Minutes of the MADONA Data Exchange Meeting

Hosted by GMGO, Trarben-Trarbach, Germany 15 - 17 June 1993

Present:

• Dr Bl`dorn, Host President of GMGO, Germany
• Dr David Ride, Chairman CBDE, UK
• Mr Ron Cionco, Chairman ARL, USA
• Mr Tim Higgs, Secretary CBDE, UK
• Mr Welf aufm Kampe GMGO, Germany
• Mr Harald Weber GMGO, Germany
• Mr Kenneth NyrJn FOA, Sweden
• Mr Per-Erik Johansson FOA, Sweden
• Dr Reinier Sterkenburg PML-TNO, Holland
• Mr Jerry Jensen JAYCOR, USA
• Dr Ray Kamada Kamada Science & Design, USA
• Mr Jurgen Streicher DLR NE-OE, Germany
• Dr Torben Mikklesen RISØ, Denmark
• Dr Søren Thykier-Nielsen RISØ, Denmark
• Dr Hans Jørgensen RISØ, Denmark
• Mr Josep Moreno-Santabarbara RISØ, Denmark

1. Apologies were received from:

• Mr C Biltoft (Dugway Proving Ground, USA),
• Dr ART Hin (TNO, Netherlands)
• Dr C D Jones (CBDE, UK).

2.Dr Bl`dorn, President of the German Military Geophysical Office (GMGO), opened the Data Exchange Meeting by welcoming the attendees to GMGO. He stressed the importance of the results to be gained from the MADONA experiment as being very relevant in this turbulent political climate, not only to be used for NBC defence but also in accident & disaster planning and Meteorological model development.

3.The administrative, transportation and social details were dealt with quickly. The provisional agenda was modified to allow CBDE to present Dr M Bennett's data from the scanning LIDAR (RASCAL), and for Søren Thykier-Nielsen and Jose Moreno-Santabarbara from RISØ to show diffusion and meteorological simulations respectively.

4. Mr T Higgs from CBDE was proposed and accepted as the meeting secretary.

SESSION ONE

Meteorological Data Chairman - Mr R Cionco

5. A brief summary of MADONA was given by the chairman and a showing of the CBDE 10 minute MADONA video followed.

6. MADONA became a larger international collaborative project than originally planned owing to forced delays brought on by operation GRANBY. Sweden and Denmark, with personnel from the Netherlands and Spain on detachment to CBDE and RISØ respectively, were able to join the UK (CBDE and Met Office), USA (ASL and DPG) and Germany (GMGO, DLR and WWDBw) in providing the largest assembly of high resolution meteorological and sensor equipment for this type of experiment. A total of 16 hours of continuous smoke and 99 discrete puffs were generated over the CBDE range.

7. The audience were impressed by the content and felt it presented MADONA accurately. A copy of the video was issued by CBDE to each participating country in PAL or NTSC format as appropriate. All expressed high regard for the work done by CBDE Photographic Section for their high-quality low-light work during MADONA and in all, 3-4 hours of video were distilled into the 10 minute summary video presented.

Synoptic weather and Upper air data - Meteorological Office UK

8. The data available were described by David Ride. They were provided by John Kimber of the Met Office team with CBDEs on-site forecasting unit. On floppy disk were the following:- Boundary layer sonde (2s intervals) of pressure vs ht -- up to 500mb Doppler sodar of Temperature and RH vs ht -- up to .10000m

9. There is also a compilation of 1 minute readings on paper recorded at the Met Building at CBDE, and synoptic met maps photocopied from routine facsimile broadcast.

10. Reinier Sterkenberg standing in for ART Hin, representing TNO, asked if the forecast winds relating to those actually measured were available. Discussion followed and CBDE were actioned to seek the Chemical Downwind Message information from Def Met O -- Sept 93.

11. Torben Mikklesen (RISØ) commented that it would be useful to input forecast data into the dispersion models and watch the error grow compared with the real situation.

Surface Wind datasets.

12. David Ride presented an initial analysis of the 15 analogue Casella cup and vane anemometers. The raw data were first corrected for Magnetic to True north and anemometer misalignment, then converted from pulse counts to m/s using calibration values. The analysis took the form of averaging the wind speed over the 14 anemometers at 10m for each 5 minute averaged period over 31 days, subtracting this value from the average for each anemometer and dividing by the average. This algorithm gave the fractional increase or decrease from the mean with the results partitioned in 10E sectors. The graphs shown for each anemometer clearly identified directional arcs where forests, buildings and terrain features caused an increase or decrease in wind speed is strongly influenced by these features. Unfortunately, sampling errors in the presented data looked odd, primarily due to low wind speed data. Further analysis is to be performed using the spatial averaging distance (SAD) to be equal or greater than the separation of the anemometers coupled with an appropriate wind speed threshold (based upon the SAD e.g <0.8m/s for 600s averages); angular distribution analysis compared to the input field; and possibly use of anemometer No.9 at 30m as a reference instead of the means - as it is high enough to be unaffected by terrain and other features.

