Home

"HmFBA User Manual"

1. i i f f f t f i i fi i i Aiii WANE Example of pi ed mostly from refracted energy and seismic data from a line that passes over the OBS Areal picks contours in samples 7 T T T T 720 samples 1440ms f The poor near zero picking South North Ps 8 8 Here the picking is excellent Smooth contours is in the farthest offsets not used for positioning West East 16 Northing 2 82 2 D Mode Processing 2 2 D Mode versus 3 D Mode Acoustics e g USBL are a refined precise observable First breaks are crude imprecise in a random sense and they are biased by picking and instrumental delays Acoustics are expensive to acquire First breaks are essentially free on an OBC OBN OBS crew HmFBA calibrates biases by having a good balance of azimuth and offset A skilled user assures this A lead selection and or by automatic geometry balancing described on the next slid
2. 2 If a GUI configuration for your prospect has been saved you may read it now with the Read button Later you can save the current configuration with Save 3 HmFBA processes SEG Y and Seismic Unix SU receiver gathers Push Load gather for a single gather or Load pick swath for an entire swath 4 The Screen gt File button will dump the screen to a file The Clear screen button will clear the screen if it becomes jumbled due to word wrap 5 The Pick button will pick the first breaks of the current gather Choose desired pick method and parameters Swath loading and picking happens with Load pick 6 The Plot button in Plot seismic with picks will display as many traces as selected with picks See manual for more information about selecting picks 7 The Trace highpass filter conditions the traces before picking by subtracting a LOESS smoothed trace of specified length thus removing low frequencies 8 The X Y gather button begins the least squares process for a single gather resulting in coordinates plus several plots and much output 9 Choose Water if processing water arrivals Modify the VP and Bias as necessary compute them automatically or constrain depth 10 Choose Refractor if processing refracted energy Regression order is relevant to refractor processing but VP and bias are not 11 The distribution of picks in a swath or gather can be automatically balanced in azimuth and o
3. 3 Mode 2 D or 3 D applies only to single node receiver and X Y sequential swath activities It switches the processing mode from 3D for use with direct water arrivals to 2D for use with seismic energy that arrives by water and or by one or more refractors 3 D mode water produces different but similar plots and statistics which are exhibited near the end of this user s manual In 3 D mode water the VP acoustic velocity of propagation and bias sum of picking and instrumental fields become active Enter those values if you know them otherwise use the defaults Trace high pass low cut filter conditions the traces before picking by subtracting a quadratic LOESS smoothed trace from the actual trace This effectively cuts out the low frequencies that might be due to tides swell or waves in deep water chop in shallow water instrumental characteristics et cetera Specify the length of the LOESS smoother This may or may not improve picking An example of the high pass filter is shown in the next two slides of sample traces from an ocean bottom seismometer OBS 2465m deep on a mid Atlantic ridge near the Azores There is more information on the plots exhibited in this user s manual on a later page of the manual and in the Acknowledgments Traces wni sce meren IIN 4 iy ji Y 5 Ji A i NK A Pa j sf ii i 4 WES j ARIN A QJ 2 oO D Q Q Wik ye rit Sample OBS traces with l
4. Azimuth histogram 10 degree bins SD2Mean 0 018997 Excellent balance Number of picks S 8 8 8 8 3 Azimuth deg Range gu versus picks ms Very linear relationship between pick time and distance after processing Computed range in gu 1400 1500 1600 1700 1800 1900 2000 2100 2200 Picks in ms 37 Concluding Comments 1 There are many ways to position OBS receivers Dedicated high frequency positioning acoustics e g USBL are the most common way and the most expensive in time and equipment Direct seismic airgun water arrival first break positioning lines are also possible Extra time is required but no extra equipment Unfortunately the first obreak observable is much cruder than the acoustic observable and there are first break picking delays and instrumental delays that are difficult to calibrate Therefore direct water arrival first breaks are not the same as dedicated acoustics A third technique is to use wide azimuth far offset production seismic data lots of them This is the cheapest technique since no dedicated positioning lines are required Vastly more data are available than in water arrival first break positioning so the statistics of large numbers make up for the coarse quality of the first break observations Because data are observed at all azimuths and offsets picking and instrumental delays are easily calibrated Concluding Comments 2 On the other hand far
5. picking and instrumental delays is Just a guess best left at zero for starters 3 D Mode Processing 2 The four approaches to 3 D mode processing in HmFBA are A3 D mode solution can be computed using only a priori VP and bias inputs HmFBA provides a linear regression first order of the picks versus the straight line distances between the sources and the computed 3 D position that provides a posteriori VP and bias By splitting the differences between the a priori and a posteriori parameters the process can be iterated until both are the same HmFBA can do this automatically by checking Automatic VP bias but less successfully than manually because of the high correlation especially in poor offset geometry between the static bias and depth It should be obvious by the number of iterations the depth and the a posteriori parameters when Automatic VP bias is not successful If the receiver depth is known perhaps by prospect bathymetry then depth can be constrained The depth constraint is most successful if Automatic VP bias is also checked the default Depth can also be constrained with manually entered VP and bias but large residuals are possible if there is a conflict between the depth and the entered VP and bias due to their correlation In all these four approaches the reported graphics and statistics iterations a posteriori VP and bias unit variance percentage of used picks computed depth pro
6. some plots are based on the inner and outer pick limits AIF plots all the picks 16 Show manual for instructions about modeling Node oscillator drift and for Trim amp seed with GroupXY Note This screen may be scrolled Output can be highlighted with the cursor copied ctr C and pasted elsewhere or use Screen gt File button Meter foot or arc second source coordinates Central Meridian if arc seconds Pick methods amp parameters Least squares parameters C Absolute amplitude i SES Begin by reading the legal disclaimer Show manual This enables other controls 12 Sample length 15 Threshold max this traces with picks C Gradientthreshoid 2 Samples amp gt 1 Traces Plot gt Piot m 2 miliseconds 4 Show manual Close all plots www Screen scroll span lines 200 Tau non centrality Meter foot or arc second coordinates 100 meter foot arc sec 33 CM Relative depth in gu TT Anisotropy angularity T Balance azimuth Escape to the web Receiver ID cdi I Balance distance p SourceMeasurementunit z Node oscillator drift day microseconds a IT Trim amp seed with Groupxy _Sereen gt Fite FEE ax onsen as Clear screen EJ HmFBA_v1b103 Hydrometronics LLC HmFBA v1b103 Wide Azimuth Far Offset First Break OBC OBN Positioning Hydrometronics Wide Azimuth Far Offset First Break Positioning Output will display here Brief instruct
7. 2500 3000 3500 Pick time in ms 22 Inversed distance in grid units Velocity in grid units per millisecond Picks after processing with regression fit This regression plot after processing is very clean with an outlier rejecting tau non centrality of 4 This accounts for the Unit Variance of 1 18 with pick SDs of 20m and resulting coordinate SDs of 54cm See a later graphic for an expanded view of the regression plot L 1 1 1 600 800 1000 1200 Pick time in ms Slope of regression equation interpreted as velocity 2 6 This OBS was processed in 2 D mode refractor amp water with a fourth order Chebyshev regression equation This curve is the numerical differential of the equation interpreted as velocity Notice that the velocity is well above that of water 1 5m ms thus indicating refracted arrivals 4 al L 600 800 1000 1200 Pick time in ms 23 Inversed distance in grid units Picks after processing with regression fit For comparison I ve opened up the offset range from 1400 to 3000ms i e everything picked is displayed This range however is unsuitable for positioning for reasons of inadequate geometry balance Notice that the farther offsets are noisier and that the red regression line bends 4 L 1 i 1000 1500 2000 2500 Pick time in ms 2 D Mode Processing 7 Chebyshev Regression and Velocity In 2 D mode refractor amp water the Chebyshev regression of f
8. Energy is the same as above except that energy amplitude squared is used instead of absolute amplitude NB Both absolute amplitude and energy use the same parameters but their performance with those parameters will differ Because there are two parameters selection Is an iterative process of experimentation In general a larger threshold which means a smaller divisor called this tends to pick later in the trace smaller thresholds pick earlier Gradient threshold method computes the normalized mathematical gradient along the trace and picks the first above a specified threshold Increase the default threshold for noisy data e g 10 decrease for clean data e g 1 Larger thresholds the only parameter tend to pick later in the trace smaller thresholds pick earlier See Acknowledgements at end of this manual for a reference Least Squares Parameters 1 Pick a priori SD in gu is an assessment of the standard deviation in grid units gu of a first break pick as a surveying observation Decreasing this SD will increase the computed UV Iteration stops when the shift in coordinates from the last iteration is less than the Tolerance per Rx in gu Regression order is the order of the Chebyshev regression equation that converts picks in milliseconds into observations in grid units 2 D mode refractor amp water only Pick inner and outer limit ms define the selection of picks Iterat
