A dataset observed by solar radio spectrometer at 4.50 – 7.50 GHz (2002 – 2013)

No comments yet

Submit questions or advice:

You are not logged in, please[Login]or[ Register]!

A dataset observed by solar radio spectrometer at 4.50 – 7.50 GHz (2002 – 2013)

Total number of views and downloads

View in HTML Paper download
995 6

A dataset observed by solar radio spectrometer at 4.50 – 7.50 GHz (2002 – 2013)

The author's papers

Sorry, failed to retrieve the author's related papers.

            Data source: Chinese Science Citation Database(CSCD)

A dataset observed by solar radio spectrometer at 4.50 – 7.50 GHz (2002 – 2013)

Yang Zherui1, Ning Zongjun1, Lu Lei1, Meng Xuan1, Gao Na1, Liu Liang1*

1Purple Mountain Observatory, Chinese Academy of Sciences, Nanjing 210008, P. R. China

*Email: LiangLiu@pmo.ac.cn

Abstract: The sun is the nearest star to  Earth. For long, solar observations have been an important component of fundamental astronomy researches. Earlier researches show that the sun has its own atmosphere which has been almost entirely ionized to plasma due to its high temperature. Changes of its built-in magnetic fields could result in spontaneous and complex solar eruption phenomena, such as the solar flare and the coronal mass ejections(CMEs). Observations on these eruptions and understandings of their physical processes have been the frontiers of solar physics. Since solar eruption is often accompanied by significant changes in the radiation intensities of different frequency bands, multi-band observations are very important for the study of solar eruption phenomena. The spectrometer at Purple Mountain Observatory could carry out multi-channel solar observations in radio frequency band. Its working frequency ranges from 4.5 GHz to 7.5 GHz, which was divided into 300 channels. It has a temporal resolution of 5 ms and its sampling rate reaches 10 M/s. The spectrometer has observed more than 200 solar flares, and our online database stores the

data observed from 2002 to 2015.

Abstract: solar radio; multi-channel; frequency spectrogram; IDL; FITS

Database Profile:

Chinese title

4.5~7.5 GHz太阳射电频谱观测数据集(2002~2013年)

English title

A dataset observed by solar radio spectrometer at 4.50 – 7.50GHz (2002 – 2013)

Corresponding author

Liu Liang (Liangliu@pmo.ac.cn)

Data authors

Yang Zherui, Ning Zongjun, Lu Lei, Meng Xuan, Gao Na, Liu Liang

Time range

2002 – 2007,2010 – 2013

Observation target

Solar atmospheric radiation

Data format


Data amount

40,000 MB

Data service system

<http://www.pmo.csdb.cn/index1.php>; <http://www.sciencedb.cn/dataSet/handle/24>

Sources of funding

National Natural Science Foundation of China (2014, Grant No.11333009);

National Natural Science Foundation of China (2015, Grant No. 11573072)

Dataset composition

The dataset contains solar radio spectral data observed during 2002 – 2007 and 2010 – 2013. The data of every day are in FITS format, recorded every 10 seconds during the observation.



Solar radio emissions are the touchstone of the fundamental physical radiation theory. These emissions are mainly from plasma radiation, synchrotron radiation and MASER radiation, attributable to high-energy electrons accelerated by solar eruption events when moving in the plasma with a strong build-in magnetic field. Therefore, observations and studies on solar radio emission are one of the basic ways to understand the acceleration process of solar flare.

The solar radio telescope (spectrometer) at Purple Mountain Observation (PMO) is a professional instrument used to record the solar radio emissions generated during solar activities since the late 1990s. The instrument samples every 10 MHz from 4.5 GHz to 7.5 GHz, amounting to 300 sampling channels in total. Each sampling case takes 5 minutes. The instrument archived data during 2002 – 2007 and 2010 – 2013, The period of 2008 – 2009 was not archived since the telescope was in technical problems during this time period. These archived data usually need to be calibrated into the units of standard solar flux before applying to research1. The fine structures distinguished from the dynamic spectrum are closely related to the solar flares, which are very important for us to study the relevant emission mechanism, energy release and particle acceleration processes2 – 7.

