Data Analysis Software

Yukio Takahashi(takahashi(AT)nict.go.jp)

Communications Research Laboratory
4-2-1 Nukui-kita, Koganei, Tokyo 184-8795, Japan

1. Introduction

I introduce the VLBI software developed by CRL (Communications Research Laboratory). As the first generation, CRL had developed our own software system (K-3 software system) which is compatible with the Mark-III system. It had been used for international and domestic VLBI during 1983-1989. CRL developed a compact K-4 VLBI system which uses the K-4 recorder, and also develop the data processing software for the K-4 recorder and new data analysis software. This system is a second generation which was used during 1989-1997. Since 1993, CRL had developed the crustal deformation monitoring system for Tokyo metropolitan Area. We call it "Key Stone Project (KSP)". CRL developed the new software. This is the third generation which is available since 1995. Figure 1 shows the outline of VLBI data processing and analysis software. These software are referred to the Review of the Radio Research Laboratories Vol.30 in 1984 (Japanese), some papers [Takahashi et al., 1991; Takahashi et al., 1995] and the Review of CRL vol.42 in 1996 (Japanese).


Figure 1. Software developed by CRL.

2. K-3 software

The software system consists of 8 software programs, namely; scheduling software (KASER), data base setting software (KASET), correlation processing software (KROSS), bandwidth synthesis software (KOMB), tape format conversion software (KONV), a priori calculation software (KAPRI), parameter estimation software (KLEAR) and data base handling software (KASTL). The software systems were developed using HP1000-45F and 10L computer system. The characteristic of our system is to be used the same data base (K-3 data base) by all software from scheduling to analysis. The data base used the HP data base handler (IMAGE1000). The k-3 software system was developed from 1981 to 1984 according to the following considerations; Data are combined for common attributes, and the key word (key item) of the attribute is added. The data base is made up of the network of key items which are linked to detail values. The data for each item can be easily accepted at random by selecting the key item. The K-3 data base is useful for random access, that is, the reading and modification of individual items. All software can use the same values and data is transferred between software through the common data base. No inconsistencies exist for common values, such as physical constants, positions of sources and stations for all software. However, the sequential access for the K-3 data base is slower than that for the Mark-III data base. The K-3 data base requires a large region of the disk and any change in the structure would be laborious. The Mark-III data base is better in a minicomputer since access speed is a serious problem, but the conception of K-3 data base is better in a high performance computer and for the many modifications of the data. The data base setup software and data processing software must be developed by our own. A priori software had the tide model with the 484 periodical terms [Manabe et al., 1984].

3. K-4 software

CRL developed the "K-4" system using the K-4 recording system [Hama et al., 1991] and the analysis system for the new computer system HP1000 A900. For the precise compatibility, we adopted the Mark-III data analysis software, such as the software for the data base handler, the software (CALC) for the calculated delay and delay rate, and the parameter estimation software (SOLVE). Thus, they are easy to install in the new computer system and they are compatible with the worldwide data base used in geodetic VLBI. However, we had developed our own software to set up the data into the Mark-III data base. The new correlation processing software for K-4 system is controlled by the new computer system HP9000-330. The BASIC language is used since it can control the commands of GP-IB sequentially, and it is especially easy to check and modify. The correlation data send to other computer system HP1000 A900, and the bandwidth synthesis software runs in the minicomputer. A priori delay and rate are necessary to make a correlation processing for every parameter period (PP). Their delay and delay rate are obtained using the 0th deviation (), the 1st deviation (), the 2nd deviation (), and the 3 order derivation () with respect to time. In the old software, these values are the coefficients of Taylor expansion in the middle of the observation. The difference between the calculated delay and the true delay is larger at the beginning and end of each observation in according to be far from the middle of the center. In this software, the 0th, 1st, 2nd,3rd deviations are obtained using the approximation of the four order polynomial formulation. The entire duration of each observation is divided into 4 parts (-2 to -, - to 0, 0 to , to 2 while 4 is the complete duration). Furthermore the delays are calculated for the 5 points (-2,-,0, , 2). Calculated values for each observation are calculated for the least square fittings. The calculation method in the correlation processing software are based on CALC. For atmospheric delay, we corrected for height in the zenith path delay since some stations are located in elevated regions which are over 1000 m above sea level. This software can facilitate data processing both among K-3 (Mark-III) recorders, and among the new K-4 recorders, and also the correlation between K-3 recorders and K-4 recorders. In correlation between the K-3 and K-4 recorders, the K-4 recorder serves as the master for synchronization and it is not controlled. This control of the synchronization is the same as the control among the conventional K-3 systems. One command "COR" in the recording system is used to automatically synchronize K-4 system with other K-4 systems. Before fine automatic adjustment, rough tape synchronization to within 1 sec is achieved by using the position search command "PRL". Figure 2 and 3 show the correlation processing for each system combination.