9. Least squares parameters 8 Picka priori SD in gu Tolerance per Rx in gu Regression order 1 2 Pick outer limit ms Ea Maximum iterations 2 milliseconds Show manual Trace high pass low cut filter T Fitter length gt 11 Close all plots Areal pick plots amp interval Se www BEN Screen scroll span lines Screen gt File 200 Clear screen Table of Contents Overview A tour of the HmFBA application 2 D mode refractor amp water processing of single nodes including 0 1 4 00 Pick inner limit ms 000 50 4 Tau non centrality ro Relative depth in gu I Anisotropy angularity T Balance azimuth Balance distance 7 Trim amp seed with GroupXY S009 Max range in gu Picking quality control 3 D mode water versus 2 D mode refractor amp water Automatic geometry balancing on distance and azimuth Comments on Snell s Law Chebyshev regression equation and velocity Least squares LS adjustment residuals and statistics Other topics Swath processing 2 D and 3 D modes 3 D mode water processing Concluding comments Appendix 1 Acknowledgements Appendix 2 Saved file format Appendix 3 Working with Matlab plots Appendix 4 Hardware software and security requirements read Appendix 4 before installing or running HmFBA Appendix 5 Glossary of terms Overview Hydrometronics First Break Analysis HmFBA loads picks and adjusts direct water arrival and or wide azimuth
10. box in the lower left of the HmFBA GUI HmFBA processes receiver gathers either individually in receiver activities or en masse in swath activities by selecting loading and picking a number of gathers all together It is handy especially in swath activities for the receivers to have unique ID numbers SEG Y and SU receiver gathers are prepared by geophysicists but it is apparently not standard among geophysicists where among the available SEG Y headers to place the receiver IDs Consequently the list box in the lower left of the HmFBA GUI allows the user to choose which header to associate with the receiver ID This choice must be made before picking the gather If you don t know which header it is you may have to use a SEG Y viewer to find out If the receiver ID is not among the headers then use the top default choice Null none 0 In this case the receivers will be numbered sequentially during swath processing Screen gt File dumps the entire display screen up to 3000 lines to a text file with a name of your choice The file can then be edited Clear screen if needed before Screen gt File or if the screen gets jumbled due to incorrect word wrap Screen scroll span is the number of text lines the display screen will scroll More lines slower performance but more lines may be necessary for swath activities to see all receiver coordinates 11 Pick sample Other Controls
11. far offset refracted arrival OBS receiver gathers in SEG Y or Seismic Unix SU formats receiver by receiver or an entire swath for best position Computations are done in map ee grid units gu which are meters or feet determined by the SEG Y or SU Described later HmFBA picks first breaks using three very different methods with user selectable parameters saves and loads first breaks as CSV files see Appendix 2 for format and optionally conditions seismic traces with a high pass filter for better picking Seismic traces with their first break picks plotted can be viewed Picks can be viewed in areal contour plots as additional QC HmFBA solves for receiver or swath vertical velocity gradient in 2 D mode optionally balances geometry optionally compensates for anisotropy angularity trims amp seeds with GroupXY and provides diagnostic QC statistics and graphics HmFBA provides rapid feedback from picking to positions If interested in having your data processed in HmFBA or in trying HmFBA yourself then contact Hydrometronics LLC www hydrometronics com noel zinn hydrometronics com A Tour of the Application E HmFBA_vib103 Hydrometronics LLC HmFBA v1b103 Wide Azimuth Far Offset First Break OBC OBN Positioning Hydrometronics Wide Azimuth Far Offset First Break Positioning Output will display here Brief instructions follow 1 Begin by pushing the Show manuaf button This will display the legal notice and the m
12. only effective if Automatic VP bias is also selected onstraining depth without automatic VP and bias computation is likely to cause large residuals Also constraining o requires a fully populated weight matrix which for an extremely large job probably in very deep water considering Snell s Law may strain computer resources See 3 D mode water guidelines later in the manual 10 Other Controls 1 Meter foot or arc second coordinates Source coordinates in SEG Y and SU headers are typically reported in map projection grid units by industry meters or feet Academic institutions may report in geographical arcseconds If arcseconds then choose a Central Meridian in degrees so that internal computations can be done in wide zone Transverse Mercator in meters GUI configuration Read or Save all the parameters set in the GUI Use these controls for sharing set ups with colleagues or for returning to a previous project You may need to click the 2 D or 3 D buttons to enable the right displays Stop stops receiver picking swath load pick and receiver and swath adjustments all of which can be time consuming w w w gets you to the World Wide Web via the Hydrometronics home page Close all plots Matlab plots consume memory and thus affect performance HmFBA generates a lot of plots Regularly close plots not in use Other Controls 2 Receiver ID list
13. regression fit 2000 2500 3000 3500 4000 4500 5000 5500 6000 6500 7000 o a O O D 2 a D a N Z 5 ar _ D gt O 2 o gt 1 38 Pick time in ms Numerically computed slope interpreted as velocity Notice that it s close to the speed of sound in water This means that despite the offset up to 7000ms direct water arrivals have been picked perhaps due to the depth about 2500m 2000 2500 3000 3500 4000 4500 5000 5500 6000 6500 7000 Pick time in ms 30 Pick residuals or t ma i ates Need a Een pE ee os eS r 200 sun pub ul 39U8 SIp p SJ AUI Jo JenpIs yY 2000 2500 3000 3500 4000 4500 5000 5500 6000 6500 7000 Pick time in ms Histogram of pick residuals This histogram reveals that the picks are noisy histogram is a good also noted in the approximation to a normal pick a priori SD of 60m confirmed by the UV of 1 06 distribution uoneindodg n a Ao E e 2 H a x 2 oO 31 i moar teore Swath Output Statistics 2 Easting o X Y all at once is al 4 SP Easing all at once is always 5 SD Northing z 6 Number of used first breaks for this receiver 2 D d 7 Latitude only with SEG Y in arcseconds l n m O e 8 Longitude only with SEG Y in arcseconds Elapsed time for simultaneous adjustment 3 874 RxID grid_coordinates UV SDX SDY used_picks HDOP geogr
14. the least squares process for a single gather resulting in coordinates plus several plots and much output Choose Water if processing water arrivals Modify the VP and Bias as necessary compute them automatically or constrain depth 10 Choose Refractor if processing refracted energy Regression order is relevant to refractor processing but VP and bias are not 11 The distribution of picks in a swath or gather can be automatically balanced in azimuth and or distance by checking the appropriate boxes 12 The Save buttons will save a picked swath or gather respectively as a CSV file for later loading with the Load buttons See manual for format 13 X Y sequential will process all the receivers sequentially one at time and X Y all at once will process all the receivers simultaneously 14 Check anisotropy angularity to compensate for these effects by leveling the pick residuals as a function of azimuth 15 Set the areal plot contour level The some plots are based on the inner and outer pick limits All plots all the picks Node oscillator drift and for Trim amp seed with GroupXY C ompens ate for jhlighted with the cursor copied ctr C and pasted elsewhere or use Screen gt File button oscillator drift in an Pick methods amp parameters GUI configuration Least squares parameters ocean bottom node Display screen control Dump screen f text to a ASCII file ip eee I a Se eeping Clear sc
15. the likelihood quantified between 0 and 1 of a random event to happen A probability of 0 is no likelinood a probability of 1 is certainty A probability distribution is the mathematical relationship between event such as the value of an observation and it s probability of occurrence The two probability distributions discussed in HmFBA are the normal and the tau Random error is a deviation from the truth for stochastic reasons having to do with the imperfections of the measurement process Random error averages out to the truth unlike bias or blunder Redundancy See degrees of freedom Regression is a statistical model that defines the expected value of one variable in terms of the value s of one or more other variables Linear regression is first order Quadratic regression is second order Higher order regressions are possible as in HmMFBA Regression equation See Chebyshev regression equation Regression order See Chebyshev regression equation A residual is the difference between an observation and its adjusted value SEG Y is a standard format for storing seismic data developed by the Society of Exploration Geophysics SEG Seismic Unix SU is a format for storing seismic data a variation of SEG Y part of an open source seismic utilities package supported by the Center for Wave Phenomena at the Colorado School of Mines Semi major and semi minor are the axes of an error ellipse derived by rotating the variance covar