2.Data collection and processing

Located at the Purple Mountain Observation Station, the solar radio telescope is
used to monitor solar radio emissions during daytime. Observation  is performed from 01:00 to 08:00 UT (Beijing Time: 09:00 – 18:00). The input voltage of this instrument  ranges from -1 to 0 V and it has 300 channels during 4.5 – 7.5 GHz, with 10 MHz per channel. The observation records variations of the solar flux over the time period for the 300 frequency channels. Data sampling is conducted at a precision of 1 ms. It uses 5-group-time-division-multiplexing method to collect data of 300 channels and there are 60 frequency channels in every group. So data could be collected from each channel every 5 ms. The instrument uses the second pulse provided by GPS as the criterion for data acquisition. The error of M-GPS second pulse is less than 0.1us. Collection starts as soon as the second pulse occurs. And we use 50-MHz crystal oscillator as the collection criterion to ensure the accuracy of the raw data. Data for all channels are recorded every 5 ms. After being transferred through the antennas and the receiver, solar radio signals are converted into electrical signals for storage.

The recorded solar radio spectral data are stored in files with a .num extension. The file name is a set of digital numbers in the format of “ddhhmmss”, which stands for the day, hour, minute and second when the first observation starts.  One .num file is generated every 10 s, and its first 64 bytes make up the header, followed by the intensity values with 2 bytes per channel for every sampling circle.

Information of each file is contained in the file header, as shown below:

The raw data were stored in disks in its original .num format, which are then moved to the PMO research laboratory using mobile storage for further processing. To maximize data readability, the .num files are converted into standard .fits files using an IDL routine “NUM_to_fits.pro”8. The source code can be found at: <http://www.pmo.csdb.cn/spectrum/program/NUM_to_fits.pro>.

3.Sample Description

  The name of the file which stores the solar radio data consists of 8 digital numbers in the format of “ddhhmmss”, designating the day, hour, minute and second of the observation, with .fits extension. The file is generated every 10 s during the observation and it consists of a header file and the radio data. The header file records parameters of the observation, followed by intensity values for the 300 frequency channels.

Take for example the file “25123549.fits” obtained on April 25, 2005. The data record solar radio intensities for the 300 frequency channels from 4.5 to 7.5 GHz (10 MHz per channel) observed every 5 ms from 12:35:39 to 12:35:49. The file header is shown in Table 1.

Table 1  The header file of 125123549.fits 

The middle column shows the parameter names. The right column shows the corresponding parameter values. From the header file we know that the .fits file was generated at 12:35:39 on November 23, 2013 (local time), and recording began at 12:35:39 on April 25, 2005. The spectral data is a two-dimensional float array with first dimension being 2,000 and second dimension being 300.  It was observed at the solar radio observation station of PMO; the minimum radio intensity is 57, and the maximum intensity is 4,095; the observation frequency range is 4.5 – 7.5 GHz, the frequency resolution is 10 MHz and its temporal resolution is 5 ms.

The recorded solar radio data are a two-dimensional floats array with 300 rows and 2,000 columns. 300 rows stand for 300 frequency channels and 2,000 columns represent 2,000 sampling cycles within 10 s. These data are in AU(Arbitraty Units). For the massive volume, we only list data of the first 50 columns at the 100th row here. The IDL command is:

 —>  print,data[0:49,99]

 The output :

The output indicates that the radio spectral data at the 100th channel (with a frequency of 4.95 GHz) on April 25, 2005 from 12:35:39 to 12:35:49 (with an interval of 5 ms) are 346, 338, 342 … 339.

4.Quality control and assessment

The solar radio spectral data include spectrum intensities of the 300 channels. Some channels may be invalid due to channel malfunction or environmental interference. To overcome this, we first select the radio intensities for each frequency channel at certain observing time t0. Suppose the intensity values of the 300 channels at the point are x1, x2, x3, …, x300, and we can calculate their average value x and their standard deviation value s.

Then we calculate the absolute values of the difference between x1, x2, x3 …, x300 and x one by one and suppose the results are d1, d2, d3 …, d300.  Compare them with 3 s, if dn > 3 s, then the nth channel at t0 is invalid. The average value of the adjacent channels would be considered as the value of the invalid channel at t0. The number of invalid channels is generally about 15 to 30, which is about 5% – 10% of the total channels.

5.Value of data

Data observed by solar radio spectrometer at 4.50 – 7.50 GHz  present the curves showing all-round changes of solar radio in different channels at various times. They could help study the microwave radiative fine structure of solar flares and analyse the physical processes behind them. Since the solar radio telescope at PMO was put into use in April 1999, the observation data has been used to discover and analyse a lot of solar radio bursts and some specific spectral fine structures within this frequency band9 12.

The dataset was produced at a frequency band of 4.5 – 7.5 GHz, which have a high temporal resolution of 5 ms and a high spectral resolution of 10 MHz. In comparison, the solar radio spectral data at Yunnan Observatories have a frequency band of 0.6 GHz – 1.2 GHz. The spectrometer of National Astronomical Observatories works at a frequency band of 5.2 – 7.6 GHz and has a temporal resolution of 5 ms, but it only have a spectral resolution of 20 MHz.

