination angle of the radiosonde optical sensor has been proposed. To study the radiative balance, experiments have also been carried out using an inclination angle stabilization system (Philipona et al. 2012). The use of optical sensors based on a conventional photodiode without an inclination angle stabilization system has also been proposed by Nicoll and Harrison (2012). The experiments were conducted prior to the present study. However, at the time when we carried out our experiments, we were unaware of these studies, even though we employ a similar approach.
2. Description of the experimental equipment
Our experiments were carried out using two types of optical sensors and two types of radiosondes. Figure 1 contains photographs of a radiosonde with installed sensors. A distinctive feature of our work is the use of a photodiode in a generator mode. In this mode, the photodiode generates a voltage proportional to the incident luminous flux. A conventional commercial photodiode for home appliances was installed on the radiosonde. Its working band is from 0.4 to 1.1 μm. When illuminated by the sun at noon, the voltage at a load of 100 kΩ is about 200 mV in the direction of maximum sensitivity, which is perpendicular to the plane of the photodiode. The signal is halved when the photodiode is deflected by 60°. Thus, the output signal is proportional to the cosine of the angle between the perpendicular to the plane of the photodiode and the direction toward the radiation source.
External view of a radiosonde with (a) a visible light sensor and (b) a radiosonde with an IR wavelength sensor before being launched at the Dolgoprudnaya station (station No. 27713). White arrows show the direction of the maximum of the sensors’ radiation sensitivity. The radiosonde with the IR sensor was also equipped with an aerosol sensor (the blue block on the bracket).
Citation: Journal of Atmospheric and Oceanic Technology 38, 9; 10.1175/JTECH-D-20-0158.1
External view of a radiosonde with (a) a visible light sensor and (b) a radiosonde with an IR wavelength sensor before being launched at the Dolgoprudnaya station (station No. 27713). White arrows show the direction of the maximum of the sensors’ radiation sensitivity. The radiosonde with the IR sensor was also equipped with an aerosol sensor (the blue block on the bracket).
Citation: Journal of Atmospheric and Oceanic Technology 38, 9; 10.1175/JTECH-D-20-0158.1
External view of a radiosonde with (a) a visible light sensor and (b) a radiosonde with an IR wavelength sensor before being launched at the Dolgoprudnaya station (station No. 27713). White arrows show the direction of the maximum of the sensors’ radiation sensitivity. The radiosonde with the IR sensor was also equipped with an aerosol sensor (the blue block on the bracket).
Citation: Journal of Atmospheric and Oceanic Technology 38, 9; 10.1175/JTECH-D-20-0158.1
The photodiodes are tested in a refrigeration chamber with a cooling capacity down to a temperature of −70°C. The experiments are carried out using radiosonde with an analog signal input. The sampling rate is 1 Hz. The visible light optical sensors are assembled in such way that the maximum of the radiation pattern is oriented along the rope connecting the radiosonde and the balloon (Fig. 1). During the flight, the radiosonde–rope–balloon system behaves like a pendulum with an average deviation from the vertical of about 18°. When the radiosonde flies through a cloud, the intensity of the direct light decreases, the intensity of the light reflected from the cloud particles increases. The change in the nature of signal fluctuations is used to detect cloudiness.
The IR sensor is tested for nighttime operation. The experiments with IR sensors are carried out using an upgraded Russian radar radiosonde, which is used at upper-air stations of the Russian Federation, Fig. 1. The sampling rate in this experimental radiosonde is 3 Hz. The IR sensor is assembled on the InfraRed Thermometer MLX90614 microcircuit (https://mel-prd-cdn.azureedge.net/-/media/files/documents/datasheets/mlx90614-datasheet-melexis.pdf), which is widely used, for example, in handheld devices for measuring human body temperature. The microcircuit is a receiver of infrared radiation in the range of 8–12 μm with a viewing angle of ±50°. It is calibrated by the manufacturer and indicates, in code form, the temperature of an object located in the sensor’s field of view.