16. 1000m tan 49 1150m or 1000m tan 16 287m In 1000m of water that direct water arrival slant range is on cos 49 1524m 1016ms or 1000m cos 16 961m ms In 100m of water divide these numbers by 10 These facts imply that when positioning receivers with dedicated first break runs more picks than anticipated may be arriving through a refractor depending upon your offset from the receiver and your picking parameters which can be tuned to pick later water arrivals 18 2 D Mode Processing 5 The next five plots from the same ION Geophysical prospect show distance and azimuth distribution as HmFBA configuration Plan view of the receiver gather before processing Plan view of the receiver gather after processing A 40 bin sloping histogram of distance after processing A 10 degree flat histogram of azimuth after processing Inner and outer limits of distance processing a donut are one way to achieve balance in azimuth and distance manually Automatic geometry balancing is another 2 D Mode Processing continues after the plots EJ HmFBA_vib103 Hydrometronics LLC HmFBA v1b103 Wide Azimuth Far Offset First Break OBC OBN Positioning Grid units in meters picking sample in miliseconds 2 i Picking by absolute amplitude sample length 6 threshold 70 high pass yes with length 41 LS parameters pick SD 20 tolerance 0 3 order 6 inner 300 outer 1400 max iter 50 t
17. 1681 HmFBA test data was provided by instruments from the Ocean Bottom Seismograph Instrument Pool hittp www obsip org which is funded by the National Science Foundation under cooperative agreement OCE 1112722 A link to the R V Marcus G Langseth Endeavour Tomography Expedition MGL0910 ETOMO survey follows http ds iris edu data reports 2009 09 014 The Principal Investigators of the NSF funded survey on the mid Atlantic ridge near the Azores MGL 1305 provided data for the testing of HmFBA Two plots in this user s manual are derived from this survey A link to the MGL 1305 survey report follows http www whoi edu sbl liteSite do 7litesiteid 90993 Appendix 1 Acknowledgements 2 The Statistics of Residuals and The Detection of Outliers Allen J Pope 1976 NOAA Technical Report NOS 65 NGS 1 the outlier detection scheme in HmFBA https www ngs noaa gov PUBS LIB TRNOS65NGS1 pdf Using Cross Correlated Head Wave and Diving Wave Seismic Energy To Position Ocean Bottom Seismic Cables a University of Houston GEOL 7333 Seismic Wave and Ray Theory term paper by Noel Zinn 1999 overall approach and a generic first break picker http www hydrometronics com downloads GEOL 207333 20Term 20Paper paf SegyMAT Read and Write SEG Y and SU files using Matlab http segymat sourceforge net Agus Abdullah PhD Ensiklopedi Seismik Online gradient threshold first break picking method ht
18. T Trim amp seed with GroupxY 0 Screen gt File 200 Clear screen 5000 Max range in gu Regression order El HmFBA_v1b103 Hydrometronics LLC HmFBA v1b103 Wide Azimuth Far Offset First Break OBC OBN Positioning Hydrometronics Wide Azimuth Far Offset First Break Positioning Output will display here Brief instructions follow zJ 1 Begin by pushing the Show manual button This will display the legal notice and the manual detailed instructions and enable activities 2 If a GUI configuration for your prospect has been saved you may read it now with the Read button Later you can save the current configuration with Save 3 HmFBA processes SEG Y and Seismic Unix SU receiver gathers Push Load gather for a single gather or Load pick swath for an entire swath The Screen gt File button will dump the screen to a file The Clear screen button will clear the screen if it becomes jumbled due to word wrap The Pick button will pick the first breaks of the current gather Choose desired pick method and parameters Swath loading and picking happens with Load pick The Plot button in Piot seismic with picks will display as many traces as selected with picks See manual for more information about selecting picks The Trace highpass fiter conditions the traces before picking by subtracting a LOESS smoothed trace of specified length thus removing low frequencies The XY gather button begins
19. User s Manual for Hydrometronics LLC HmFBA Hydrometronics First Break Analysis LEGAL NOTICE THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS AS IS AND ANY EXPRESS OR IMPLIED WARRANTIES INCLUDING BUT NOT LIMITED TO THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE ARE DISCLAIMED IN NO EVENT SHALL THE COPYRIGHT OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT INDIRECT INCIDENTAL SPECIAL EXEMPLARY OR CONSEQUENTIAL DAMAGES INCLUDING BUT NOT LIMITED TO PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES LOSS OF USE DATA OR PROFITS OR BUSINESS INTERRUPTION HOWEVER CAUSED AND ON ANY THEORY OF LIABILITY WHETHER IN CONTRACT STRICT LIABILITY OR TORT INCLUDING NEGLIGENCE OR OTHERWISE ARISING IN ANY WAY OUT OF THE USE OF THIS SOFTWARE EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE Copyright 2014 2015 Hydrometronics LLC Wide Azimuth Far Offset First Break Positioning A User s Manual for HmFBA Hydrometronics First Break Analysis Noel Zinn Hydrometronics LLC www hydrometronics com 12 October 2015 EJ HmFBA_v1b103 Hydrometronics LLC HmFBA v1b103 Wide Azimuth Far Offset First Break OBC OBN Positioning Hydrometronics Wide Azimuth Far Offset First Break Positioning Output will display here Brief instructions follow 1 Begin by pushing the Show manual button This will display the legal notice and the manual detailed instructions and enable activities
20. anual detailed instructions and enable activities 2 If a GUI con ct has been saved you may read it now with the Read button Later you can save the current configuration with Save er gathers Push Load gather for a single gather or Load pick swath for an entire swath i n ear screen button will clear the screen if it becomes jumbled due to word wrap Display screen Basic Teee TA many traces as selected with picks See manual for more information about selecting picks instructions Output icking by subtracting a LOESS smoothed trace of specified length thus removing low frequencies for a single gather resulting in coordinates plus several plots and much output scrolls here VP and Bias as necessary compute them automatically or constrain depth ression order is relevant to refractor processing but VP and bias are not 11 The distribution of picks in a swath or gather can be automatically balanced in azimuth and or distance by checking the appropriate boxes 12 The Save buttons will save a picked swath or gather respectively as a CSV file for later loading with the Load buttons See manual for format 13 X Y sequential will process all the receivers sequentially one at time and X Y al at once will process all the receivers simultaneously 14 Check anisotropy angularity to compensate for these effects by leveling the pick residuals as a function of azimuth 15 Set the areal plot contour level The
21. au 4 depth 100 mode 2 D balancing no anisotropy angularity correction no drift 0 UV 1 1822 Scaled SDX 0 53816 Scaled SDY 0 5338 HDOP 0 034857 20 bin distance SD mean ratio 0 15448 40 bin distance SD mean ratio 0 19623 10 degree azimuth SD mean ratio 0 16786 20 degree azimuth SD mean ratio 0 1192 Used picks 3306 Selected picks 3337 Total picks 6931 Receiver coordinates 285219 68627 7437308 0728 Receiver ID 0 Time seconds processing 0 64921 iteration 8 position jump in grid unit 0 053007 H H siaaa a ah asa ae Configuration of HmFBA for these data iteration 6 position jump in grid unit 0 25729 iteration 5 position jump in grid unit 0 23565 iteration 4 position jump in grid unit 38 9437 iteration 3 position jump in grid unit 163 4053 iteration 2 position jump in grid unit 703 808 Swath activities Receiver activities Pick methods amp parameters GUI configuration Least squares parameters Load pick swath Load gather Absolute amplitude Read Save 20 Pick a priori SD in qu sequentia Pick Piot Energy Mode 2 D or 3 D Tolerance per Rx in gu 2d0 C 6 Sample length gt n Regression order Load Save Load Save 70 Threshold max this Pick inner limit ms Eas i Pick outer limit ms Plot seismic traces with picks C Gradient threshold 0 5 stomatic VP bias Gow 1 2001 lt Samples amp gt 1 nstrain depth Maximum iterations 79104 lt Traces P
22. au non centrality of 2 will trim about 5 of the data 3 will trim lt 0 3 of the data and so on Trace 1 Sequence of recorded seismic amplitudes 2 sum of the diagonal terms of a matrix The Unit Variance UV is a the sum of the squares of the weighted residuals divided by the degrees of freedom If the a priori standard deviations are a correct assessment of the true random errors of the observations biases and blunders excluded then the UV computed in the adjustment will equal unity 1 USBL is Ultra Short Baseline an acoustic system providing one range distance a inclination angle and an angle relative to vessel centerline Variance is the mean of the squared residuals See residual The square root of the variance is the standard deviation Variance covariance matrix See normal matrix Velocity of Propagation VP Speed of sound in water A vertical velocity gradient is a variation in seismic velocity as a function of offset between the source and the receiver Energy traveling farther are more likely to dive into deeper faster refractors Weight is the inverse square of the a priori standard deviation assigned to an observation 47
23. be improved by lowering the regression order N N S 3 Population g 40 20 0 20 40 Pick residuals in grid units 2 D Mode Processing 8 A good least squares adjustment converges on the final coordinates at the rate of about an order of magnitude per iteration HmFBA accomplishes this especially in the early iterations despite the refracted observations themselves changing slightly at each successive iteration due to revised Chebyshev regression coefficients despite outlier rejection and despite automatic geometry balancing all of which slow convergence in the later iterations The least squares criterion is that the sum of the squares of the residuals is a minimum Those residuals are shown on the previous three slides Finally we turn to the numerical statistics provided by HmFBA Anything can be copied from the display screen and pasted elsewhere as required e g as in a report Alternatively the Screen gt file button will save the display screen as a text file with a name selected by the user The next slide shows the results for the adjustment so copied 2 D Mode Statistics Grid units in meters picking sample in milliseconds 2 Picking by absolute amplitude sample length 6 threshold 70 high pass yes with length 41 LS parameters pick SD 20 tolerance 0 3 order 6 inner 300 outer 1400 max iter 50 tau 4 depth 100 mode 2 D balancing no anisotropy angularity correctio