CBDE were actioned to provide analyses above as soon as possible. [graphics as *.pcx and *.cgm to be put into final database when assembled]

A closer look at the statistics for the anemometers only over the release periods coupled with the directional and speed analysis is planned in the near future. There is a high degree of confidence in the Casella anemometer data because of pre and post calibrations at the Met Office wind tunnel; comparison of the Met Office continuous observations and the RISØ sonic data were favourable.

13. RISØ commented that it would be useful to use 222E intervals rather than 10E and also show the frequency or number of data points in that segment. GMGO commented that the distributions and initial interpretations were very interesting. ARL suggested that the models not only needed to assess ground contour effects but also terrain above surface features such as forests and groups of buildings.

14. Dr Reinier Sterkenberg, representing Dr ART Hin from TNO Netherlands, announced that the processed calibrated data was available. The raw data recorded at 1Hz throughout the release period would be averaged in accordance with previous agreements and given to CBDE for incorporation into their wind dataset. The data were being used for NATO ATP45 predictions.

15. A brief discussion followed on the use of the data and several actions given:

TNO would supply the raw 1Hz and averaged data to CBDE to then incorporate it into their dataset for dissemination. CBDE were also actioned to obtain Dr Jones' Bivane and Sonic data gathered in the Met Field Propylene Dispersion experimental site.

Turbulence data

16. "Single level turbulence data". Torben Mikklesen presented RISØ's data from a single sonic anemometer using proprietary software . Data at 10Hz and averaged are available although 5 minute average were disseminated on floppy disk. Others averages on request. The values that would be available in the next release due September 1993 are : U, Dir, Temp, u* (stress), T*, heat flux, TKE (total kinetic energy) and z/L. RISØ would also provide software to analyze the raw data which would allow the user to specify and calculate as many or as few of the values as required. A thermo-couple at 7m provided a temperature check for the sonic value.

17. CBDE stressed the point that the wind direction values should be given as True North not Magnetic as distributed and assured that this would be corrected in the next release. It was also suggested that the mean sonic values for speed and direction could be added to the end of the analogue wind dataset. CBDE would consider this.

18. "Profiles of Turbulence", was presented for Chris Biltoft (DUGWAY) by Kenneth Nyrén (FOA) and showed some of the comprehensive dataset from the Dugway Sonic anemometers. He stated that the rest of the processed data would slowly become available as time permitted. These data were important as there were measurements from 2 towers having 2m and 6m 2D and 4m and 8m 3D Sonics as well as high resolution temperature probes. The smoke plumes were noted as having engulfed the sonic towers during several experiments. Kenneth explained that the raw data were available but an adaptive filtering QC procedure to despike the data from the ageing sonic heads was necessary and for the alignment correction to create the Ascii averages. Documentation on the QC procedure was made available.

19. RISØ was actioned to contact Chris Biltoft to find out when the data could be expected in Europe and to seek research money from the US Army Research Office in London, UK, to fund a short-term contract with RISq for the analysis of the Sonic data from the Dugway towers. Subsequent to this meeting, CBDE persuaded a US research group to fund DPG directly for the Sonic analysis to be completed by October 1993.

20. Several puffs passed through the DPG sonic arrays and the sonic data for these should be given priority because of the turbulence information they provide. [J262: .17.04 - 17.08 GMT and 17.13 GMT at the stop butt array J259: 13.45 GMT puff at the HS array ]

Temperature Profiles m

21. Welf aufm Kampe (GMGO) presented data from the PT100 resistor thermocouples located at the HS tower, also used by the DPG sonics, recorded as a paper trace. The paper records were then scanned and the data calibrated and corrected for inconsistencies. The dry bulb and wet bulb temperatures at 0.5, 2, 4, 6 and 8m above the ground were distributed on floppy disk with documentation in 5 minute averages. A Pasquill stability value was estimated from the 4m and 0.5m readings. Values are available from 1330 GMT September 14 to 1900 GMT September 23. A drainage flow development was noted .1600-1900 GMT on September 16.