The IR sensors are mounted so that the field of view is oriented at an angle of 45° to the rope connecting the radiosonde and the balloon. To determine the presence of clouds, the cloud detection algorithm implemented in the Boltwood Cloud Sensor II (Diffraction Limited; https://diffractionlimited.com/product/boltwood-cloud-sensor-ii/) was used. The Boltwood device has a vertically oriented IR sensor and a thermometer for measuring the air temperature near the device. The IR sensor measures the temperature of the sky. If there is no cloudiness, then the difference between the readings of the vertically oriented IR sensor and the air temperature is more than 14°C. The difference of less than 4°C is observed with a continuous cloud cover. This difference is independent of the presence of sunlight, which allows measurements to be made both during the day and at night. The radiosonde–rope–balloon system swings like a pendulum during flight, so the sensor sequentially measures the temperature at different angles to the horizon. The average angle of deviation from the vertical is 18° (Dubovetsky and Kochin 2015), and the width of the radiation pattern of the IR sensor is ±50°. In clear skies, this causes large fluctuations of the signal, and in the presence of cloudiness, fluctuations are small. Before being launched into flight, the prototype sensor is oriented at different angles to the horizon, both in the presence of clouds and in clear weather. The readings are consistent with the detection principles described above.
3. Observation results
Radiosondes with visible light optical sensors were launched at the standard time 1200 UTC at the Dolgoprudnaya upper-air station (WMO index 27713) near Moscow, Russia (55.93°N, 37.52°E). The launch time was at 1500 local time. The sun’s elevation above the horizon varied from 30° to 50°. More than 20 launches were made in various synoptic conditions (Kochin 2018). To confirm the convergence of the results, two radiosondes were launched simultaneously. The value of the variance of the optical sensor signal was obtained from the radiosonde data, calculated as the arithmetic mean of the squared deviations of the instantaneous signal values from the arithmetic mean of the 20 corresponding radiosonde values. If there are fewer than 20 radiosonde values, then the variance fluctuations increase sharply. The length of the array corresponds to a spatial resolution of 100 m since the radiosonde rises at a speed of 5 m s−1 and the sampling rate is 1 Hz.
The variance array was formed as follows. For the first point, the measured values from 1 to 20 were taken. The calculated variance value was assigned to the average height; that is, the first point is 50 m above ground level. The second point was the values from 2 to 21. Accordingly, the height of the second point in the variance array was 55 m, the third was 60 m, and so on. For the analysis, the launch of the radiosonde at 1200 UTC 10 June 2015 was selected, which was made during the experiment to verify the weather radar software for calculating CTH.
In the experiment, the CTH measurement is carried out with a lidar (laser rangefinder) from the flying laboratory Yak-42D of the Russian Federal Service for Hydrometeorology and Environmental Monitoring (Roshydromet) (Borisov et al. 2012). The radiosonde crossed the upper boundary of the cloud at the same moment. The distance between the aircraft and the radiosonde was about 1 km. According to the aircraft data, continuous stratus clouds with a flat top boundary were observed. Figure 2 shows the measurement results of that launch.
(a) Raw signal of the optical sensor (blue line). The variance of the sensor signal is shown in green. The x axis is the height above the sea level (m); the left y axis is the signal level in units of the analog to digital conversion (ADC) code at a reference voltage of +3.3 V. (b) The same variance of the optical sensor signal in units of the ADC code (green line), and the readings of the radiosondes’ humidity sensors in the paired launch—by MODEM (red line) and Russian MRZ-3 (blue line). The right y-axis scale is given as a percentage of relative humidity.
Citation: Journal of Atmospheric and Oceanic Technology 38, 9; 10.1175/JTECH-D-20-0158.1
(a) Raw signal of the optical sensor (blue line). The variance of the sensor signal is shown in green. The x axis is the height above the sea level (m); the left y axis is the signal level in units of the analog to digital conversion (ADC) code at a reference voltage of +3.3 V. (b) The same variance of the optical sensor signal in units of the ADC code (green line), and the readings of the radiosondes’ humidity sensors in the paired launch—by MODEM (red line) and Russian MRZ-3 (blue line). The right y-axis scale is given as a percentage of relative humidity.
Citation: Journal of Atmospheric and Oceanic Technology 38, 9; 10.1175/JTECH-D-20-0158.1
(a) Raw signal of the optical sensor (blue line). The variance of the sensor signal is shown in green. The x axis is the height above the sea level (m); the left y axis is the signal level in units of the analog to digital conversion (ADC) code at a reference voltage of +3.3 V. (b) The same variance of the optical sensor signal in units of the ADC code (green line), and the readings of the radiosondes’ humidity sensors in the paired launch—by MODEM (red line) and Russian MRZ-3 (blue line). The right y-axis scale is given as a percentage of relative humidity.
