1. Ørsted DTU 1 The COSMOS Airborne Campaigns. Status October’06 N. Skou, S. S. Søbjærg, J. Balling, S. S. Kristensen, and S. Misra ØrstedDTU Technical University of Denmark ns@oersted.dtu.dk

2. Ørsted DTU 2 L-band Radiometer System EMIRAD-2 is a fully polarimetric radiometer operating in the 1400 - 1427 MHz protected band EMIRAD-2 consists of: –2 antennas, one pointing 40 deg aft, one pointing nadir. The antennas are Potter horns with no sidelobes –radiometer unit with dual inputs –EGI (INU + GPS) for attitude and navigation –industrial PC for fast data recording –laptop for instrument control and normal data recording Installed on 2 small aircraft

3. Ørsted DTU 3 40 deg Potter Horn

4. Ørsted DTU 4 40 deg Horn Pattern HPBW=30.6° i.e.: FPL = 932 m FPX = 714 m from 1000 m altitude

5. Ørsted DTU 5 Radiometer Description Digital radiometer with subharmonic sampling. A to D converters directly sample the L-band signals with a clock frequency of 139.4 MHz. The data from the converters are fed into an FPGA where correlation, calculation of second and fourth order moments of the PDF, and integration is performed digitally Data integrated to 8 msec. is stored on the laptop computer also controlling the system. These data will be available in near real time. A second data stream - fast data - is implemented for RFI mitigation, done off-line for optimum performance. In the normal mode of operation, data only pre-integrated to 1.8  sec is recorded on a fast HD in an industrial PC. The fast data channel can also be operated in a special mode where raw data from the converters are stored. 2 x 32 K samples are stored with a 25% duty cycle. The normal fast data is pre-integrated to 14.7  sec in this mode.

6. Ørsted DTU 6 Data Output 8 msec. integration: – for H-pol – for V-pol – for H-pol – for V-pol – 0° for 3’rd Stokes – 90° for 4’th Stokes Fast data (1.8  sec integration): as above. Fast data alternatively raw samples plus above integrated to 14.7  sec.

7. Ørsted DTU 7 EMIRAD-2 Specifications Correlation radiometer with direct sampling Fully polarimetric (i.e. 4 Stokes) Frequency: 1400 - 1427 MHz (-60 dB BW; about 22 MHz -3 dB BW) Digital radiometer with 139.4 MHz sampling Advanced analog filter for RFI suppression. Data integrated to 8 msec recorded on PC Off-line digital RFI filtering in frequency and time domains. Fast data pre-integrated to 1.8  sec or raw data is recorded on HD Sensitivity: 0.1 K for 1 sec. integration time Calibration: internal load and noise diode 2 antennas - one nadir pointing, one pointing 40 deg. aft Antennas are Potter horns (no sidelobes) with 37.6° and 30.6° HPBW

8. Ørsted DTU 8 Block Diagram

9. Ørsted DTU 9 Temperature Stabilized Enclosure 2 digital PI-regulators stability of microwave section better than 0.02 °C for 15 °C change in ambient temperature DFE stability better than 0.1 °C for same change

10. Ørsted DTU 10 Problems under Warm Conditions Internal temperature (normally around 40 °C) cannot be kept stable Calibration severely affected Eventually the radiometer overheats This situation prevailed in Australia due to failing aircraft air-condition Solution: –base calibration on internal load and noise diode –make model for noise diode output as function of temperature by operating radiometer in lab under elevated temperatures –re-process all data Result: –calibrated data but with less accuracy (under normal conditions calibration depends directly on primary LN2 cal. - here only indirectly) –OK for soil moisture where requirements are modest

11. Ørsted DTU 11 Radiometer Control - screen dump

12. Ørsted DTU 12 Large Antenna on C-130

13. Ørsted DTU 13 EMIRAD-2 on Aero Commander

14. Ørsted DTU 14 EMIRAD-2 on Aero Commander

15. Ørsted DTU 15 EMIRAD-2 on Aero Commander

16. Ørsted DTU 16 CoSMOS “Down Under” Campaign

17. Ørsted DTU 17 EMIRAD on HUT Skyvan

18. Ørsted DTU 18 Two Flight Patterns off Norway

19. Ørsted DTU 19 CoSMOS-OS Campaign

20. Ørsted DTU 20 Re-processing Status Radiometer has been characterized in the lab with internal temperatures in the range 37 - 48 °C OMTs have been measured (NWA): –loss in side port: 0.08 dB, in end port: 0.05 dB –S 11 around -15 dB –X-pol below 30 dB Cable losses are 0.3 dB Processing algorithms established this week Bulk processing starts next week Quality checks Data delivery before Christmas

21. Ørsted DTU 21 RFI - 8 msec Data and 15  sec Data

22. Ørsted DTU 22 RFI - 15  sec Data: TB and Moment Ratio

23. Ørsted DTU 23 Example from Australia (zoom in on 15  sec data)

24. Ørsted DTU 24 Example from Australia (raw data)

25. Ørsted DTU 25 What are the Dangers of RFI? Strong RFI of long duration may elevate brightness temperature by unreasonable amount or even blank radiometer Result: loss of data, but you know! More likely, but far more dangerous situation: RFI (low level or short pulses) may contribute to your signal with a power corresponding to a Kelvin for example. Very difficult to know!

