The Indian Space Research Organization (ISRO)’s Vikram Sarabhai Space Centre (VSSC) has developed a new GPS radiosonde, called Pisharoty sonde, with its ground station for atmospheric research and operational meteorology. The latest version of this radiosonde weighs 125 g and uses a bead thermistor, a capacitive humidity sensor, and a GPS receiver module. It computes geopotential heights and pressures using the temperature and GPS altitude profile. This paper describes the radiosonde and its associated ground system, summarizes its different versions, and discusses balloon tests of the newly developed version, including comparisons with internationally accepted high-quality radiosondes.
The Indian Space Research Organization (ISRO)’s Vikram Sarabhai Space Centre (VSSC) has developed a new, low-cost GPS radiosonde system called the Pisharoty sonde system. It is named after P. R. Pisharoty, the famous Indian physicist and meteorologist who founded the Indian Institute of Tropical Meteorology. To meet ISRO requirements for detailed profiles, especially in the lower troposphere, simple operation, light weight, and low cost, the current operational Pisharoty sonde (weighing 125 g, including batteries) has a single electronics board, a bead thermistor, a capacitive relative humidity (RH) sensor, and a GPS module to obtain height and wind data. Pressure is computed hydrostatically. The system provides detailed lower-troposphere profiles at a lesser cost than its available counterparts in the international market. The Pisharoty sonde system supports all ISRO space launch activities and atmospheric research programs. It is also used by agencies like the India Meteorological Department (IMD) and universities for their meteorological studies. The technology developed by ISRO-VSSC is transferred to another ISRO center, Semi-Conductor Laboratories (SCL) in Chandigarh, for manufacturing the sondes. The following sections summarize the latest version of the radiosonde, the ground station and data processing, different versions, validation, and special applications.
2. System configuration
The Pisharoty sonde system consists of two subsystems, namely, the sonde and the ground station. Figure 1 gives a schematic representation of the Pisharoty sonde system.
a. Pisharoty sonde
The sonde consists of sensors for measuring temperature and relative humidity; a sigma delta analog-to-digital converter (ADC) to process sensor data; a GPS module to get the navigation parameters; a microcontroller for initialization, data acquisition, and frame formatting; a transmitter module for carrier generation, modulation, and transmission; an antenna; and a battery.
A photograph of the sonde is given in Fig. 2a and the block diagram in Fig. 2d. The temperature and RH sensor are mounted outside the sonde package on a flexible printed circuit board (PCB), but the batteries and other electronic parts are inside the insulated radiosonde case.
Pressure is calculated from temperature and geopotential height using software in the data processing and display system of the ground station. The temperature sensor is a negative temperature coefficient glass bead thermistor with a base resistance of 1 kΩ at 25°C (0.4-mm diameter and without antiradiation coating). The humidity sensor gives an output voltage proportional to the relative humidity (sensor module of 8 mm × 4 mm × 3 mm covered by an aluminized plastic cap of 15-mm diameter). The voltage outputs from the sensor circuits are collected and processed by the sigma delta ADC sequentially.
The GPS receiver module with an integrated patch antenna gives altitude, time, location, and velocity of the balloon by processing the signals from GPS satellites. The output from the GPS receiver is in National Marine Electronics Association (NMEA) standard messages format.
The microcontroller acquires the data from ADC and GPS, multiplexes the data, and applies error correction coding. Reed–Solomon coding, being best suited for systems prone to burst errors, is used in the sonde system for forward error correction. The telemetry frame with a frame synchronization pattern, multiplexed data, and error coding bytes are fed to the transmission block. Telemetry data are frequency-shift keying (FSK) modulated on a carrier with frequency programmability in the range of 402–406 MHz.
Acquisition, processing, and transmitting electronics, based on commercial off-the-shelf (COTS) components operating at 3.3 V or less, are placed on a single electronic board. Including two AA lithium-thionyl chloride batteries (allowing for operation for more than 4 h), the radiosonde weighs 125 g. Sensor calibration coefficients are stored on chip, and the preparation and initialization procedure takes less than 5 min. Initialization includes programming the transmission frequency to any desired 125-kHz step between 402 and 406 MHz.
The bead thermistor is subject to radiation errors, including heating by sunlight and cooling by radiation to space, as well as a lag in responding to temperature changes as the balloon ascends, with all errors being larger at high altitudes. A solar and IR radiation correction, varying with pressure and sun angle, is applied to compensate for all of these errors. The correction table was derived using comparison flights described in section 3b. For example, at 50 hPa the correction is −2.5°C (the reported temperature is reduced) at high sun angles and +0.6°C at night. While the RH sensor is similarly subject to errors, the RH data are currently not adjusted.