Citation: Journal of Atmospheric and Oceanic Technology 38, 9; 10.1175/JTECH-D-20-0158.1
The optical sensor signal variance increases sharply before crossing CTH. This happens because the radiosonde exits the clouds. To calculate the CTH value, the following algorithm is chosen. The reference value of the variance was calculated for the stratosphere, where there is no cloudiness. The data analysis began at a height of 100 m above the ground surface to eliminate reflections from local objects. If the variance was more than 0.3 of the reference, then the atmosphere was considered cloudless. The coefficient 0.3 was chosen empirically from the consideration that the threshold should be 3 times the maximum observed variance in clouds, which did not exceed 0.1 of the reference. Then, the points of the array were sequentially checked. The height was determined, as occurring where the variance increased to 0.3 of the reference value and then the next height, where the variance increased to 0.7 of the reference value. The average between these two heights was defined as the CTH. According to the aircraft data, the optical thickness of stratus clouds is 30 m (Feigelson 1984). When the depth of immersion in the cloud is 30 m, the intensity of the direct light of the sun decreases by a factor of e. Therefore, the variance of the light intensity falling on the photodetector is reduced by at least a factor of 2. Thus, the height at which the variance is halved relative to the variance in a cloudless atmosphere is the optical (visible) CTH.
The presence of cloudiness and the absence of precipitation were detected by visual observations, ground-based radar, and the research aircraft. According to the aircraft data, the CTH was 2800 m. According to the algorithm in the above example, the CTH height turned out to be 2770 m. Figure 2b shows the graphs of the variance change of the optical sensor signal (green line) and graphs of the readings change of the humidity sensors of the radiosondes in a paired launch: the data of the MODEM radiosonde (red line) and the MRZ-3 radiosonde (blue line). At an altitude of about 6000 m, there is a peak in the humidity. Taking into account the calibration errors of the sensors, it can be interpreted as a midlevel cloud. However, according to the aircraft data, there was no midlevel cloud. Thus, using the readings of a humidity sensor alone is not enough for detecting cloudiness using radiosonde measurements.
Figure 3 shows two more examples of changes in the signal due to clouds. It is notable that the variance in a powerful cumulonimbus cloud with intense precipitation decreases to almost zero at a distance of 600–800 m from the CTH. This result for a powerful cumulonimbus cloud is quite expected. The direct radiation of the sun in the cloud attenuates, but the radiation scattered by the drops increases. Large drops of precipitation ranging in size from 0.5 to 4 mm are located below the cloud’s optical boundary (Doviak and Zrnić 2006) and mainly backscatter radiation. As a result of multiple scattering, the radiation becomes anisotropic and the orientation of the sensor does not affect the intensity of the incident radiation.
Features of the change in the raw signal (blue line) and the variance of the optical sensor signal (green line) in clouds: (a) stratus cloud without precipitation and (b) powerful cumulonimbus cloud with intense precipitation.
Citation: Journal of Atmospheric and Oceanic Technology 38, 9; 10.1175/JTECH-D-20-0158.1
Features of the change in the raw signal (blue line) and the variance of the optical sensor signal (green line) in clouds: (a) stratus cloud without precipitation and (b) powerful cumulonimbus cloud with intense precipitation.
Citation: Journal of Atmospheric and Oceanic Technology 38, 9; 10.1175/JTECH-D-20-0158.1
Features of the change in the raw signal (blue line) and the variance of the optical sensor signal (green line) in clouds: (a) stratus cloud without precipitation and (b) powerful cumulonimbus cloud with intense precipitation.
Citation: Journal of Atmospheric and Oceanic Technology 38, 9; 10.1175/JTECH-D-20-0158.1
The behavior of the signal in clouds without precipitation has not yet been explained. As noted previously, the optical thickness of stratus clouds according to the aircraft data is 30 m (Feigelson 1984). At a distance of 100 m or more, the direct solar radiation should decrease to almost zero. The variance should also decrease to zero, but this does not occur. The only explanation is the reemission by small particles. Stratus clouds consist of cloud particles with a characteristic size of about 5 μm. They do not reflect light back but scatter it forward (Van de Hulst 1957). Possibly, this fact leads to self-focusing of direct solar radiation, when cloud particles play the role of reemitting points. This is confirmed in other launches of radiosondes and is not the result an of experimental error. An indirect confirmation of the self-focusing effect is the observation of astronomical objects that are visible through stratus clouds (Kochin 2019). If the direct radiation of stars were determined only by damping, then observations would be impossible.