Figure 2. The correlation processing among K4 recorders.


Figure 3. The correlation processing between K4 recorder and Mark-III recorder.

The data base setup software inputs various data for VLBI into the data base. We originally developed this software. The bandwidth synthesis software produces the files including the observation delay and delay rate for every observation and baseline. First, we collect the data from these files to produce a new observation data file. Secondly, we calculated the ionospheric delay corrections to input into the ionospheric correction data file. Thirdly, we calculate the ephemeris data using the JPL ephemeris "DE200/LE200" to input into the ephemeris data file. Fourthly, the temperature, the atmospheric pressure, the humidity at the site and the cable delay calibration are retrieved from the log file, and the interpolated data is calculated for each observation to input into the new calibration data file. Fifthly, we prepare for the precise earth rotation parameters, and we set up the earth rotation parameters into the earth rotation parameter file. Finally, we collect the all data and we set up them into the data base. We also developed the software for converting the recording data from K-4 tape to Mark-III tape for both high and low density. It is necessary for the case that we conducted the VLBI experiments using K-4 system and we must send the observation data to NASA. Furthermore, we developed the new parameter estimation software (VLBEST). This estimation method is the almost same as the estimation software (SOLVE) except for the QR solution. We can modify the estimated software easily, and it is used for KSP software.

4. KSP software

Since 1993, CRL had developed the KSP projects. In KSP, VLBI and SLR (Satellite Laser Ranging) facilities are established at four observation sites in the Tokyo metropolitan area and the precise position of four sites are measured every day or every other day [Takahashi et al., 1994; Kurihara et al., 1996]. The data of KSP is useful to research the crustal deformation and the dynamics in this area. The precision reaches less than 1mm and the repeatability is about 2mm for each experiment. The real time VLBI system is necessary to monitor the abnormal movement of the stations related with the big earthquakes. The station movements are measured in mm precision just after the experiment. The observation is conducted by a remote control and the data analysis is carried automatically [Koyama et al., 1996a; Koyama et al., 1996b]. We need the rapid results, and we realize the real time VLBI system in KSP [Kiuchi et al., 1996]. New data correlation software had developed corresponding to real time VLBI [Kondo et al., 1996; Sekido et al., 1996]. The system is completed and the regular experiments have been conducted every two days automatically without no operator. There is one operator at correlation center to check the system at the begin of the experiment and to monitor the system only during working time. The software is very advanced.

5. Other software

We proposed the new estimation using the differential method [Takahashi, 1992]. This method was applied to the clock estimation. It did not depend on judgments of analyst or need clock estimation. The estimation parameters in the geodetic VLBI are station positions, earth rotation parameters, atmospheric delay parameters, and the clock offset and clock rate of a frequency standard. The clock epoch is divided into the estimation period. The estimation parameter increases according to the clock parameters, the computer needs more time and a large work area. The clock information is unnecessary for the geodetic VLBI, and it is undesirable to include it in the estimated parameters. We propose an analytical method which uses the modified values without the clock information. This method can be applied to other estimated parameters such as atmospheric delay parameter. We assumed that the clock changes could be approximated to the linear change during the three observations. The modified delay value was calculated to cancel out the clock and clock rate using three sets of data as following formula;

where i is each observed delay, ti is the observation time, and i is observation number. We obtained the differential values for each observations. The number of data is decreases by only 2. However, the error matrix has the correlation between the observations. We must consider the effects of the correlation between the neighbor differential values. The detail is referred to the paper [Takahashi, 1992]. Our method is also suitable for the use of frequency standard whose stability is best for the 1000 seconds period.