24. by energy sample length 5 threshold 20 high pass yes with length 11 LS parameters pick SD 60 tolerance 0 4 order 4 inner 2000 outer 7000 max iter 50 tau 4 depth 2500 mode 2 D balancing no anisotropy angularity Swath activities Receiver activities Pick methods amp parameters GUI configuration Least squares parameters Load pick swath C Absolute amplitude Read Save 60 Pick a priori SD in gu Energy Tolerance per Rx in gu s Sample length 20 C30 Regression order 20 Threshold max this Pick inner limit ms Plot seismic traces with picks Gradient threshold 6 Automati jai Pick outer limit ms 1 2001 lt Samples amp gt 7 __ c Maximum iterations Trace sample rate 1 9104 lt Traces Plt gt Plot Ga mpo A milliseconds Meter foot or arc second coordinates A Trace high pass low cut filter Relative depth in gu C meter foot arcsec 128 CM I Filter length gt 11 Close all plots I Anisotropy angularity Receiver ID Areal pick plots amp interval I Balance azimuth cdp All 4 Some www T Balance distance SourceMeasurementUnit a i Node oscillator drift day microseconds Screen scroll span lines T Trim amp seed with GroupxY _sereen gt rite 200 _Giear screen E vere Receiver gather after processing 29 11000 10000 9000 8000 7000 6000 5000 Inversed distance in grid units 4000 3000 Picks after processing with
25. caled SD in depth 0 0512 automatically Not always Slope VP in gu ms 1 4947 intercept bias in gu 28 8619 Intercept bias in ms 19 3099 possible Time seconds processing 0 16973 iteration 6 position jump in grid unit 0 076186 iteration 5 position jump in grid unit 1 1119 Swath activities Receiver activities Pick methods amp parameters GUI configuration Least squares parameters Load pick swath Load gather Absolute amplitude 8 Pick a priori SD in gu X Y sequential ie t Energy 0 1 Tolerance per Rx in gu XY al at once E 5 Sample length 4 1 5 VP in gu ms Load Save Load Save 20 Threshold max this GE besin ms Pick inner limit ms oH Pick outer limit ms Plot seismic traces with picks C Gradient threshold 6 Automatic VP bias is 1 2001 lt Samples amp gt 1 IT Constrain depth Maximum iterations a aes Trace sample rate 1 9104 Fee a tea 2 slozeni Tau non centrality Se oo Trace high pass low cut filter meter foot C arc sec 33 CM PM Fiter length gt 11 Close all plots I Anisotropy angularity Receiver ID Areal pick plots amp interval T Balance azimuth cdp B Stop SourceMeasurementUnit All 40 Some www Balance distance ce x Node oscillator drift day microseconds Screen scroll span lines T Trim amp seed with GroupXy o Screen gt File 200 Clear screen 9000 Max range in gu 3 D Mode Output Statistics Grid units in meters picking samp
26. ce versa The coordinate uncertainties from the inverse normal matrix are scaled by the UV and therefore are independent of pick SD Choose a convergence tolerance that is small but not so small that it runs up the iterations to the limit In 2 D mode refractor amp water regression order has an effect Experiment Order 2 should always work Orders 3 and 4 are generally good Higher orders are possible For a flat refractor regressions order 1 yields additional information slope VP and intercept sum of the biases See next slide Choose inner and outer limits to isolate the water arrivals or to find a clean refractor for example The maximum number of iterations is a fail safe number If the convergence tolerance is too low iterations may increase 27 2 D and 3 D Swath Processing Swath Processing X Y sequential swath processing is nothing more than en masse automatic single node 2 D mode or 3 D mode processing without the time and space consuming graphics X Y all at once swath processing brings more to the party All nodes are processed simultaneously in a single 2 D mode refractor amp water adjustment with the same Chebyshev regression coefficients In many prospects this is an advantage but not always Multi receiver plots in this manual are generated from ae ong of several ocean bottom seismometers OBS from the NSF funded MGL0910 ETOMO survey west of Oregon A See Acknowledgem
27. dump the screen to a fie The Clear scr tolerance regression order inner and 5 The Pick button will pick the first breaks of the current gather Choose 6 The Piot bution in Plt seismic with picks wil display as many traces outer pick limits maximum number of 7 The Trace highpass filter conditions the traces before picking by subtra f Fe e e nee iterations tau non centrality parameter a a lea otal E ome A NS T he E tae estimated receiver depth 12 The Save buttons will save a picked swath or gather respectively as 13 UY sequentia wal process al the receivers sequentialy one at tine anisotropy angularity balance azimuth balance distance trim amp seed Plot seismic traces with picks for the selected samples and traces and at cnr ant i Areal contour Ska poten a the amplitude selected Tolerance per xin gu plots of pick samples and ae pick interval Seer Plot seismic traces with picks trend lt Samples amp gt 1 sran depth Maximum iterations tend lt Traces Pot gt Piot 2 ee E Tau non centrality S So ee Trace high pass low cut fiter 100 Relative depth in gu meter foot arcsec 33 CM I Fitter tength gt 11 Close all plots I Anisotropy angularity Receiver ID Areal pick plots amp interval I Balance azimuth a o em www TE Sa SourceMeasurementUnit i s Node oscillator drift day microseconds Screen scroll span ines
28. e HmFBA has two modes of operation 2 D refracted or water arrivals or both single receiver or swath 3 D water arrival only single node or swath X Y sequential In very deep water direct water arrivals can shots are close enough to the detector provide for an adequate balance of offset and azimuth and sufficient numbers to compensate for pick imprecision See the 3 D Mode Processing section later in this manual for more details In shallower water and in many deep water prospects too most useful first break seismic energy arrives at the receiver via the water AND one or more refractors in which the velocity of propagation may vary In 2 D mode HmFBA solves for this velocity gradient with a Chebyshev regression equation that relates pick time in milliseconds to distance in grid units while also solving for delay biases Depth is not solved in 2 D mode 17 2 D Mode Processing 3 Automatic Geometry Balancing In both 3 D mode water and 2 D mode refractor amp water first break positioning works best if the source locations are balanced in azimuth and offset distance with respect to the receiver HmFBA provides tools to achieve this manually HmFBA also provides for automatic geometry balancing in azimuth and distance for rapid processing when manual intervention is not timely or possible Also HmFBA supports simultaneous swath adjustment which minimizes the effects of geometry imbalance Automatic geom
29. e based on the inner and outer pick imis AIF plots all the picks nstructions about modeling Node oscilator drift and for Trim amp seed with GroupXY parameters y be scrolled Output can be highlighted with the cursor copied ctri C and pasted elsewhere or use St Receiver activities Pje s i Least squares parameters 8 Picka priori SD in gu E 0 1 Tolerance per Rx in gu Select r E receiver ID 100 Pick inner limit ms header me m bins Si C e eee mel constrain depth Stop selected __tZend __ Teese 2 i m processes Trace high pass low cut fiter Fiteriength gt 1 Close all plots Enisotropy angularity Areal pick plots amp interval Balance azimuth p All 40 Some www T Balance distance SourceMeasurementUnit Node oscillator drift day microseconds Screen scroll span lines T Trim amp seed with Groupxy 0 Screen gt File 200 Clear screen 5000 Max range in gu Meter foot or arc second q EJ HmFBA_v1b103 Hydrometronics LLC HmFBA v1b103 Wide Azimuth Far Offset First Break OBC OBN Positioning Hydrometronics Wide Azimuth Far Offset First Break Positioning Output will dis 1 Begin by pushing the Show manual button This will display the legal no spa 2 If a GUI configuration for your prospect has been saved you may read it LS param eters pick S D convergence 3 HmFBA processes SEG Y and Seismic Unix SU receiver gathers Pus 4 The Screen gt File button wil
30. ents at the end for more details about the survey The following X Y all at once plots are HmFBA configuration Plan view Chebyshev regression equation Numerically computed velocity Pick residuals versus pick time Histogram of pick residuals Swath output statistics 28 EJ HmFBA_v1b103 Hydrometronics LLC HmFBA v1b103 Wide Azimuth Far Offset First Break OBC OBN Positioning Elapsed time for simultaneous adjustment 3 874 RxD grid_coordinstes UV SDX SDY used_picks HDOP geogrphicals_if_arc sec 420427 43 530367232 048 263 2 17 747 0 082 47 8814507 129 0642908 421530 43 5308181 74 063 2 70 2 54 809 0 078 47 9221525 129 0503610 423740 95 531361210 048 2 24 2 01 924 0 073 47 9712699 129 0217393 415779 39 5302468 18 0 88 3 09 2 97 822 0 076 47 8700257 129 1262119 417371 46 5307229 39 Bh 4 3 60 3 41 939 0 072 47 9130628 129 1058382 419017 39 5311962 95 1 84 417 3 73 1030 0 069 47 9558544 129 0847047 420639 67 531670458 086 294 262 987 0 071 47 9987117 129 0638554 412842 03 5306238 66 1 63 445 429 775 0 081 47 9035511 129 1662438 414304 90 5311031 64 0 77 284 271 878 0 075 47 9468634 129 1476269 10 415848 64 5315863 11 1 51 3 92 3 46 974 0 071 47 9905296 129 1279034 UV_total total_pi selected_picks used_picks used TRX H a a a a g X Y all at once swath processing results Processing done results follow nr Grid units in arc seconds with a Transverse Mercator Central Meridian of 128 Note the pick a priori SD of 60 meters Picking