Radiosondes with IR sensors were also launched at the Dolgoprudnaya upper-air station. Two radiosonde launches with an IR sensor were made. One launch was made on 7 February 2020 in cloudy weather, the other launch was made on 28 February 2020 in clear-sky conditions. Before the launch, the sensors were tested by changing the orientation from horizontal to vertical. In clear skies, the IR sensor in the vertical orientation gave the temperature value 14°C lower than in the horizontal orientation. In cloudy weather, the difference was less than 2°C.
The data obtained fully confirm the proposed hypothesis (Fig. 4). In the cloud, the fluctuations of the IR temperature are small. After crossing the CTH, the fluctuations increase sharply.
IR signal sensor when flying (a) through a clear sky and (b) through the cloud. The CTH is well traced by the humidity sensor readings. The blue line indicates the humidity readings from the radiosonde, the red line indicates the ambient temperature reading from the IR sensor, and the green line indicates the object temperature reading from the IR sensor.
Citation: Journal of Atmospheric and Oceanic Technology 38, 9; 10.1175/JTECH-D-20-0158.1
IR signal sensor when flying (a) through a clear sky and (b) through the cloud. The CTH is well traced by the humidity sensor readings. The blue line indicates the humidity readings from the radiosonde, the red line indicates the ambient temperature reading from the IR sensor, and the green line indicates the object temperature reading from the IR sensor.
Citation: Journal of Atmospheric and Oceanic Technology 38, 9; 10.1175/JTECH-D-20-0158.1
IR signal sensor when flying (a) through a clear sky and (b) through the cloud. The CTH is well traced by the humidity sensor readings. The blue line indicates the humidity readings from the radiosonde, the red line indicates the ambient temperature reading from the IR sensor, and the green line indicates the object temperature reading from the IR sensor.
Citation: Journal of Atmospheric and Oceanic Technology 38, 9; 10.1175/JTECH-D-20-0158.1
4. Profile of the attenuation coefficient of visible light in the atmosphere
The raw optical sensor signal changes by 30% from ground level to tropopause according to Fig. 2a. It is close to the known attenuation of visible light in the atmosphere, taking into account the optical thickness of the atmosphere at 1500 local time. This indicates the possibility of obtaining information about the visible light attenuation. To calculate an attenuation, it is necessary to differentiate the average sensor signal. Before differentiation, the raw signal is corrected for the elevation angle of the sun. The raw signal fluctuates greatly. Simple averaging, as in Philipona et al. (2012), does not give a satisfactory result, because vertical resolution gets worse and profile details are lost. After considering the options, an empirical method for constructing a signal from the maxima was chosen. In a sliding window of 30 values, the maximum of the signal was selected, and this value was written to the file. The resulting file was differentiated in height. The results are shown in Fig. 5.
(a) Graphs of the derivative of the average signal level of the optical sensor (green line) and temperature (blue line). The x axis is the height (km), the left y axis is the ADC (code units per meter), and the right y axis is temperature (°C). (b) The wavelet spectra of the signals from the optical sensor (red line) and the IR sensor (blue line). The y axis is the normalized value of the spectral component; the x axis is time (s).
Citation: Journal of Atmospheric and Oceanic Technology 38, 9; 10.1175/JTECH-D-20-0158.1
(a) Graphs of the derivative of the average signal level of the optical sensor (green line) and temperature (blue line). The x axis is the height (km), the left y axis is the ADC (code units per meter), and the right y axis is temperature (°C). (b) The wavelet spectra of the signals from the optical sensor (red line) and the IR sensor (blue line). The y axis is the normalized value of the spectral component; the x axis is time (s).
Citation: Journal of Atmospheric and Oceanic Technology 38, 9; 10.1175/JTECH-D-20-0158.1
(a) Graphs of the derivative of the average signal level of the optical sensor (green line) and temperature (blue line). The x axis is the height (km), the left y axis is the ADC (code units per meter), and the right y axis is temperature (°C). (b) The wavelet spectra of the signals from the optical sensor (red line) and the IR sensor (blue line). The y axis is the normalized value of the spectral component; the x axis is time (s).