The activities for bulk production of the latest version of the sondes were initiated after building up confidence with sufficient trial and comparison ascents, and this version of the system is planned to be operational by mid-2014.
b. Ground station
The ground station consists of three main systems: (i) antennas and low noise blocks (LNBs), (ii) a Pisharoty sonde receiver, and (iii) a data processing and display system.
For full hemispherical coverage and high-quality signal reception, two independent antennas [a monopole antenna and a quadrifilar helix (QFH) antenna] are used for receiving signals radiated from the sonde. The output of each antenna is fed to the respective LNB, which contains a low-noise amplifier (LNA) and a narrow bandpass filter (BPF). The LNA gives sufficient amplification to the received signal to compensate the signal attenuation due to cable loss. A picture of the antenna assembly is shown in Fig. 2b.
The Pisharoty sonde dual-channel FSK receiver accepts signals from both the antennas simultaneously, demodulates, decodes independently, and sends the data from both the channels to the data processing and display system. The dual-channel receiver system with high sensitivity and Reed–Solomon decoding ensures a good telemetry link (even up to a range of 300 km), and the data loss is less than 0.2% (i.e., fewer than 12 frames out of a total of 6000 frames) in the high-resolution 1-s data file for most of the cases. Error detection schemes, including checksum verification, ensure good-quality data throughout the ascent.
The data processing and display system, developed by ISRO and called Indian Radiosonde Software (IndRoS), is installed on either a Windows XP, Windows-7, or Windows-8 desktop, or a laptop computer with Ethernet interface for Transmission Control Protocol/Internet Protocol (TCP/IP) connectivity to connect to the receiver for data collection or system configuration.
3. Development and testing
a. Pisharoty sonde versions
ISRO-VSSC initiated the development of radiosondes in 2005. In 2006, prototype Pisharoty sonde was first flown from the Tata Institute of Fundamental Research (TIFR) Balloon Facility at Hyderabad and data was received using a prototype sonde receiver. This sonde prototype model had a platinum resistance temperature detector (RTD), a microelectromechanical systems (MEMS) pressure sensor, and a GPS receiver module for acquiring various parameters.
The first operational version of the system includes the PS01_A1 Pisharoty sonde, with similar sensors and GPS receiver as the prototype, a capacitive humidity sensor, a modified data acquisition module, and an upgraded transmitter. Also, the ground system, including the signal acquisition antennas and the data processing software, was further improved. However, this version had no on-chip coefficient storage, no frequency programmability as part of initialization, and no temperature or humidity radiation correction. This system was tested by radiosonde releases from the Vikram Sarabhi Space Centre at Thumba (WMO station 43373) and the ocean Research Vessel Sagar Kanya (call sign VTJR) between September 2008 and October 2010 (Bala Subrahamanyam et al. 2012; Bala Subrahamanyam and Anurose 2011; Anurose et al. 2012; Rao 2008).
Highly accurate pressure sensors capable of working at low temperatures make the sondes heavy and costly. The second operational version of the Pisharoty sonde (PS01_B2) is used routinely at Thumba and in various Indian research programs starting in late 2011. It adopts the software approach to compute pressure using the temperature profile and geopotential height, which is calculated accurately from the geometric height obtained from GPS (Nash et al. 2011). The PS01_A1 and PS01_B2 sensors are identical (although the temperature–humidity sensor PCB is modified), but the data processing and display software allows the user to select pressure data either from the software or from the pressure sensor. The PS01_B2 software also corrects temperature readings to minimize solar radiation errors (corrections were derived from comparison flights with imported radiosondes in mid-2011). Other modifications in sonde, receiver, and software incorporate features like frequency programmability and onboard coefficient storage.
The first two operational versions of the Pisharoty sondes had a platinum wire RTD as a temperature sensor, but because of large radiative heating errors, the third (latest) operational version (PS01_B3) of sondes was developed with a very small glass bead thermistor as a temperature sensor (custom made for VSSC), with solar and infrared radiation corrections in ground software. The pressure sensor is omitted and pressure is computed in ground software. Development and validation of this system has been completed, and the results are discussed in detail in the following section. Specifications of this latest version of sonde are given in Table 1. These are 2σ values (95.5% of the errors in the measurements are within the listed values).