31. eplacing the last seed position with the latest coordinates until convergence that is until the change from one iteration to the next is less than some tolerance See observation equation Linear describes an equation or an expression in which all variables are of degree 1 that is no higher powers or transcendentals Linearization See observation equation LOESS is an unweighted version of LOWESS which is locally weighted scatter plot smoothing basically a rolling quadratic used as a smoother of time series data Measurement is the physical process of determining the value of a quantity such as a distance or angle or time Also called an observation All measurements have error Multicollinearity also collinearity is a statistical phenomenon in which two or more predictor variables in a multiple regression model are highly correlated meaning that one can be linearly predicted from the others with a non trivial degree of accuracy Wikipedia Non centrality See Tau Normal or Gaussian distribution is the bell shaped probability distribution that describes most random errors It is characterized by a mean and a variance Named after the mathematician Karl Friedrich Gauss 1777 1855 Normal matrix and inverse normal matrix The normal matrix is a product of a least squares adjustment It is the transpose of the design matrix times the design matrix There may be weighting too See design matrix which leads you to observa
32. ers for your prospect but HmFBA offers three kinds of QC plots to help Pick samples versus chronological sequence Seismic trace plots with picks Pick sample areal contour plots all picks or some i e only those between the specified inner and outer limits can be created The next three plots in this manual exhibit these QC plots from an example ION Geophysical prospect See Acknowledgements Additionally a plan view of the shot lines is shown with the innermost direct water arrival picks edited out Least squares adjustments after picking are quick and can be rapidly repeated with different parameters to judge which picks work best in the least squares adjustment Least squares adjustments provide their own quality metrics that by inference help guide the picking First break picks T Pick samples versus chronological sequence Some poor pickin MeL ye poor picking aS dN L L 1 1000 2000 3000 4000 5000 6000 7000 8000 9000 10000 Sequence of source events 15 Same prospect with excellent picks out to
33. etry balancing works by decimating azimuth sectors and or offset rings you can choose one or the other or both that are overpopulated with respect to the average In order to achieve a uniformly balanced geometry the picks in the overpopulated rings or sectors are decimated randomly Therefore if automatic geometry balancing is chosen every solution is unique It cannot be repeated because the observation set will be different the next time the adjustment is run Nevertheless the results will be consistently within the error bars i e uncertainties SDs reported by the application 2 D Mode Processing 4 Comments on Snell s Law Snell s Law states that the ratio of the sines of the angles of incidence and refraction is equal to the ratio of the velocities of the respective media When the refractive angle is 90 that is when the seismic energy travels along the boundary of the media then the critical angle of incidence is asin v1 v2 If the angle between the source and the receiver is more than the critical angle then direct water energy arrives after the refracted energy which may be weaker than the later direct arrival V1 for water is about 1 5m ms V2 for shale is about 2 0m ms V2 for igneous rock is about 5 5m ms Therefore the critical angles are about 49 for shale and 16 for igneous In 1000m of water the farthest horizontal distance from the receiver to assure that energy through the water arrives first is
34. hen distant sources present with near zero picks as often happens with large far offset files Those picks will be edited out as unreasonable without polluting the adjustment If you want seeding only chose a large Max range in gu If you want trimming and seeding then choose a reasonable Max range in gu You cannot have trimming without seeding LS residuals as function of azimuth T T T T T Signature of anisotropy or angularity 200 150 100 grid units LS residuals as function of azimuth T Afte anisotropy angularity compensation 7 5 4 200 150 100 3 D Mode Parameters Certain parameters are operative only in 3 D mode water These are VP in gu ms velocity of propagation or slope Bias in ms static offset or intercept Automatic VP bias Constrain depth If Automatic VP bias and Constrain depth are not selected then the adjustment is done with the VP and bias entered The adjustment does report a slope and intercept that serve as a guideline for adjusting VP and bias manually if desired Automatic VP bias computes the VP slope and bias intercept automatically with or without the constraint of depth but restraining depth gives best results Constrain depth constrains the entered depth found among the least squares parameters as an observation with an SD of 20cm in appropriate m units but this is
35. iance matrix to the orientation at which the covariances become zero Snell s Law states that the ratio of the sines of the angles of incidence and refraction is equal to the ratio of the velocities of the respective media 46 Standard deviation or Sigma o Standard deviation is the square root of the variance Sigma is the lower case Greek letter o that is generally used to represent the standard deviation See variance Standard deviation of unit weight is the square root of the unit variance UV often reported as SDO or o0 A stochastic process is one in which the effect is randomly related to the cause in some non deterministic way that can only be described probabilistically See deterministic Systematic error See bias Tau Tau Method non centrality Tau is an obscure probability distribution that for large degrees of freedom is extremely close to the normal distribution but which differs for low degrees of freedom The Tau Method is an outlier rejection scheme developed by Allen J Pope an American geodesist in the 1970s See Acknowledgements for a link to his paper The Tau Method is an alternative to the Delft Method developed by W Baarda a Dutch geodesist in the 1960s The non centrality parameter is the number of tau statistics to use for outlier rejection Since HmFBA adjustments typically enjoy high degrees of freedom one tau statistic is about the same as one normal distribution standard deviation In HmFBA a t
36. ion and solved for i e calibrated Also called systematic error Blunder or Outlier or Spike A blunder is a mistake that is an observation occurring outside of the expected probability distribution An example in surveying might be using the wrong back sight Other examples might be an acoustic reflection or erroneous data communication 43 Chebyshev regression equation is a mathematical expression of the form y aoTo Z a T4 Z a T Z where ao ay a are empirically determined coefficients where T z 1 T z z and T Z 2zT z T z and where z x min x max x x max x min x The regression order is the highest positive integer power in the equation These Chebyshev terms of the first kind T are orthogonal in the domain 1 to 1 thus the compression of x into z This orthogonality eliminates the multicollinearity of normal polynomial regression and thus is an improvement over previous methods The x s are pick times and the y s are distances corresponding to the picks C O is computed minus observed another expression for residual Convergence See least squares adjustment Correlation is a measure of the statistical dependence between variables A correlation coefficient is the covariance divided by the product of the associated standard deviations varying between 1 and 1 where 1 is complete positive dependence 1 is complete negative dependence and 0 is no depende
37. ion stops when Maximum iterations is reached Tau non centrality is the number of tau statistics used for outlier detection and elimination For high degrees of freedom tau is close to the normal statistic A non centrality of 2 is aggressive trims about 5 of the data 3 is relaxed trims lt 0 3 of the data and 4 trims only the worst outliers Use 100 if you don t want outlier rejection See Acknowledgements and Glossary for more on the Tau Method Relative receiver depth in gu difference between gun and receiver depth in grid units is used for computing slant ranges in 2 D mode refractor amp water though slant ranges are compensated by the regression equation so accuracy is not critical Least Squares Parameters 2 Anisotropy angularity is either on or off Anisotropy is variation in seismic velocity as a function of direction or travel Angularity is a directional source array effect Both can sometimes be seen in the least squares adjustment plot LS residuals as a function of azimuth as a red wavy line of the mean residuals with two peaks and two troughs over 360 degrees Selecting anisotropy angularity complements the picks with the diminishing mean residual as a function of azimuth until the red line is nearly flat Of course this will have some effect on the coordinates computed so use anisotropy angularity prudently The next two plots offer examples of a prospect with anisotrop
38. ions follow 1 Begin by pushing the Show manual button This will display the legal notice and the manual detailed instructions and enable activities 2 If a GUI configuration for your prospect has been saved you may read it now with the Read button Later you can save the current configuration with Save 3 HmFBA processes SEG Y and Seismic Unix SU receiver gathers Push Load gather for swath for an entire swath Three pick methods and parameters Single node receiver activities Load Pick Adjust X Y coordinates of the gather Load and Save Read and save GUI configuration its AIF plots all the picks d with GroupXY and pasted elsewhere or use Swath activities Load pick swath Trace high pass filter before picking cdp SourceMeasurementUnit Node oscillator drift day microseconds 0 Pick methods amp parameters Absolute amplitude Energy 12 Sample length 15 Threshold max this Gradient threshold 2 Trace sample rate 2 milliseconds Trace high pass low cut filter I Fitter length gt 11 Areal pick plots amp interval a Ta s GUI configuration Read Save Mode 2 D or 3D 20 C30 EEn BA E Automatic VP bias M co th Close all plots www Screen scroll span lines Screen gt File 200 Clear screen Least squares parameters Pick a priori SD in gu 0 1 Tolerance per Rx in gu 4 Regression