Meteorological simulations

22. " S program visual simulation ". Josep Moreno-Santabarbara (attached to RISØ from Spain) presented his visual simulation program, written in Pascal, which showed the vector magnitudes of the CBDE Casella anemometers and RISq sonic overlaid onto a contour map of the CBDE range with markers for certain features. The simulation clearly showed excellent agreement of the sonic with the analogue anemometers. 23. The new version of Lincom (with data fitting and multiple wind initialisation) was demonstrated by Dr Srren Thykier-Nielsen (RISØ). The Lincom output was used as the input for the dispersion model Rimpuff. A PC demonstration of a single case for September 22 from 1100 to 1700 hrs GMT in Pasquill D was given.

24. There was great interest in the new version of the Lincom wind flow generator using multiple wind dataset initialisation and data interpolation. Copies were distributed on floppy disk to those interested.

25. Harald Weber presented GMGO's display software (written in Fortran and Starbase for a HP 9000/340). The output was X windows-like but of Starbase origin and showed multiple windows, each with a different dataset displayed -- The windows can be moved to suit each display application and configured for size and content. The background is the CBDE range map contours with 15 analogue CBDE anemometers and/or RISØ sonics with the option to add additional data as when made available e.g SF6 concentration data, DPG sonic arrays.

26. Harald Weber also showed viewfoils of model comparisons between the Ball & Johnson and Lincom plots with actual data. The B&J model was shown to be very terrain sensitive. The demo on the workstation was given on June 17. 27. Ron Cionco presented the results from the HRW model initialised with the upper air winds from the SODAR and BL Sonde. The HRW model is based on the Freeman et al model and modified for ARL's use. Various scenarios were shown in the PC demo.

28. David Ride opened the second day with two slides graphically illustrating still photographs taken by CBDE Photographic personnel. They showed contrasting meteorological conditions, one unstable and the other stable.

29. The experimental log of the MADONA field trial named RUNDOK.ORG compiled primarily by Kenneth Nyren (FOA) and Harald Weber (GMGO) was presented. The latest incarnation was made widely available on floppy disk. The proposed layout for the distributed CDROM would mimic the RUNDOK.ORG file structure. It was suggested that it should have 3 parts:- a general section with position data, meteorological data and dispersion data. Each main section having appropriate subsections with their own documentation files describing file formats and layouts of the experiments.

30. Everyone present agreed the format of the RUNDOK.ORG file would allow users of the database to find the relevant information easily when on CDROM.

31. Actions were placed on GMGO for the SF6 concentration data layouts, DLR for the lidar details, CBDE for 2 hourly mean wind speeds during the releases for stations 6 to 11, and CBDE for a copy of the Meteorological daily summary provided by the Met Office.

32. "Propylene concentration data" was presented by David Ride (CBDE) for Dr Chris Jones. It was explained that the new experimental UVIC sensors had calibration problems and difficulties with widespread background contamination (sheep faeces). Experiments with southerly winds may be OK with reduced contamination from this direction.

33. RISØ asked for some of the data and CBDE would request some data from Chris Jones.

Smoke plume and puff data including smoke behaviour videos

34. "RASCAL lidar data" was presented by Tim Higgs (CBDE) for Dr M Bennett (UMIST). RASCAL is an acronym for UMIST's RApid SCAnning Lidar. The data presentation consisted of a PC demo showing a movie of PCX graphics files in a series of different access modes from processed lidar data. A shot by shot, vertical scanned profile and horizontal scanned profile modes were available for display. The demo used a puff release to show the dispersion of the cloud using a display of arbitrarily scaled contour concentration data.

35. The PC demo was well received and many thought that the visualisation of the plume movement could be invaluable in understanding effects such as terrain influences and drainage flows. Henning Van Raden (GMGO) thought that the visual smoke behaviour in his videos compared well with that of the RASCAL demo.

36. Henning then showed his video sequences of plume and puff generation with background subtraction. This form of analysis allowed a vertical slice through the centre of the monitor picture to depict relative concentration fluctuations of the moving cloud profile. The system resolution is 700 lines at full screen at 2Hz update rate with 16 grey scales.

37. Hans Jørgensen (RISØ) presented his LIDAR data. The RISØ LIDAR has a cycle time of 3 seconds between shots at 1.5m resolution (0.6m in HiRes mode) and was 2-3m above the ground level. A comparison between the LIDAR plots and the GMGO SF6 concentration data was given and showed good agreement. 55 Puffs and 10 continuous runs were collected of which, 10 to 12 hours are of good quality. Some of the runs used the vertical scan mode available with this LIDAR with 512 points across the plume. Approximately 20Mb of data is available in total. The presentation also showed concentration vs distance (shot mode) and a 2hr integrated profile slice through the cloud with height for the run designated mad22f.

