1. A radar data simulator for SuperDARN A.J. Ribeiro, P.V. Ponomarenko, J.M. Ruohoniemi, J.B.H. Baker, L.B.N. Clausen, R.A. Greenwald ------ 5/29/2012 SuperDARN Workshop 2012

2. Background In order to test the performance of SuperDARN processing routines (e.g. FITACF) we need to know what the correct output is This can be accomplished by using simulated radar data, where the inputs are known At the Dartmouth workshop in 2011, the Software Working Group decided that a data simulator would be useful Ponomarenko et al. [2008] developed a physically based, robust system for simulating data We have adapted this concept, and made it more flexible and user-friendly 29 May 2012 SD Workshop 2012 1

3. Components of the Simulator The simulator can be though of as consisting of four fundamental components – A scatter model Fundamental components – An irregularity model One per range gate Made up of collections of scatterers with common characteristics – A model radar Samples the irregularities using SuperDARN pulse sequences – Post-processing Some data manipulation to make the simulator more flexible and the data more realistic 29 May 2012 SD Workshop 2012 2

4. The Big Picture 29 May 2012 SD Workshop 2012 3 irregularities Range gates scatterers radar Post-Processing

5. Scatterer Model Fundamental components of the simulator Perfectly reflecting point in 1-D space with a discrete lifetime Each has a random appearance time within the sampling interval, t app Each has a random ‘off’ time to model particle disappearance, t off Each has a Gaussian random velocity component, v G – This velocity gets added to the bulk velocity of the irregularity 29 May 2012 SD Workshop 2012 4 t app time t off amplitude of reflected signal

6. Irregularity Model One Irregularity per range gate Composed of 2000 discrete scatterers Each scatter is assigned a random spatial location within the 1-D range gate The irregularities have a bulk LOS velocity component, v LOS Each irregularity has a characteristic growth and decay time constant, t g and t d 29 May 2012 SD Workshop 2012 5 time amplitude of reflected signal decaygrowth

7. Model Radar Responsible for sampling the irregularities In general, any arbitrary pulse sequence can be used, but katscan, tauscan, and old normalscan are available by default Sampling is performed in series Starting from the time of each pulse, calculate returned samples, one range gate at a time Voltage samples are calculated by summing the returns from individual scatterers Amplitude and phase of returned signal are calculated separately 29 May 2012 SD Workshop 2012 6

8. Model Radar - Calculating Amplitude 29 May 2012 SD Workshop 2012 7 t app time t off amplitude of reflected signal decay growth

9. Model Radar - Calculating Phase 29 May 2012 SD Workshop 2012 8

10. Model Radar - Calculating Voltage 29 May 2012 SD Workshop 2012 9

11. Post-Processing Integrated ACFs have lag 0 powers which fluctuate about some random average value – This value is arbitrary Each of the integrated ACFs is normalized to have an average lag 0 power of 1 – This allows scaling of ACF amplitude to some user-selected value If desired, ACF amplitude can be made to decay with range (1/r 2 ) If desired, δ-correlated white noise can be added to the signal – This is done by adding noise ACFs to the signal ACFs – Noise ACFs are generated by sampling irregularities with a decorrelation time shorter than the smallest interpulse separation, mpinc 29 May 2012 SD Workshop 2012 10

12. Results - Output The output of the simulator can be either ACFs in RAWACF format or voltages in IQDAT format The files can be manipulated using any standard SuperDARN routines 29 May 2012 SD Workshop 2012 11 (a)A real SuperDARN ACF. This is from a katscan period, N ave = 21. The fitted parameters for this data are: t d = 55 ms, v LOS = 365 m/s, and SNR = 9 dB. (b)An example of a simulated ACF. This ACF was generated with katscan, N ave = 21, t d = 50 ms, v LOS = 350 m/s, and SNR = 9 dB.

13. Results – Data Quality Because of the random nature of the simulator, realistic statistical fluctuations are present in the data 29 May 2012 SD Workshop 2012 12

14. Results – Data Quality 29 May 2012 SD Workshop 2012 13

15. Inputs to the Simulator 29 May 2012 SD Workshop 2012 14

16. What can we do with it? Use simulated data to test new SuperDARN routines – E.g. effectiveness of ACF fitting algorithms (See my poster) Test the effect of Gaussian velocity spread versus irregularity decay on the shape of ACFs Other things which we haven’t thought of 29 May 2012 SD Workshop 2012 15

17. Conclusions We have developed a data simulator which produces reliable and realistic data Data is available in both IQDAT and RAWACF format Hopefully people in the community find uses for the simulator 29 May 2012 SD Workshop 2012 16

18. References Ponomarenko, P. V., C. L. Waters, and F. W. Menk (2008), Effects of mixed scatter on SuperDARN convection maps, Ann. Geophys., 26, 15171523, doi:10.5194/angeo-231 26-1517-2008. 29 May 2012 SD Workshop 2012 17

19. Thank You. Questions? 29 May 2012 SD Workshop 2012 18