39. irst break pick time ms to horizontal distance in gu is refreshed at every Iteration of the least square adjustment The red Chebyshev regression line relates a pick time to a horizontal distance for the adjustment of the receiver s position Note that the Chebyshev regression equation curve does not start at 0 0 This static offset near the origin accounts for instrumental delays on average and picking bias on average The numerical differential of the regression line estimates velocity in the refractor First breaks are seismic data Skill and interpretation are as important as science in processing seismic data We ve looked at pick residuals The next three plots are least squares LS residuals LS residuals as functions of azimuth LS residuals as functions of distance LS residuals as histogram 24 grid units grid units residual LS residuals as function of azimuth Suitably random over the entire abscissa The red line is the Fourier series mean which does not have the signature of anisotropy or angularity but which can be made zero with comperisation 200 150 100 50 degrees LS residuals as function of distance Suitably random over the entire abscissa Mean as a function of distance is always zero due to the regression equation 1500 2000 2500 grid units distance 25 Histogram of pick residuals Should ideally be a normal distribution Some bi modality here which can
40. is required for HmFBA HmFBA was not tested on 64 bit Vista or Windows 8 or 10 but expect that it will run on those OSs XP is not recommended HmFBA was not tested on Intel i3 or is CPUs or any equivalent AMD CPU but expect that it will run on them 4GB of RAM may be adequate for small gathers but at least 8GB and as many as 48GB may be required for production depending upon the size of the gathers thus the requirement for 64 bits HmFBA is provided with a KEYLOK III blue security dongle which enables you to run HmFBA on any Windows computer The application may be copied freely The blue KEYLOK III dongle does not require the installation of drivers finding them in the Windows OS Demonstration versions of HmMFBA will have time limited dongles 41 Blue Security Dongle Contents of the Dongle driver and utility folder of the USB drive on which HmFBA was provided VerifyKey exe checks for proper dongle installation USBKey64 sys driver for 64 bit CPU required by HmFBA The blue security dongle installs its drivers automatically upon installation Wait for the process to complete The utility VerifyKey exe will confirm proper installation of the driver Troubleshooting The driver itself is located in the dongle driver and utility folder To manually install the driver use Control Panel gt System gt Device manager gt double click on USBKey or USB Dongle gt Driver gt Update Driver gt browse
41. iver gathers in SEG Y or Seismic Unix SU binary format and uses the selected pick method pick parameters and optional high pass filter to pick the first breaks and save them to memory X Y sequential processes the swath of receivers in 2 D mode refractor amp water or 3 D mode water one by one using different regression coefficients for every receiver and adjusts the coordinates using the least squares parameters selected X Y all at once processes the swath of receivers only in 2 D mode refractor amp water simultaneously using the same regression coefficients for the entire swath and adjusts the swath using the least squares parameters selected Save saves the swath of picks in memory to a CSV file for later loading into HmFBA or for analysis outside of HmFBA Load loads a previously saved swath of picks in CSV format into memory Load pick swath can be time consuming but Load is quick Therefore always Save your picked swath Load Save CSV format in Appendix 2 Pick Methods and Parameters Absolute amplitude requires sample length and a divisor called this to determine the threshold The mean absolute value of the 10 largest trace amplitudes is determined called max The threshold is max this The mean absolute amplitudes of a rolling sample length are determined When the threshold is exceeded the pick is the mean of the current rolling sample length
42. le in milliseconds 2 Picking by energy sample length 5 threshold 20 high pass yes with length 11 3 D mode parameters VP 1 5 and bias 0 and auto VP bias yes and constrain depth no LS parameters pick SD 8 tolerance 0 1 order 4 inner 1300 outer 2150 max iter 50 tau 3 5 depth 2100 mode 3 D balancing no anisotropy angularity correction no drift 0 Easting 661987 2566 Northing 2925837 1434 Depth 2096 2881 Total picks 314241 Selected picks 10831 Used picks 10315 Percentage used 95 2359 Unit variance 0 26443 SD of unit weight 0 51423 PDOP 0 03452 Scaled semi major 0 093803 Scaled semi minor 0 093521 Orientation deg 20 7089 Scaled dRMS 0 13246 Scaled SD in depth 0 0512 Slope VP in gu ms 1 4947 Intercept bias in gu 28 8619 Intercept bias in ms 19 3099 Time seconds processing 0 16973 iteration 6 position jump in grid unit 0 076186 Rapid convergence iteration 1 position jump in grid unit 1576 896 Processing begins 3 D mode plan view in 3D Can i i be rotated to suit the viewer The Prospect 2000 2 928 N gt 2 97 N E A x 10 2 926 NS ae 2 925 N An iiie 6 61 2 924 Northing Easting 35 Pick range Residuals C O Regrettably skewed 8 Number in bin S 5 grid units 3 8S B Number of picks 8 2200 2400 2600 2800 3000 Distance gu 3200 36
43. lot gt Plot sample rate 2 milliseconds Meter foot or arc second coordinates Trace high pass low cut filter 00 Relative depth in gu meter foot C arc sec 33 CM PM Fiter length gt 41 Close all plots I Anisotropy angularity Receiver ID a Areal pick plots amp interval T Balance azimuth ransductionConstantMantissa Stop TransductionConstantPower All 40 Some www Balance distance gt Node oscillator drift day microseconds Screen scroll span lines T Trim amp seed with Groupxy 0 Screen gt File 200 Clear screen 9000 Max range in gu Tau non centrality 19 _49 Receiver gather before processing 2 84 D pa O ai 2 82 2 83 2 84 Sources used with OBS in red 20 Distance histogram 40 bins SD2Mean 0 19623 g Distance geometry More picks with distance but the number needs to increase linearly as it does 8 8 g Number of picks 8 1500 2000 Distance Azimuth histogram 10 degree bins SD2Mean 0 16786 140 2 g B Number of picks gt Azimuth geometry 50 100 Azimuth 21 Inversed distance in grid units 2 D Mode Processing 6 Notice that the gather has shrunk through the use of pick limits An inner limit can eliminate the direct water arrivals as determined by depth and Snell s Law An outer limit or creative picking can be used to eliminate refracted arrivals for water only processing Decisions
44. n no drift 0 UV 1 1822 Scaled SDX 0 53816 Scaled SDY 0 5338 HDOP 0 034857 20 bin distance SD mean ratio 0 15448 40 bin distance SD mean ratio 0 19623 10 degree azimuth SD mean ratio 0 16786 20 degree azimuth SD mean ratio 0 1192 Used picks 3306 Selected picks 3337 Total picks 6931 Receiver coordinates 285219 68627 7437308 0728 8 iterations are mandatory Receiver ID 0 in 2D mode refractor amp Time seconds processing 0 6028 water iteration 8 position jump in grid unit 0 053007 iteration 7 position jump in grid unit 0 037679 Read bottom to top iteration 6 position jump in grid unit 0 25729 5 iteration 5 position jump in grid unit 0 23565 Notice the configuration iteration 4 position jump in grid unit 38 9437 parameters in the red box iteration 3 position jump in grid unit 163 4053 at the top iteration 2 position jump in grid unit 703 808 iteration 1 position jump in grid unit 607 5953 See glossary for Processing begins definitions 2 D Mode Processing 9 The least squares adjustment can be controlled by the LS parameters pick SD convergence tolerance regression order if in 2 D mode inner and outer pick limits maximum number of iterations tau non centrality depth relative to source anisotropy angularity geometry balancing trimming amp seeding If you increase the pick SD the unit variance UV will compensate by going down and vi
45. nce at all that is completely random Covariance is a measure of the linked variation of the two random variables It is a product of the inverse normal matrix See normal matrix CSV Comma separated value Degrees of freedom DoF are the number knowns observations minus the number of unknowns coordinates or parameters in an adjustment Also called redundancy Design matrix See observation equation Deterministic A deterministic process is one in which no randomness is involved in the development of future states of the process that is it will always produce the same output from a given starting condition Compare stochastic DOP is Dilution of Precision a measure of adjustment geometry HDOP horizontal is 2D and PDOP positional is 3D DRMS is Distance Root Mean Square or radial error the square root of the sum of the variances in the X and Y axes See normal matrix An error can be a blunder a bias or a random error gu grid unit The unit meter or foot of the map projection of the source coordinates Inverse normal matrix See normal matrix Iteration See least squares adjustment 44 Least squares LS adjustment is an algorithm for adjusting systems of observation equations by finding the minimum value for the sum of the squares of the residuals Because observation equations are often linearized the adjustment begins with a seed value for the coordinates and iterates repeats the adjustment by r