Throughout the development process, the Pisharoty sonde ground station has also improved significantly, of which the major milestones were the antenna design, development of a dual-channel receive chain, the implementation of coding schemes, and modifications in processing software to make it fault tolerant. Development of the receiver system is not discussed in detail because the ground system changes have little effect on the output data biases.
b. Comparisons and operational testing
The latest version (PS01_B3) of Pisharoty sonde (with bead thermistor, and computed pressure, as there is no pressure sensor) was validated by comparison ascents, where the Pisharoty sonde and another radiosonde (Vaisala, Meisei, or Graw) were attached to a 1.5-m stick suspended in a horizontal plane by a 30-m string below the balloon. The balloon was inflated in such a way that the mean ascent rate is 5 m s−1. Comparison ascents with Vaisala RS92-SGPD were conducted at IMD in New Delhi (WMO station 42182, 5 flights from 10 to 13 December 2010), with Graw DFM-06 radiosondes at Kochi Naval Base (WMO station 43353, 4 flights from 27 to 29 January 2011), and with Meisei RS-06G at the National Atmospheric Research Laboratory (NARL) in Gadanki (10 flights from 21 to 23 July 2011). Ascents were performed every few hours to validate system performance during the entire diurnal cycle. Comparisons in Table 2 and Figs. 3–7 include radiation corrections on temperature data, and each difference is the PS01_B3 Pisharoty sonde reading minus the other sonde reading.
Comparison plots are generated as follows:
Pisharoty sonde raw data are available at a 1-Hz rate and vertical profiles of wind parameters are smoothed through a 61-point moving averaging technique to compensate for the disturbances, including balloon oscillations. A total of 15 consecutive points are averaged for temperature, pressure, and RH.
Since the data from Vaisala are available once every 2 s and data from Graw and Meisei are available every second, data points for comparison are taken at 4-s intervals (i.e., points corresponding to time count 1, 5, 9, etc.), which provides sufficiently high-resolution data. If any of these points are not available (a missing point), then linear interpolation is done, provided an adjacent point is available within 4 s. Otherwise, the data point is treated as a missing point in the 4-s interval profile and is not considered for error computation.
Differences between the corresponding values of two sondes for various parameters are calculated.
Before plotting, a 61-point moving averaging is done on differences in the 4-s interval vertical profile to remove random noises.
Typical comparison plots of the Pisharoty sonde with Vaisala and Meisei are shown in Figs. 3–6 (temperature, pressure, humidity, and wind speed and direction, respectively). In Fig. 4, computed Pisharoty sonde pressures are compared with that of Vaisala. For RH, Pisharoty sonde RH is compared only until the temperature drops below −40°C. In all plots, the time count is the number of seconds after the balloon release. In Figs. 3–6, the difference is calculated as the Pisharoty sonde data value minus the Vaisala or Meisei data value at each level. Launch times are in Indian standard time (IST), where IST = UTC + 5.5 h. Data from other comparison ascents also have similar differences for the measured parameters.
To analyze the repeatability and accuracy in the performance of the system with respect to other sondes used in comparison ascents, scatterplots combining data from all the ascents (19 ascents) are generated as described below.
Vertical profiles are smoothed through the moving averaging technique (61 points for wind parameters and 15 points for temperature, pressure, and RH) to compensate for the disturbances, including balloon oscillations. Then, scatterplots for different parameters are drawn by taking the corresponding data points from both the sondes at 10-s intervals (which provides a sufficiently large number of points for the statistical analysis). Linear interpolation is done to obtain the missing points, provided a valid data point is available within 10 points from the missing data.
Scatterplots for various parameters are given in Fig. 7. “PS01” indicates the Pisharoty sonde. Total number of points (N), correlation (R), and standard deviation (SD) are found and these values are given in the graphs. A best-fitting straight line is drawn to study the linearity. The equation of the best-fit line with slope and intercept is also indicated in the figure. In 2 (out of 5) ascents with Vaisala, 2 (out of 10) ascents with Meisei, and 2 (out of 4) ascents with Graw, humidity sensors were not used in Pisharoty sondes, as the main intention of these comparison ascents was the validation of temperature data. So, for these ascents RH is not compared. For other ascents, RH is considered until the temperature drops below −40°C.
To further analyze the Pisharoty sonde performance during the entire diurnal cycle with respect to each of the imported sondes, ascents are classified into groups and the correlation with other sondes is found as follows:
The entire dataset of comparison ascents is split into three groups based on the imported sonde used (Vaisala, Meisei, or Graw).
Each of these groups is further divided into subgroups based on time of ascent (day, night, dawn, or dusk).
Scatterplots for different parameters are drawn for each of these subgroups and a best-fit straight line is also found. Obtained plots are similar to the combined scatterplot shown in Fig. 7. The total number of points considered, correlation, standard deviation, slope, and intercept of all parameters for the various subgroups are summarized in Table 2.