We would take some examples to compare our observations with observations from other instruments. Observations for the same event could be used to check and demonstrate the characteristics of different telescopes. Figure 1 shows observations of solar flare on March 18, 2003 by different instruments. The 4th panel is the light curve observed by the PMO spectrometer at 4.5 GHz . The abscissa values represent the time and the ordinate values represent the radio radiation intensity. The PMO spectrometer has observed a large number of temporal fine structures within the frequency band of 4.5 – 7.5 GHz due to its high temporal resolution and spectral resolution. The bottom panel shows observations with the abscissa representing the time and the ordinate representing the frequency (channel). The brightness of the dynamic spectrum indicates the radio intensity recorded by different frequency channels at different times. The whole fine structure lasted for 6 seconds and 4 pulsating drift structures were distinguished, as indicated by dashed lines. Similar to the structures at low frequencies of its kind, it consisted of many short time scale and narrow-band pulse structures. The intensity and duration of these pulse structures show a power-law distribution, which is similar to the distribution of solar flares. For the first time, we observed the pulse structures which drifted to high frequencies13, which is contrary to previous observations that pulse structures drift to low frequencies. This flare was also observed by other space-based or ground-based telescopes. The top two panels show the soft and hard X-ray light-curves observed by GOES and RHESSI14, respectively, and the 3rd panel shows the radio observation results by Nobeyama Polarimeter15. From all these observations, we find the light-curve at 3.75 GHz observed by Nobeyama Polarimeter is consistent with results observed by the PMO radio spectrometer. A flux peak was captured by both instruments. From observations by the PMO spectrometer we can distinguish many fine structures within the peak due to its high temporal resolution while from Nobeyama Polarimeter only an eruption peak was observed. This illustrates the advantages of high temporal and spectral resolution of the PMO spectrometer. Since these high-speed, narrow-band eruptive phenomena are related to the processes of particle acceleration (especially electron acceleration) in solar flares, the study of these fine structures provides first-hand material for understanding the acceleration process of electrons in solar flares.

Figure 1 Overview of a solar flare detected by some major spectrometers on March 18, 2003

6.Usage notes

We make use of solar radio spectral data by processing them into dynamic spectrum figures  for the study of solar activities. We use the IDL program “readfits.pro”16 to read the data from .fits files. The source code can be found at: <http://www.pmo.csdb.cn/program/readfits.pro>. For example, the IDL command for reading data from “25123539.fits” is:

—> data=readfits(‘/home/yangzr/20050825/25123539.fits',h)”

After getting the data, we can apply IDL routines to make different kinds of figures. We could plot the light-curve for one frequency channel or we can plot the spectrum at a fixed time point. The span between two adjacent points along the temporal axis is 5 ms, equal to the interval of data sampling over 300 frequency channels. The dynamic spectrum of high temporal resolution is able to show detailed spectrum variations occurring within a small time range, which is helpful to study the temporal fine structures of microwave radiation evoked by solar flares.

We use IDL routines to show the data stored in “25123549.fits”. For example, in Figure 2, we get the flux variation of the 201th frequency channel (6.5 GHz frequency); the abscissa values represent the recording time with a time cadence of 5 ms, while the ordinate values represent the corresponding radio radiation intensity.

The IDL command to plot such a figure is:


Figure 2  Radio intensity at the 200th channel (6.5 GHz bandwidth) on March 25, 2005 from 12:35:49 to 12:35:49

We can also plot the whole dynamic spectrum to get radio intensities of all the channels over corresponding time periods of observation (Figure 3). The abscissa values denote the time for each data sampling circle, while the ordinate values represent the 300 frequency channels. The brightness of  spectrum indicates the radio radiation intensity.

The IDL command is:

 —-> plot_image,data

Figure 3  Spectral image showing intensities of all the channels (4.5 7.5 GHz bandwidth) on March 25, 2005 from 12:35:49 to 12:35:59

Since one .fits file could only record the spectral data for 10 s, many events which last a longer time could not be recorded completely. In order to study such events, we need to connect the .fits files. For example, da1, da2… are the data of 10 s,  and the  command to connect them together is:


After connecting the data, we can use the IDL command “plot, daydata[*,x]” to show the variation of flux intensity at frequency channel x during the connected time period. Using the command “plot_image,daydata” we can show the flux variation over all the channels during the connected time period. The upper-left panel of Figure 4 shows radio intensities of all the channels on April 25, 2005 at 4:36:51.5. The upper-right panel shows the dynamic spectrum for the whole 300 frequency channels during the entire connected time, and the lower-right panel shows the radio flux variation at 6.31 GHz during the connected time.  