38. After lunch Hans showed a PC demo and stated that a LIDAR data analysis program was being prepared for distribution.

39. Jurgen Streicher (DLR) presented his car mounted LIDAR data. The LIDAR is connected to a PC by a serial port and mounted in the boot of an estate sized car. Relative mass concentrations are available because absolute calibration is not possible without knowing many unavailable variables. The plume on September 20 was noted as being of brown colour and had a different backscatter response than on other days. This behaviour is thought to be due to the very low humidity level on that day which caused the particles to remain relatively small (i.e low backscatter) and not to grow by absorbing moisture giving different optical properties. Puffs were viewed in the perpendicular and parallel to the direction of travel. Approximately 20Mb of ascii data is available for dissemination. The data are in the form of 3 second averages (normally of .2000 shots) with 3m resolution. The number of shots used in the average is dependent upon the plume/puff distance and is averaged until the signal to noise goes to a value above the DC transfer threshold. The data are available in TAB SEPARATED FORMAT on floppy disk.

40. Welf aufm Kampe presented the data for the WWDBw group from Munster. Paper documentation was distributed as well as a floppy disk with the ascii data. A total of 133 profiles through plumes were recorded. The van mounted flame spectrometer was driven at 1.3m/s " 0.3 m/s along the predefined routes with 250m markers manually inserted into the data files as reference points. It was found that a 7s delay (15s FWHM) occurred owing to the length of the instruments sampling line from inlet to ionisation. Once the stream was in the system, though, structure within the cloud was observed. For recording puffs, the van was stationary at or near the centre of the puff and volume concentration vs time at one location was recorded. The flow rates for the SF6 releases are given as m3/hr or m3/puff on the floppy disk.

41. Harald Weber (GMGO) presented comparison plots for experiments 141633f and 141654f. The closest wind dataset was used for the wind initialisation with RIMPUFF providing the comparison between a Gaussian model and measured SF6 concentrations. The two STANAG 2301 (CBW warning) diffusion coefficients Fy and Fz were derived from data by GMGO for increased accuracy and results indicate concentration values within x2 and directions very close.

42. CBDE were informed that their J day value in the averaged data of the anemometer dataset distributed pre-Christmas 1992 was in error by -1 day -- leap year not accounted for !

CBDE will rectify by October 1993 for database dissemination.

43. RISØ stated that it was possible to archive their LIDAR data as a self extracting archive Zip file onto a 6250bpi 1600' magnetic tape at 1600bpi. A binary sonic file to Ascii conversion routine and analysis program would also be provided.

44. Everyone felt that it was essential to appoint a "DATA Management committee" to oversee the collection of the whole database. There was a unanimous decision to allow Kenneth Nyren (FOA), Per-Erik Johansson (FOA) and Tim Higgs (CBDE) to collate and disseminate the final database. CBDE would master a CDROM with the whole dataset archived as laid out in the RUNDOK.ORG file. The cost to MADONA participants would be .,50 per CDROM to recover media and time costs only.

45. The data would be in ASCII format. Any deviations from this will require a comprehensive documentation file, and conversion or analysis software for users of the database CDROM. All wind datasets will have the same format, as will the sonics, and the LIDARs.

46. The Office of Federal Meteorology in Maryland USA may hold the database for the MADONA participants, although it was generally thought that they may charge us for the service which would not be a possibility.

47. Everyone endorsed the idea from CBDE that the final database format should be on a CDROM which CBDE were in a position to master for dissemination.

48. The CDROM format must be accessible using a MSDOS PC with a FAT (File Allocation Table) structure. The files would be primarily in Ascii and any in binary would be supplied with software to process the data.

49. The RUNDOK.ORG file was discussed again and several important points were reiterated. The root directory of the CDROM would hold the RUNDOK.ORG documentation file, PKUNZIP.exe, and text files giving the introduction and synoptic weather for the experimental period. There would be a readme.txt file in each subdirectory with information about the contents and essential details.

50. CBDE would discuss with Mike Bennett (UMIST) regarding the RASCAL dataset for incorporation onto the CDROM in a RISq lidar compatible format so that a standard layout may be adopted.