The major observations are as follows:
Temperature data are matched with the data obtained from imported sondes irrespective of the time of ascent. This is evident from the small standard deviations and a high correlation value of 0.999 (Table 2) across various subgroups.
Pressure data are also matched, especially at higher altitudes (within 0.5 hPa), where the accuracy required is higher. Scatterplot analysis also indicates a high correlation of Pisharoty sonde pressure with that of Vaisala, Meisei, and Graw.
It is uncertain whether the RH readings of the other radiosondes are accurate. For example, according to Vömel et al. (2007), Vaisala RS92 has a dry bias. This uncertainty is the main reason why no RH corrections are made, as stated in section 2a. Qualitative comparisons such as in Fig. 5 show that uncorrected Pisharoty sonde RH readings still measure small-scale RH variations similar to the other radiosonde in thin atmospheric layers. The statistical study shows that the sonde has good correlation with Vaisala and Graw (Table 2).
Wind data from the Pisharoty sonde is closely (within 1 m s−1) matched with the data obtained from Vaisala and Graw, as evident from Table 2. The correlation of Pisharoty wind data with that of Meisei is slightly lower when compared with that of other sondes. The large cluster of points along the diagonal in the scatterplot (Fig. 7) indicates close agreement between sondes.
In addition to the comparison ascents, a 1-month operational test was performed by launching Pisharoty sondes as operational soundings at the IMD Thiruvananthapuram Observatory (WMO station 43371) and transmitting WMO TEMP messages over the Global Telecommunications System (GTS) throughout February 2011. Preproduction PS01_B3 radiosondes were used, the same as in the flights listed in section 3b, but only preliminary solar radiation corrections were applied, derived from the December 2010 and January 2011 comparison flights. Since no WMO instrument code is assigned for the Pisharoty sonde, the 31313 group reported the five-digit system and status code as 59008, where 90 indicates an unspecified radiosonde.
4. Concluding remarks
ISRO-VSSC has designed and developed an advanced radiosonde system that provides a cost-effective solution for atmospheric parameter measurement. More than 7000 sondes were produced and delivered to different user agencies for atmospheric modeling, meteorological studies, and weather prediction.
Because of extremely low weight, the sondes can be used with smaller balloons (less hydrogen–helium gas), which eventually reduces the cost of the balloon ascent. Another version (PS02_A1) of the Pisharoty sonde was developed for wind-only applications, weighing around 75 g, which can be used to replace the optical-theodolite-based tracking of balloons.
The Pisharoty sonde receiver has data storage facility and is capable of an ac power supply or battery operation, and hence it can be used in remote locations. A handheld version of the receiver is also developed.
After balloon release and real-time data acquisition, already recorded data can be replayed, if required, in the software with another set of user-selectable parameters, say, smooth lengths. A potential research application of the system is the high-resolution boundary layer studies. In addition to the usual balloon ascents, time-staggered ascents, which are essential for the day-of-launch wind biasing scheme of launch vehicles, are also possible with the Pisharoty sonde system. In this scheme, data from sondes released every few minutes and captured by multiple receivers in the ground station allow for better atmospheric modeling by making simultaneous measurements of wind and atmospheric parameters at different altitudes.
This system has been validated by comparison ascents from different places, during different seasons of the year. The results have shown that the Pisharoty sonde is a reliable and efficient system, best suited for atmospheric studies, weather prediction, and other related aerospace applications.
We would like to mention our team members of Avionics at VSSC, who have shouldered the responsibility of design and development of the Pisharoty sonde system along with us: Gopakumar R., Femina Beegum S., Eden Evans Samuel K., Binil Roy T. S., Arun Alex, Mukundan K. K., Resmi R., and Satya Bhushan Shukla.
We are grateful to the anonymous reviewers, who spent their much valuable time reviewing and refining our paper. We thank the team at the National Atmospheric Research Laboratory (NARL) in Gadanki, IMD in New Delhi, the Space Physics Laboratory (SPL) of VSSC, the Meteorology Facility (METF) Thumba Equatorial Rocket Launching Station (TERLS) of VSSC, and the Kochi Naval Base for their wholehearted support in conducting the comparison ascents at these locations. We also thank our colleagues at Satish Dawan Space Centre (SDSC) in Sriharikota and various other centers of ISRO for their valuable advice and support in system realization and evaluation. Our sincere thanks are due to the director of VSSC for the inspiration and encouragement provided during the various phases of system development.