6. Conclusion and Future

CRL have been developed the VLBI software. The correlation mixing the different VLBI system was realized at the first time. The software was also corresponding to the new K-4 VLBI system. The software in KSP may the most advanced system in the world. The many results was obtained by our software, such as domestic VLBI, Antarctica VLBI, Japan-China VLBI, KSP etc.

The future plan of the software is to develop the differential VLBI software for the new observation by the high speed switching, which NAO in Japan plans as VERA project. CRL has been developing the wideband VLBI system more than 1 Gbps. In the system, the number of channels is a few (one or four channels). We will develop the data processing and analysis software for the system. Furthermore, we want to develop the VLBI software available for both for astronomy and geodesy (astrometry).

References

Hama, S., H. Kiuchi, III.3 K-3 and K-4 VLBI Data Recorders, the special issue of the J. Commun. Res. Lab. ``Results of VLBI Experiments at the Communications Research Laboratory", 1991.

Kiuchi, H., M. Imae, T. Kondo, M. Sekido, S. Hama, T. Yamamoto, H. Uose, T. Hoshino,Real-time VLBI of the KSP, Proceeding of the Technical Workshop for ATP and APSG 1996.

Kondo, T., H. Kiuchi, M. Sekido, KSP Correlator and Data Reduction System, Proceeding of the Technical Workshop for ATP and APSG 1996.

Koyama, Y., Automated Remote Operation System and Data Analysis System for the Key Stone Project VLBI Network, Proceeding of the Technical Workshop for ATP and APSG 1996a.

Koyama, Y., Y. Takahashi, T. Gotou, K. Heki, Data Analysis Software, The Review of CRL, Vol.42, No.1, 1996b in Japanese

Kurihara, N., K. Uchida, F. Takahashi, M. Imae, T. Yoshino, S. Hama, Y. Takahashi, KSP Group of CRL, The Crustal Deformation Monitoring System for the Tokyo Metropolitan Area (Key Stone Project),Proceeding of the Technical Workshop for ATP and APSG 1996.

Manabe, S., Y. Takahashi, H. Hanada, T. Ishikawa, M. Fujishita, T. Sato, K. Koike and T. Yoshino, Physical Models Adopted in KAPRI, The Publication of the International Latitude Observatory of Mizusawa, Vol.XVIII, No2 pp 93-104, 1984.

Takahashi, F., M. Imae, T. Yoshino, Y. Takahashi, K. Heki, C. Miki, T. Kondo, Y. Kunimori, N. Kurihara, T. Otsubo, H. Takaba, T. Iwata, H. Kiuchi, Y. Koyama, Y. Hanado, M. Sekido, K. Imamura, J. Nakajima, A. Sugiura, The plan of Metropolitan Crustal Deformation Monitoring System by Communications Research Laboratory, Journal of Geodetic Society of Japan (CRCM'93 proceedings special isuue), 1994.

Takahashi, Y., S. Hama, T. Kondo, III.5 K-3 software system for VLBI and new correlation processing software for K-4 recording system, the special issue of the J. Commun. Res. Lab. ``Results of VLBI Experiments at the Communications Research Laboratory", 1991.

Takahashi, Y., The Baseline Analysis by the Differential Method in the Geodetic VLBI, Journal of the Geodetic Society of Japan, Vol.38, No.3, 1992.

Sekido, M., T. Kondo, H. Kiuchi, Y. Takahashi, Y. Koyama, Correlation Process Control Software, The Review of CRL, Vol.42, No.1 1996 in Japanese.

Updated on June 2, 1998. Return to CONTENTS