Citation: Journal of Atmospheric and Oceanic Technology 38, 9; 10.1175/JTECH-D-20-0158.1
Figure 5a shows that at heights of 25 and 5 km there are sharp jumps in the derivative of the average signal level of the optical sensor, which coincided with the inversion layers. This corresponds to the well-known concepts of a decrease in the concentration of aerosols in the inversion layer and an increase in their concentration below it, which yields changes in the light attenuation coefficient (Feigelson 1984). In the stratosphere, visible light is almost not attenuated (Feigelson 1984), as compared with in the troposphere. This result was confirmed in all our experiments. However, the existence of such a sharp boundary in the attenuation coefficient has not been previously established.
The reason for the signal fluctuations in clear skies has also remained unclear. In Fig. 5 the spectrum of the visible optical sensor and the IR sensor signal are shown. The fluctuations due to a change in orientation are clearly expressed in the wavelet spectrum of the IR sensor. However, there is a noiselike component in the spectrum of the optical sensor, which cannot be explained by a changes in sensor orientation. Similar results were obtained in Philipona et al. (2012), in which a sensor with a 180° angle of view and a system of angular position stabilization of the radiosonde was used. There are no signal fluctuations due to changes in the sensor orientation, but they were, in fact, observed.
5. Ability to check humidity sensors
The radio-sounding data undergo a multistage check by the sounding quality monitoring system. The monitoring system reliably detects errors in the temperature measurement. Detecting errors in humidity is very difficult. Recently, the Russian manufacturer of radiosondes changed the opaque caps covering the sensors with transparent caps. Transparent caps were often accidentally left in place. This problem was detected by temperature errors, and no humidity errors were detected.
Potentially, a radiosonde with an optical sensor makes it possible to assess the quality of a humidity sensor according to several criteria. The humidity profiles in clouds and cloudless atmospheres differ from each other. Therefore, the humidity profile can become control information when cloudiness is present according to the optical sensor readings. If the optical sensor indicates the presence of clouds at a certain height, then the humidity sensor readings cannot be less than 50%, taking into account calibration errors. This will eliminate cases in which, for example, the protective cap is not removed from the sensor, leading to an underestimation of the humidity values. The optical sensor data can also detect precipitation particles in the clouds. If there are precipitation particles in the clouds, then the relative humidity approaches 90%–100%. A properly calibrated humidity sensor should show a value between 70% and 100% in the presence of clouds with precipitation particles, according to the optical sensor readings.
The measurement of light attenuation can be used to evaluate the quality of the humidity sensor in the stratosphere. Light attenuation occurs due to the presence of aerosols. Aerosols are practically absent in the stratosphere, resulting in no light attenuation, as can be seen in Fig. 5a. Both aerosol and water vapor are transported from Earth’s surface to the atmosphere by turbulent transport processes. Turbulence abruptly decreases in the tropopause region. Therefore, above the tropopause, the concentration of both aerosol and water vapor is low. The results of the relative humidity measurement are shown in Fig. 2b. The MODEM radiosonde (red line) above the tropopause gives close to zero humidity values that correspond to current ideas. The MRZ-3 sensor (blue line) gives nonzero humidity values, which are erroneous. The height of the boundary between the troposphere and the stratosphere can therefore be determined by changing the attenuation coefficient. Its vertical profile according to the optical sensor data is shown in Fig. 5a. The change in the attenuation coefficient (Fig. 5a) and the vertical relative humidity profile (Fig. 2b) occur at the same height. Thus, the erroneous nonzero data of the MRZ-3 radiosonde above the tropopause can be corrected or removed from processing.
Another problem is that the humidity sensor gets wet in precipitation. Therefore, above the clouds, a wetted sensor results in overestimated humidity values. The Vaisala company produces the RS92 radiosonde, which had two sensors, heated alternately. While one heats up to dry, the other measures the humidity. Then the dry sensor measures the humidity, and the other one heats up (Jensen et al. 2016). This is a good solution, but it leads to a noticeable increase in the cost of the radiosonde. Other radiosonde manufacturers do not use this solution. In this case, an alternative may be a radiosonde with an optical sensor that determines CTH. Above the clouds, the humidity corresponds to a free atmosphere. If the humidity sensor is wet and shows a humidity value of 90%–100% above the cloud, then its readings can be adjusted.