46. offset seismic data may arrive horizontally through one or more refractors These refracted data are subject to geological velocity gradients that must be calibrated They are in HmFBA with a Chebyshev regression equation for a single node or an entire swath and with anisotropy angularity compensation It s alla matter of statistics With a crude observable like a first break the statistics are in your favor with all the data in a wide azimuth far offset receiver gather And the outliers are easy to clean up with all those data too HmFBA will process one receiver gather at a time while providing copious QC graphics and statistics for the analysis of the best parameters or process a swath of receiver gathers both sequentially and simultaneously which provides added benefits Automatic geometry balancing in HmFBA is an effective way to achieve excellent results on a receiver by receiver basis 38 alco Acknowledgements 1 Thanks to FairfieldNodal and to ION Geophysical for permission to exhibit in this manual HmFBA plots derived from their data The facilities of IRIS Data Services and specifically the IRIS Data Management Center were used for access to some waveforms related metadata and or derived products used in testing HmMFBA IRIS Data Services are funded through the Seismological Facilities for the Advancement of Geoscience and EarthScope SAGE Proposal of the National Science Foundation under Cooperative Agreement EAR 126
47. on the right limits can be based upon the histograms previously shown discussed next and the least squares statistics discussed later In addition to histogram presentation do they look balanced histograms have SD2Mean statistics the ratio of the SD of the bin variations to the mean or regressed bin size that should be as small as possible SD2Mean lt 0 3 is OK especially for the finer bins 40 bins for distance and 10 degrees for azimuth but not essential Get the best you can get manually or balance geometry automatically After achieving a good balance of distance and azimuth manually or automatically turn your attention to the first break picks The next four plots show The picks in ms versus the distance from the source to the mean source before processing The picks in ms versus the distance from the source to the adjusted receiver position after processing with Chebyshev regression equation in red The slope of the Chebyshev regression equation as velocity in 2 D mode Another Chebyshev regression equation with longer offset for comparison 2 D Mode Processing continues after the plots Picks before processing T Pick times ms versus source to mean source distance before processing A LAG s PER J WY A The shoddiness of this plot is due to inaccurate pre processing OBS position actually the source barycenter unless Seed with GroupXY is chosen 1 l 1 1 1 1 500 1000 1500 2000
48. order lt Pick inner limit ms Close all plots I Anisotropy angularity I Balance azimuth J Balance distance I Trim amp seed with Groupxy S000 Max range in gu EJ HmFBA_vib103 Hydrometronics LLC HmFBA v1b103 Wide Azimuth Far Offset First Break OBC OBN Positioning asap Wide Azimuth Far Offset First Break Positioning Output wil display here Brief instructions folow display the legal notice and the manual detailed instructions and enable activities may read it now with the Read button Later you can save the current configuration with Save ers Push Load gather for a single gather or Load pick swath for an entire swath Swath a activities Load Clear screen button will clear the screen if it becomes jumbled due to word wrap i i Ch desired pick method and ramet Swath loadir ind picking happen ith Load pick Pick swath Adjust X Y nytrsees as serce win pers See manusi fr more oration about acing pia coordinates sequentially fs ingie gatner resuting in coordinates plus several piots an7 nd Bias as necessary compute them automatically or con H a or all at once Load and foresrssceanw enaar poces t vrare 27D OF 3 D mode ically balanced in azimuth and or distance by checking the 4 Save picks ctively as a CSV file for later loading with the Load butto processing and one at time and X Y all at once will process all the receive 3 D mode Gur level The some plots ar
49. ow frequency high amplitude noise which may be swell in the 2500m water column 7000 Biter asi et pis iis ys Fis TERELID EX UH S IVIL If 1h J 1575 1580 1585 1590 1595 1600 1605 1610 1615 1620 1625 Sequence of source events 12 Traces with picks superimposed w Q Pick sample gt The same traces after high pass filter enabling better picking woo E 1575 1580 1585 1590 1595 1600 1605 1610 1615 1620 1625 Sequence of source events Other Controls 4 Plot seismic traces with picks The previous two pages exhibit plots made with this control Samples and traces are selected with Matlab syntax 1000 3000 is every trace or sample from 1000 to 3000 1000 2 3000 is every even trace or sample from 1000 to 3000 1001 2 3000 is every odd trace or sample from 1001 to 2999 An amp of 1 means that the maximum trace amplitude will occupy the division between the traces An amp of 2 means that the maximum amplitude will occupy twice the division between the traces and so on for amp values less than 1 too After picking a gather the number of traces and samples are shown in the output screen for reference and written to the samples and traces window Remember these numbers They can be found again by scrolling the output screen Picks are plotted on the traces as QC of their quality The seismic trace plot can be zoomed for more de
50. path Another way to add the path to use the System Properties dialog box Open Control Panel gt Performance and Maintenance gt System In the box that opens click the Advanced tab to obtain the dialog box Click the button Environment Variables The dialog box lists variables that apply only to the current user and those that apply to the whole system Add a path to the MCR as above Finally using the command prompt PATH can be appended by the command path path path_to_MCR Appending the path this way lasts only until reboot Better to use one of the previous methods If the MCR path needs to be added a reboot may be required Appendix 5 Glossary of Terms Adjust Adjustment Correct s observations to compensate for random error The least squares criterion is that the sum of the squares of the corrections residuals be minimum See least squares adjustment Angularity Variation in seismic energy onset as a function of source array geometry and direction or travel Anisotropy Variation in seismic velocity as a function of direction or travel A posteriori is Latin for from what comes later that is statistical values determined after an adjustment based on posterior experience A priori is Latin for from what precedes that is statistical values assumed before an adjustment based on prior knowledge Bias is a deviation from the truth in some systematic way that can be written into an observation equat
51. phicals_if_arc sec 420427 43 5303672 32 0 48 2 63 2 17 747 0 082 47 8814507 129 0642908 421530 43 5308181 74 0 63 2 70 2 54 809 0 078 47 9221525 129 0503610 423740 95 5313612 10 0 48 2 24 2 01 924 0 073 47 9712699 129 0217393 415779 39 5302468 18 0 88 3 09 2 97 822 0 076 47 8700257 129 1262119 417371 46 5307229 39 1 32 3 60 3 41 939 0 072 47 9130628 129 1058382 419017 39 5311962 95 1 84 4 17 3 73 1030 0 069 47 9558544 129 0847047 420639 67 5316704 58 0 86 2 94 2 62 987 0 071 47 9987117 129 0638554 412842 03 5306238 66 1 63 4 45 4 29 775 0 081 47 9035511 129 1662438 414304 90 5311031 64 0 77 2 84 2 71 878 0 075 47 9468634 129 1476269 415848 64 5315863 11 1 51 3 92 3 46 974 0 071 47 9905296 129 1279034 UV_total total_picks selected_picks used_picks used_per_Rx 1 51370894 55670 8915 8884 888 4 1 Processing done results follow Configuration parameters Grid units in arc seconds with a Transverse Mercator Central Meridian of 128 Picking by energy sample length 5 threshold 20 high pass yes with length 11 LS parameters pick SD 60 tolerance 0 4 order 4 inner 2000 outer 7000 max iter 50 tau 4 depth 2500 mode 2 D balancing no anisotropy angularity correction no Iteration jump used picks elapsed time NB jump is the sum of the squares of the receiver position shifts Time seconds processing 0 71412 8 0 016503362 8884 0 67442209 SEMVNONEWN 0 02725147 8914 0 32697193 Elapsed
52. r distance by checking the appropriate boxes 12 The Save buttons will save a picked swath or gather respectively as a CSV file for later loading with the Load buttons See manual for format 13 X Y sequential will process all the receivers sequentially one at time and X Y all at once will process all the receivers simultaneously 14 Check anisotropy angularity to compensate for these effects by leveling the pick residuals as a function of azimuth 15 Set the areal plot contour level The some plots are based on the inner and outer pick limits AIF plots all the picks 16 Show manuaf for instructions about modeling Node oscillator drift and for Trim amp seed with Gro Note This screen may be scrolled Output can be highlighted with the cursor copied ctr C and pasted elsewhere or use Screen gt File button Receiver activities Load gather x a Pick Plot all at e gather Load ave oad Save Plot seismic traces with picks Tend lt Samples amp gt 7 1 Z end lt Traces Plot gt Piot Meter foot or arc second coordinates meter C foot C arc sec 33 CM Swath activities Receiver ID cdp SourceMeasurementUnit Node oscillator drift day microseconds 0 Pick methods amp parameters C Absolute amplitude GUI configuration Read Save Mode 2 D or 3D 2D C30 12 Sample length 15 Threshold max this Energy C Gradient threshold 2 Trace sample rate
53. reen ontrol the has Gasca aan sas number of text lines the screen scrolls lt Samples amp gt 1 air Sep imum iterations lt Traces Plot gt Plot 2 4 Tau non centrality Sane eee cease C arcsec 33 CM Close all plots I Anisotropy angularity ieee Balance azimuth ees p I Balance distance SourceMeasurement nit i Node oscillator drift day microseconds Screen scroll span lines D Trima O o 200 Clear screen S009 Max range in gu Receiver Activities Single Node Load gather loads a previously prepared receiver gather of seismic data in SEG Y or Seismic Unix SU binary format Pick uses the selected method parameters trace sample rate and optional high pass filter to pick and plot the first breaks X Y gather processes the picks of a single gather as observations and adjusts the receiver coordinates using the selected least squares parameters including geometry balancing in either 3 D mode water or 2 D mode refractor amp water Adjustment can be repeated any number of times with the current picks in order to get the parameters right perhaps for later swath processing Load loads a CSV file of a previously picked gather Save saves the picks to a CSV file This can be useful if the gather is large and loading and picking are slow Load Save CSV format in Appendix 2 Swath Activities Multiple Nodes Load pick swath loads a swath of seismic rece