51. RISØ would provide a program to process the raw binary files in addition to providing the ASCII 5 minute averages. Several people wished to see other computed moments in the averaged files e.g U, dir, C, U'W', U'T', U'2, V'2, W'2, T'2, Heat flux, TKE and z/L. This would be available from the processing software to be provided by October 1993.

52. CBDE raised a very important question which had been overlooked, even by CBDE itself, that the time stamp given in the averaged ASCII files must be the beginning of the 5 minute period. i.e 0000-0005 would have the time stamp 0000. Everyone was actioned to ensure compliance of their datasets with this standard.

53. TNO were actioned to provide the raw data to CBDE from their wind anemometers.

54. It was stated quite clearly that all copyrights to both proprietary graphics and other such software must be observed at all dissemination levels. SESSION THREE --- Follow on Work Chairmen - Dr D Ride and Mr R Cionco

55. It was unanimously agreed that a morning or afternoon session at the American Meteorology Society (AMS) should be sought to present MADONA data consisting of between 6 to 8 papers. The likely date for this meeting was given as April to July 1994 or perhaps as late as April 1995.

56. CBDE stressed that the initial AMS paper produced by this group must be cleared by UK MOD before any derivative papers could be freely published by individual groups.

57. A date for the MADONA Technical Exchange meeting was tentatively made for the summer of 1994 over a 4 day period to be hosted by ARL in New Mexico, USA.

58. Ron Cionco was actioned with confirming the AMS meeting date before committing the Technical Exchange meeting date. This action to be completed as soon as possible.

59. A formal list of the available data and its current state was given :

61. A draft version of the paper destined for the Bulletin of the AMS was circulated to all on the previous night and intense discussion followed as to what should be included and what should be left out.

62. CBDE stated that all topics in the MADONA study would need to be covered in the abstract so that a blanket cover for subsequent publications could be issued after the overview has been cleared as UNCLASSIFIED.

63. Everyone agreed with the opinion that the importance of the MADONA experiment should be emphasised in the paper overview in that so much high quality, high resolution (both spatially and time) data employing complementary techniques was assembled on terrain typical of much of central Europe.

64. CBDE clearly stated that the use of Ordnance Survey maps in the B of AMS or any other publication was prohibited for copyright reasons. CBDE would produce a range map from internal MOD sources suitable for open publication.

65. There was further intense discussion regarding the introduction and other components of the draft paper. FOA, CBDE, GMGO, RISØ and ARL would remain in close contact using FAX to move revisions of the draft around for comment. The FAX would also be of great benefit whilst aggregating the database information.

66. It was decided that GMGO should attempt a GRAND TABLE for the draft paper as this would simplify the overview and decrease its size dramatically.

67. CBDE emphasised that the draft should be no more than 10 pages of A4 before B of AMS formatting in order to expedite publication clearance. Publication of the paper in the B of AMS would be expected in mid 1994 if the draft could be finalised by October 1993. 68. GMGO were formally thanked for their hospitality and organisation during the data exchange meeting which was closed at 1445 (EST) followed by several hours of informal demonstrations and data exchange.

Dr M Bennett Simon Environmental Technology Centre Department of Chemical Engineering UMIST PO Box 88 Manchester M60 1QD

(for Dr T Mikkelsen) Meteorology and Wind Energy Department Risr National Laboratory PO Box 49 DK-4000 Roskilde Denmark

Director (for Mr K Nyren) National Defence Research Establishment FOA 4 Cementv@gen 20 Erdsboda S-901 UmeD Sweden

Dr Christian Werner Deutche Forschungsanstalt fhr Luft und Raumfahrt Institut fhr Optoelectronik D-8031 Oberpfaffenhofen Federal Republic of Germany

Dr Kocksholt Wehrwissenschaftliche Dienststelle der BW fhr ABC-Schutz -Denzernat 424- Postfach 13 29 D-3042 Munster/Oertze Federal Republic of Germany

President (for Mr W aufm Kampe) German Military Geophysical Office Science Department Applied Meteorology Mont Royal D-5580 Traben-Trarbach Federal Republic of Germany

Divisional Director (for Mr R M Cionco) Battlefield Environment Directorate US Army Research Laboratory White Sands Missile Range New Mexico 88002-5501 United States of America

Director (for Mr C Biltoft) Meteorology Division US Army Dugway Proving Ground Utah 84022-5000 United States of America

Dr Reinier Sterkenburg (for Dr A R T Hin) Protection and Risk Assessment Prins Maurits Laboratory Lange Kleiweg 137 PO Box 45 2280 AA Rijswijk The Netherlands