6. Conclusions
In the experiments of Nicoll and Harrison (2012), it was shown that at the border of the cloud and cloudless space there is a change in the signal variation of the optical sensor installed on the radiosonde. However, because of the lack of measurement of CTH by another method, the ratio between the height of the variation change and the actual height of the optical upper cloud boundary was not measured by them. Our experiments were made during the testing of the weather radar software for determining CTH, which used an aircraft with specialized equipment for measuring CTH. Therefore, we were able to directly compare with the aircraft data, which not only confirmed the possibility of measuring CTH but also demonstrated the accuracy of the method. The developed method made it possible to calculate CTH according to the optical sensor data. The difference with the aircraft data was 30 m. Nicoll and Harrison (2012) assumed that the reason for the change in variation is a change in the orientation of the sensor that is due to the pendulum oscillations of the radiosonde–rope–balloon system. When we started our experiments, we also adhered to this hypothesis, but it turned out to be wrong. The fluctuation spectra correspond to a noiselike signal and not to an oscillatory process. Thus, fluctuations in the output signal of the sensor arise from a certain process, the nature of which is not yet clear. Fluctuations due to pendulum oscillations are also present in the spectrum, but their amplitude is less than the noiselike ones.
We also found that the signal variation of the optical sensor does not decrease to zero in clouds without precipitation, and in clouds with precipitation, the signal variation is almost zero already at an altitude of 600–800 m below the optical upper boundary of the cloud. This conclusion is made by comparing the data of the weather radar and the optical sensor. This allows to detect the presence of precipitation particles in the clouds.
To measure CTH at night, a widespread IR sensor MLX90614 was installed on the radiosonde. Its readings also make it possible to measure CTH with close accuracy.
Radiosondes with precision sensors and a stabilized platform are used to measure the vertical profile of the light attenuation coefficient in the atmosphere. We were able to estimate the attenuation without using the sensor position stabilization, which allows us to make measurements with a conventional radiosonde. The results of the measurements revealed a sharp decrease in the attenuation coefficient at the boundary of the troposphere and stratosphere. This makes it possible to measure the integral thickness of the aerosol layer of the atmosphere.
These results can be used to monitor remote satellite methods for detecting clouds and measuring CTH. This is especially important in the Arctic, where cloud detection by satellite methods has low reliability. Also, the data of the optical sensor of the visible range allow one to monitor the operation of the humidity sensor of the radiosonde.
The ease of use and low cost of consumables are crucial factors for devices used in the meteorological network. Unlike Nicoll and Harrison (2012), we use a photodiode in the generator mode. This will simplify and reduce the cost of production, because of the lack of an external amplifier. The average price of a radiosonde is now around USD 100. The estimated additional cost of a radiosonde with an optical sensor will be about USD 2, or an overall cost that is 2% higher than it is now. Thus, the additional costs will not exceed USD 1,000 per upper-air station per year. For example, the cost of an annual supply of radiosondes with balloons is about USD 100,000. The cost of a radiosonde with an IR sensor will be 20 USD higher, that is, about USD 10,000 per year for one station. The extra cost of an IR sensor would make it suitable only for a small number of stations in industrialized countries. Most modern sensing systems allow connecting additional sensors and archiving measurement results. The software for calculating the variance and determining the CTH can be installed remotely, since all stations transfer data over the internet. Therefore, the implementation of a visible light sensor could be done without significant additional costs.
Acknowledgments
The authors thank the staff of the Dolgoprudny upper-air sounding station for their assistance in conducting radiosonde launches. The authors are grateful to Bruce Ingleby for useful discussions and especially valuable comments on the paper. We thank the reviewer David Adams for his comments and help with the English.
Data availability statement
The data of the radiosonde with an optical sensor used in this article is available in an online archive (https://data.mendeley.com/datasets/5fc36g4kfz/1).
REFERENCES
Asano, S., Y. Yoshida, Y. Miyake, and K. Nakamura, 2004: Development of a radiometer- sonde for simultaneously measuring the downward and upward broadband fluxes of shortwave and longwave radiation. J. Meteor. Soc. Japan, 82, 623–637, https://doi.org/10.2151/jmsj.2004.623.