26. Ørsted DTU 26 What are our Priorities? 1.To detect the situation 2.Mitigate if possible

27. Ørsted DTU 27 Can RFI be Detected? Huge RFI no problem - TB is clearly too large More normal RFI very difficult in normal radiometer systems having integration from milliseconds to seconds Need to have very fast sampling rate - preferably digital radiometer (EMIRAD-2 has 140 MHz sampling) Radar signals may be detected as unusually large signals with suitable but short pre-integration But most signals can be detected by investigating statistical properties: –TB is Gaussian which has a fixed ratio of 3 between 4’th and 2’nd order central moments (kurtosis) –Other signals (especially pulsed and continuous) typically have different value (beware, however: sine with 50% duty cycle also have ratio of 3!!) All this has to be done using “raw” data (before integration)

28. Ørsted DTU 28 Can RFI be Mitigated? Time domain: –Continuous signals not –Pulse type signals with low duty cycle can: following the detection in the raw data, inflicted samples are discarded before integration. For typical radar signals the loss of radiometer signal will thus be very moderate. Frequency domain: –Even our narrow band (27MHz) may be split into sub-bands. –Each sub-band is analyzed –Inflicted sub-bands are discarded –Work on both continuous and pulsed signals In both cases: consequence is increased  T - depending on how much has to be discarded.

29. Ørsted DTU 29 What is Being Done at DTU? EMIRAD-2 has collected data in Australia and in the North Sea: –Data pre-integrated to 1.8 microsec. recorded continuously. –Bursts of raw data (140 MHz sampling) recorded on special occasions. For sure, examples of RFI have been captured Analysis and theoretical considerations are ongoing.

30. Ørsted DTU 30 CoSMOS-OS Flight Line

31. Ørsted DTU 31 Power, Aft Horn, H-pol, 8 msec. Sampling

32. Ørsted DTU 32 Kurtosis, Aft Horn, H-pol, 8 msec. Sampling

33. Ørsted DTU 33 Power, Region of Interest

34. Ørsted DTU 34 Kurtosis, Region of Interest

35. Ørsted DTU 35 Power, Zoom in on Region of Interest

36. Ørsted DTU 36 Kurtosis, Zoom in on Region of Interest

37. Ørsted DTU 37 Power, one 8 msec Window with 1.8  s Sampling

38. Ørsted DTU 38 Kurtosis, one 8 msec Window with 1.8  s Sampling

39. Ørsted DTU 39 Power, Nadir Horn, 8 msec. Sampling

40. Ørsted DTU 40 Kurtosis, Nadir Horn, 8 msec. Sampling

41. Ørsted DTU 41 Power, Zoom in on Region of Interest

42. Ørsted DTU 42 Kurtosis, Zoom in on Region of Interest

43. Ørsted DTU 43 Power, one 8 msec Window with 1.8  s Sampling

44. Ørsted DTU 44 Kurtosis, one 8 msec Window with 1.8  s Sampling

45. Ørsted DTU 45 Kurtosis on Flight Line

46. Ørsted DTU 46 Power

47. Ørsted DTU 47 Power - Land Zoom

48. Ørsted DTU 48 Power - Sea Zoom

49. Ørsted DTU 49 Kurtosis

50. Ørsted DTU 50 Kurtosis - Land Zoom

51. Ørsted DTU 51 Kurtosis - Sea Zoom

52. Ørsted DTU 52 Conclusions and Plans Many potentially harmful RFI pulses present Some can be seen in 8 msec TB data - some cannot!! Can be seen in kurtosis data Some not seen in 8 msec TB data can be seen in 1.8  s TB data. Example 1: seen both in kurtosis and 8 msec TB data. RFI adds 3.5 K!! Example 2: only seen in kurtosis. RFI adds 0.9 K to 8 msec TB data!!!!!! Kurtosis powerful tool. Also fast sampling - 1.8  s (TBC) - is powerful Analysis being carried out this fall and winter Algorithms for mitigation to be developed Potential for FPGA implementation - avoids fast data rate (link to ground / recording capacity)

53. Ørsted DTU 53 What About Future Campaigns? Do not trust any L-band radiometer unless it has at least fast sampling (microsecond range) - and preferably kurtosis check This goes especially for airborne campaigns Available radiometers: –EMIRAD-2 –CAROLS (EMIRAD-2 copy, ready next year) –EMIRAD-1 after minor modification. –????