54. tail or select fewer traces and or samples This control works only for single node not swath gathers Automatic Gain Control AGC is not offered If you have to use AGC to see a refractor you probably can t pick it automatically 13 Other Controls 5 Node oscillator drift day microseconds This feature compensates for in situ oscillator drift in an ocean bottom node which can be identified by systematic shifts in first break positions if the wide azimuth far offset seismic picks are segmented on offset and by other ways beyond the scope of this manual This feature uses the relative times of the seismic traces in the SEG Y or SU data If positive it adds the drift to the earliest picks as a function of time and subtracts the drift from the latest picks as a function of time If negative the reverse This feature will change coordinates so use it with great caution This feature applies only to receiver activities single node and not to swath activities multiple nodes because multiple node will have different drifts This feature applies to both 2 D and 3 D modes receiver activities The default drift is zero 0 that is no drift applied 2 D Mode Processing 14 Pick sample 2 D Mode Processing 1 Picking Quality Control HmFBA offers three very different picking methods absolute amplitude energy and gradient It can be daunting to choose the right pick method and paramet
55. time for bookeeping and first regression 0 23694 Elapsed time for bookeeping 0 17665 Process begins please wait 3 D Mode Processing 32 3 D Mode Processing 1 Wide azimuth far offset 2 D mode processing is the strength of HmFBA but the application offers a full complement of features for 3 D mode processing of direct water arrivals no refractor involvement as a single node X Y gather or in a swath X Y sequential See earlier in the manual for more information See also the previous comments on Snell s Law to assess your offset geometry and learn whether you re getting only water arrivals HmFBA in 3 D mode will of course process refracted arrivals too but refracted geometry is different more bent than a water arrival Including refracted arrivals in a 3 D mode adjustment can skew results Although there are clues in the output graphics in the velocity plot and in the seismic data plot the HmFBA adjustment doesn t know if you re including refracted arrivals or not In the years before GNSS dominated land surveying and maybe still to some extent electronic distance measuring EDM equipment needed to be calibrated for both a scale error velocity of propagation or VP anda static offset bias A first break is also subject to VP and bias Therefore the main 3 D mode controls allow for a manual input of VP and bias VP can be measured empirically in your prospect with an ocean probe but bias e g
56. tion equation The inverse normal matrix is also a product of a least squares adjustment It is the inverse of the normal matrix It is also called the variance covariance matrix of the coordinates The diagonal terms are the variances of the coordinates The off diagonal terms are the covariances of the coordinates The square root of the trace of the inverse normal matrix is the DRMS OBC OBN OBS Ocean Bottom Cable Node Seismometer Observation In the context of HmFBA an observation is a positioning measurement typically a first break 45 An observation equation expresses an observation in terms of the knowns and unknowns The classic observation equation is that for an observed range in terms of known source coordinates s and unknown receiver coordinates r namely Range Xs Xr 2 Ys Yr 42 40 5 This is a non linear equation that is the powers of the unknowns are greater than first order or unity 1 To be used in a least squares adjustment it must be linearized by using the first order terms of a Taylor s series expansion of the observation equation not discussed further The coefficients of the first order terms of a Taylor s series expansion comprise the elements of the design matrix Outlier See blunder Precision sometimes called resolution is the consistency of a time series of observations or the coordinates derived from those observations blunders and biases having been removed Probability is
57. to the driver in the dongle drivers and utilities folder MCR Installation HmFBA is complied Matlab software that requires the installation of the Matlab Compiler Runtime MCR The MCRiInstaller supplied by The MathWorks for free and without royalty is large because it will support all of Matlab on your computer The MCR is like the NET framework for Visual Studio languages or the Java Virtual Machine JVM for Java The MCR supports compiled Matlab programs The installer can be found on the supplied USB drive or at http www mathworks com products compiler mcr index html Copy or download the 64bit Windows version for Matlab Release 2015a v8 5 to the target machine Execute the MCR installer Place the HmFBA executable in the desired folder Execute by double clicking This will launch splash png which can be any splash screen you desire by this name even your company logo 42 Troubleshooting the MCR Installation If the MCR is not seen add MCR path to the PATH variable within Environment variables One way to do that is Right Click on My Computer gt Properties gt Advanced System Settings gt Click on Environment Variables In the System Variables dialog box click on Path variable and add the MCR path to it which is typically C Program Files MATLAB MATLAB Compiler Runtime v83 runtime win64 for a 64 bit Windows system Check first to see where the MCR is located then copy that
58. tp ensiklopediseismik blogspot com 2014 05 first break picker html 39 Appendix 2 Saved file format 668734 668754 668773 668792 668811 668830 668849 668868 661987 2920570 4547 19614713 1416629008 2920602 4553 19614713 1416653252 2920634 4559 19614713 1416677496 2920666 4565 19614713 1416701740 2920699 4571 19614713 1416725984 2920731 4577 19614713 1416750228 2920763 4583 19614713 1416774472 2920796 4589 19614713 1416798716 2925837 0 19614713 0 OU uw Se ee a Co l wo Sample format above Column 1 is pick samples not ms or 99 for GroupXY Column 2 is source or group Easting or longitude Column 3 is source or group Northing or latitude Column 4 is source ID Column 5 is receiver ID Column 6 is milliseconds since the first trace Appendix 3 Matlab Plot Controls Expand to full screen Save in 16 formats Print Zoom out 9 9 Zoom in en He 9 9 Pan Rotate 3D figure is gt y 6 A Y m Mark coordinates Reset to original view right click with cursor in plot Appendix 4 Hardware Software and Security Requirements Hardware Software and Security HmFBA was developed using a 64 bit version of Matlab R201 4a on 64 bit Windows 7 on a 64 bit quad core Intel Xeon CPU with 16 and later 48GB RAM and tested on a quad core Intel i7 with 8GB RAM and a dual core Intel i7 with 4GB of RAM A 64 bit CPU and 64 bit Windows
59. vide strong guidance about which is the most successful 33 3 D Mode Processing 3 Our 3 D mode example derived from FairfieldNodal data in deep water 2100m See Acknowledgements VP and bias are computed automatically possible because there are lots of data The next two slides exhibit the configuration and 3 D mode output statistics The next slide is a plan view in 3D Then follow three histograms of the pick residuals and the distance and azimuth geometry balance Finally pick time versus offset distance is shown E HmFBA_vib103 Hydrometronics LLC HmFBA v1b103 Wide Azimuth Far Offset First Break OBC OBN Positioning Picking by energy sample length 5 threshold 20 high pass yes with length 11 3 D mode parameters VP 1 5 and bias 0 and auto VP bias yes and constrain depth no LS parameters pick SD 8 tolerance 0 1 order 4 inner 1300 outer 2150 max iter 50 tau 3 5 depth 2100 mode 3 D balancing no anisotropy angularity correction no drift 0 Grid units in meters picking sample in milliseconds 2 a Easting 661987 2566 Northing 2925837 1434 Depth 2096 2881 Total picks 314241 Selected picks 10831 Used picks 10315 Percentage used 95 2359 Unit variance 0 26443 SD of unit weight 0 51423 PDOP 0 03452 VP and bias are com puted Scaled semi major 0 093803 Scaled semi minor 0 093521 Orientation deg 20 7089 Scaled dRMS 0 13246 S
60. y angularity before and after compensation Balance azimuth and Balance distance randomly decimates picks from overpopulated azimuth sectors and or distance rings in order to achieve better balanced geometry for the purpose of solving for biases and velocity gradients in either 3 D mode water or 2 D mode refractor amp water If automatic geometry balancing is chosen every solution is unique It cannot be repeated because the observation set will be different the next time the adjustment is run Nevertheless the results will be consistently within the error bars i e uncertainties SDs reported by the application grid units Least Squares Parameters 3 Trim amp seed with GroupXY offers some special features Without checking this feature the seeding of the iterative least squares adjustment is with the barycenter of the sources selected with the inner and outer pick limits Usually this works just fine By checking this feature the seeding of the adjustment is with the GroupX and GroupY coordinates in the SEG Y or SU files This may speed up the iteration or the GroupXY coordinates may be just wrong in which case you don t want to do it Another benefit of checking Trim amp seed with GroupXY if the GroupXY coordinates are correct is that the picks are trimmed on distance i e within the limits in the Max range in gu edit box This offers a distinct advantage when picking is poor i e w

"HmFBA User Manual"

Contents

Download Pdf Manuals

Related Search

Related Contents

&quot;HmFBA User Manual&quot;

Contents

Download Pdf Manuals

Related Search

Related Contents

"HmFBA User Manual"