User Manual for M8 - Indico
Contents
1. Mathematicians and geophysicists at the International Institute for Earthquake Prediction Theory and Mathematical Geophysics in Moscow have applied methods of pattern recognition to the problem of earthquake prediction 2 V I Keilis Borok and V G Kossobokov developed the M8 algorithm which has successfully predicted several strong earthquakes 3 Earthquake predictions from this algorithm have been presented to the U S National Earthquake Prediction Evaluation Council NEPEC 4 and NEPEC has recommended that the USGS undertake an evaluation of this algorithm After successful prediction of the Loma Prieta 1989 earthquake J H Healy V G Kossobokov and J W Dewey 5 designed a rigid test to evaluate the M8 algorithm Since 1991 each half year the algorithm has been applied in a real time prediction mode to the seismicity of the entire Circum Pacific seismic belt The experiment is set to predict magnitude 7 5 earthquakes and we expect that in five years there will be more than ten of them in the territory considered This publication intends to facilitate communication and better understanding of the intermediate term earthquake prediction method as it provides ultimate description of the algorithm i e a program with a source code and a set of fixed parameters M8 is a computer program written in Fortran 77 that allows for the intermediate term prediction of earthquakes using the M8 algorithm as well as the evaluation of a time seri2. see answers to Q 4 Comment Before this question the block No 5 concerning the constants for diagnosis of TIP is displayed The values of pb and p the displayed values will be used Comment pb and p are used in definition of extremely large values of functions The value of the number Nex of steps dt in a time window for the definition of simultaneous extrema the displayed value of Nex will be used The values of the thresholds for the number of groups of functions with extremely large values G and the number of such functions H the displayed values of G and H will be used The value of the number Nau of steps dt in the duration of a TIP the displayed value of Nau will be used y the table of values of the functions considered will be written in the file for printing m8 pri n the table will not be written in the file y goes to Q 45 n the program starts calculations The name of profile in which the data will be written 25 Running the batch version of M8 In some cases it might be more convinient not to go through interactive dialog but run M8 from the command line For this purpose we provide the user with M8_BAT EXE To run it execute M8 BAT filename REM time where filename REM is the profile created in a standard of M8 see Appendix 1 and time is the file with the date of T in free format This program is used to speed up the update of the M8 algorithm test
3. Keunuc Bopox BH Kocodoxop BI Tenponmt NOBbINICHHO BepOATHOCTU BOZHHKHOBCHH CHIIHGDIUIHX 3e MIeETpAceHM MUpa MatremaTwueckue MeTOJIbI B ceHiCMOJIOrMM H reowMHamuKe M Hayka 1986 c 48 58 Buoncmpremtnag CeMCMOJIOrMA BoutycK 19 2 Gabrielov A M Dmitrieva O E Keilis Borok V I Kossobokov V G Kouznetsov I V Levshina T A Mirzoev K M Molchan G M Negmatullaev S Kh Pisarenko V F Prozoroff A G Renehart W Rotwain I M Shebalin P N Shnirman M G Schreider S Yu 1986 Algorithms of long term earthquakes prediction CERESIS Lima Peru pp 61 29 3 TonrocpouHbiit IporHo3 3eMileTpaceHun Merogageckue DekOMenmaunn Ion pen akan M A Canoscxoro M 4 3 AH CCCP 1986 127 c 4 JIMutpuesa O E Keumuc Bopox BH Kocodoxos BI Kysmenos M B Jenna T A Mopzaoen K M Hermatynnaep C X ucapenxo B Porpapn M M Ulpennep Ch J MarHocTuKa Geppgonop MOBbIIIeCHHOM BePOATHOCTH CHJIbHbIX 3CMJICETPHCeCHHM B ceHiCMOOlacHbIx perHoHax CCCP u pana apyrux cTpau B KH UucmeHHoe MOeuMpoBaHHe WM apnamnz reodu3uyueckux mpoueccop M Hayka 1987 c 99 111 BaraucnutTenbHasa ceuicmMomorua Brmvck 20 5 Keilis Borok V I and Kossobokov V G 1988 Premonitory activation of seismic flow Algorithm M8 Lecture Notes of the Workshop on Global Geophysical Informatics with Applications to Research in Earthquake Prediction and Reduction of Seismic Risk 15 Nov 16 Dec 1988 ICTP Trieste pp 17 6 Keilis Bor
4. circles 2 gt The 2 in brackets indicate that the current choice is circles Answer with a or type 2 to accept this shape For R R Mo km punch 0 for R R Mo gt answer with a or type O to accept this choice Next the program displays the area specific info listed in table and asks Correction deletion or addition c d a n gt Answer C for correction and then follow the table below Question Answer Area No gt A 19 Latitude 36 00 gt 13 Longitude 120 00 gt 145 T year month day 1989 01 01 gt 1993 7 1 Correction deletion or addition c d a n gt n The program displays the second block with your corrections and the message Is it correct y n 1 5 gt check if the corrections correspond to location at 13 N 145 E in Marianas and to J uly 1 1993 i e near the epicenter and about one month before the Guam 1993 earthquake by convention positive values are given to N and E and negative to S and W Then type y to finish changes of the settings The program will show the parameters of the M8 algorithm as if we are using it to predict the Guam 1993 earthquake as a magnitude 8 0 event in the circle centered at 13 N 145 E and will ask for Any changes n y gt Answer n For Do you want to print functions y n gt answer y For Do you want to save a profile y n gt answer y For Name the profile please gt type some file name e g guam rem The progr