Borisov, Y., and Coauthors, 2012: New Russian aircraft-laboratory Yak-42D “Atmosphere” for environmental research and cloud modification. Proc. 16th Int. Conf. on Clouds and Precipitation, Leipzig, Germany, International Commission of Clouds and Precipitation, http://www.cao-rhms.ru/iccp-2012/ICCP_2012_Programme/media/files/proceeding/24_proceeding.pdf.
Doviak, R. J., and D. S. Zrnić, 2006: Doppler Radars and Meteorological Observations. 2nd ed. Dover, 256 pp.
Dubovetsky, A. Z., and A. V. Kochin, 2015: A specialized radiosonde with a three-coordinate acceleration sensor for studying atmospheric parameters. XXIX All-Russian Symp. on Radar Research of Natural Environments, Saint Petersburg, Russia, Mozhaisky Military Space Academy, 31.
Esau, I. N., and A. V. Chernokulsky, 2015: Convective cloud fields in Atlantic sector of Arctic: Satellite and ground-based observations. Izv. Atmos. Ocean. Phys., 51, 1007–1020, https://doi.org/10.1134/S000143381509008X.
Feigelson, E. M., 1984: Radiation in a Cloudy Atmosphere. Vol. 6, Atmospheric Sciences Library, Springer, 293 pp., https://doi.org/10.1007/978-94-009-6443-3_7.
Görsdorf, U., V. Lehmann, M. Bauer-Pfundstein, G. Peters, D. Vavriv, V. Vinogradov, and V. Volkov, 2015: A 35-GHz polarimetric doppler radar for long-term observations of cloud parameters—Description of system and data processing. J. Atmos. Oceanic Technol., 32, 675–690, https://doi.org/10.1175/JTECH-D-14-00066.1.
Jensen, M. P., D. J. Holdridge, P. Survo, R. Lehtinen, S. Baxter, T. Toto, and K. L. Johnson, 2016: Comparison of Vaisala radiosondes RS41 and RS92 at the ARM Southern Great Plains site. Atmos. Meas. Tech., 9, 3115–3129, https://doi.org/10.5194/amt-9-3115-2016.
Kochin, A. V., 2018: Possibilities of atmosphere optical characteristics measurement during aerological sounding. WMO/CIMO Tech. Conf. on Meteorological and Environmental Instruments and Methods of Observation, Amsterdam, Netherlands, WMO, O2(6), https://library.wmo.int/doc_num.php?explnum_id=5469.
Kochin, A. V., 2019: Daytime and cloudiness astronomy. Proc. All-Russia Conf. Current Problems in Remote Sensing of the Earth from Space, Moscow, Russia, Space Research Institute of the Russian Academy of Sciences, http://conf.rse.geosmis.ru/files/books/2019/7711.htm.
Kondratiev, K. Y., I. Y. Badinov, G. N. Gaevskaya, G. A. Nikolsky, and M. P. Fedorova, 1964: Balloon investigations of radiative fluxes in the free atmosphere. Pure Appl. Geophys., 58, 187–203, https://doi.org/10.1007/BF00879147.
Kostyanoy, G. N., 1975: Accuracy of measurement of long-wave radiation by an actinometric radiosonde. Meteor. Hydrol., 4, 37–41.
Mokhov, I. I., and M. E. Schlesinger, 1994: Analysis of global cloudiness 2. Comparison of ground-based and satellite-based cloud climatologies. J. Geophys. Res., 99, 17 045–17 065, https://doi.org/10.1029/94JD00943.
Nicoll, K. A., and R. G. Harrison, 2012: Balloon-borne disposable radiometer for cloud detection. Rev. Sci. Instrum., 83, 025111, https://doi.org/10.1063/1.3685252.
Philipona, R., A. Kräuchi, and E. Brocard, 2012: Solar and thermal radiation profiles and radiative forcing measured through the atmosphere. Geophys. Res. Lett., 39, L13806, https://doi.org/10.1029/2012GL052087.
Van de Hulst, H. C., 1957: Light Scattering by Small Particles. John Wiley and Sons, 470 pp.
WMO, 2000: Status of the availability and use of satellite data and products by WMO Members. WMO Tech. Doc. WMO/TD 994, SAT-23, 45 pp., https://library.wmo.int/doc_num.php?explnum_id=9743.