Figure 4 Spectral image of the whole observation time (UT) on March 25, 2005      

In cases when we do not want the data to be shown with such a high temporal resolution, or we just want to see the outline of the data, we can reduce the temporal resolution by integrating the radio data over a certain time period. For example, we reduce every 40 adjacent time points to one point by taking the average value to get a temporal resolution of 0.2 s, and reduce 200 points to one point to get a temporal resolution of 1 s. The temporal-resolution-reduced data can be displayed much faster under a smaller size.


This study was supported by Purple Mountain Observatory and Computer Network Information Center, CAS.


1. Lu L, Liu S, Song Q et al. Calibration of solar radio spectrometer of the Purple Mountain Observatory. Acta Astronomica Sinica 56 (2015): 130 – 144.

2. Tan C, Tan B, Yan Y et al. Microwave observations of the Chinese Solar Broadband Radio Spectrometer at Huairou. Proceedings of the International Astronomical Union 8 (2012): 499 – 500.

3. Ning Z, Fu Q & Lu Q. Solar radio radiation bursts on 1998 April 15. Publications of the Astronomical Society of Japan  52 (2000): 919 – 924.

4. Ning Z, Fu Q & Lu Q. Special fine structures of solar radio bursts on April 15 1998. Astronomy and Astrophysics 364 (2000): 853 – 858.

5. Chernov GP, Yan YH & Fu QJ. A superfine structure in solar microwave bursts. Astronomy and Astrophysics 406 (2003): 1071 – 1082.

6. Wang M, Fu Q, Xie R et al. Centimetric type N and type M bursts. Astronomy and Astrophysics 380 (2001): 318 – 322.

7. Yan Y, Fu Q, Liu Y et al. Fine structures observed by the Chinese Solar Radio Broadband Fast Dynamic Spectrometers. The Proceedings of the IAU 8th Asian-Pacific Regional Meeting  289 (2003): 401 – 408.

8. Ning Z. NUM_to_fits.pro, 2014.  Available at: <http://www.pmo.csdb.cn/program/NUM_to_fits.pro>.

9. Xu FY, Xu ZC, Huang GI et al. A broadband solar radio spectrometer and some new observational results. Solar Physics 216 (2003): 273 – 284.

10. Ning Z, Yan Y, Fu Q et al. Microwave M burst on May 3 1999. Astronomy and Astrophysics 364 (2000): 793 – 798.

11. Ning Z, Wu H, Xu F et al. Positively drifting structures during the 18 March 2003 solar flare. Solar Physics 241 (2007): 77 – 84.

12. Ning Z, Wu H, Xu F et al. High-frequency evolving emission lines for the 25 August 1999 solar flare. Solar Physics 250 (2008): 107 – 113.

13. Ning Z, Wu H, Xu F et al. Positively drifting structures during the 18 March 2003 solar flare. Solar Physics 241 (2007): 77 – 84.

14. Lin RP, Dennis BR, Hurford GJ et al. The Reuven Ramaty high-energy solar spectroscopic imager (RHESSI), in The Reuven Ramaty High-Energy Solar Spectroscopic Imager (RHESSI), 3 – 32, eds. Lin RP et al. Amsterdam: Springer Netherlands, 2003.

15. Torii C, Tsukiji Y, Kobayashi S et al. Full-automatic radiopolarmeters for solar patrol at microwave frequencies. Proceedings of the Research Institute of Atmospherics, Nagoya University 26 (1979): 129 – 132.

16. Lu L. IDL program readfits.pro, 2013. Available at: <http://www.pmo.csdb.cn/program/readfits.pro>.

Data citation

Yang Z, Ning Z, Lu L et al. A dataset observed by solar radio spectrometer at 4.50 – 7.50GHz (2002 – 2013). Science Data Bank. DOI: 10.11922/sciencedb.590.24

Authors and contributions

Yang Zherui, Assistant Engineer; research area: computer science. Contribution: Web development and database design.

Ning Zongjun, Professor; research area: astrophysics. Contribution: program design.

Lu Lei, PhD Candidate; research area: solar physics. Contribution: data processing.

Meng Xuan, Engineer; research area: solar physics. Contribution: data scanning and retrieval.

Gao Na, Engineer; research area: computer science. Contribution: project applying.

Liu Liang, Professor; research area: computer science. Contribution: project organizing.


How to cite this article: Yang Z, Ning Z, Lu L et al. A dataset observed by solar radio spectrometer at 4.50 – 7.50 GHz (2002 – 2013). China Scientific Data 2 (2016), DOI: 10.11922/csdata.590.2015.0020