5. followed by the values of counts assembled in a table The title of the table specifies the Region with its number T and coordinates of the center Each row lists the voting score g t h t the maximal magnitude of events observed since the previous determination the date of determination and the values of functions if the function can not be evaluated the program will mark it with If the value of a function is abnormally large it is accompanied with The voting resulting in a TIP declaration is followed by in the next four rows the program excludes from consideration these four determinations of voting In addition each table is preceded by the first record and is followed by the last record of quakes considered for these counts The last lines of M8 PRI summarize the number of STIP s FTIP s CTIP s and e c s and reports statistics of the occurrence of quakes with magnitudes ranging 27 from Mg 0 5 to Mg in the TIP s Four supplementary files might be useful for analysis and presentation of the results M8 ERS contains the coordinates of centers of areas used in the run of M8 M8 STR provides all the strong quakes in the catalog during the period of investigation in a special ASCII format see Appendix 2 M8 TIP lists in a short format all the TIP s determined Each TIP is specified by its Region number time span and by the number of strong quake in M8 STR that confirmed this alarm 0 for none M8 RES s
6. from Mp shf or a magnitude cutoff that provides on average constant annual number of events act It is used to specify the highest limit of magnitude range for the count 15 Value of the lower magnitude cutoff A value consistent with the type of magnitude cutoff 16 Value of the upper magnitude cutoff A value consistent with the type of magnitude cutoff 17 Size of the time interval s The integer multiple of the increment of determinations dt e g s 12 for time interval of 6 years and dt 0 5 year 18 Size of the time interval u for v type counts only The integer multiple of the increment of determinations dt e g it equals 18 for u 9 and dt 0 5 years 19 Time to The start point for the data used in analysis should be chosen with regard to the completeness for the magnitudes For example the NEIC Global Data Base System has records of events at the beginning of the century and earlier however in the M8 algorithm test we set t to be January 1 1963 since the analysis 15 requires completeness of magnitude 5 events 20 Percentages p and pg The percentile levels to define what values of a function are abnormally high By M8 program definition the highest p or pp percent of values only vote for a TIP 21 Interval This parameter defines time span for counting votes for a TIP 22 Voting thresholds H and G When the two measures of voting pass these thresholds i e h t gt H and g t
7. in Circum Pacific You can find corresponding batch files RUN_REG BAT and RUN_TEST BAT in the M8_TEST subdirectory To run the January 1 1994 update in region BM Bonin Mariana Trench go to the MS TEST subdirectory and execute RUN_REG BM 940 To run the July 1 1993 update in entire Circum Pacific go to the M8_TEST subdirectory and execute RUN_TEST 935 Two auxiliary programs The user is provided with the two supplementary programs CAT MS and ASC2B20 The first one selects a subcatalog from a catalog in the 20 byte binary format This program permits the specification of ranges of latitudes and longitudes i e lat_min lt latitude lt lat_max and lon_min lt longitude lt lon_max this area may cross the 180 longitude but its longitudinal size should not exceed 180 degrees To run CAT_M8 type CAT MS limits input_catalog output_catalog where limits is a file with lat_min lat_max lon_min lon_max in free format input_catalog and output_catalog are the file names the input catalog and the output subcatalog The program is useful when the catalog available covers significantly larger territory then an area of investigation e g we use it to speed up execution of updates of the M8 algorithm test in Circum Pacific The other program converts an ASCII catalog of mainshocks to the 20 bytes binary format which is used by M8 Each line of an ASCII catalog should contain the following as numbers separated by commas year AD month day
8. of shocks per year a the displayed value of a will be used 32 Type of MM c the magnitude threshold will be constant MM aa goes to Q 33 s the magnitude threshold will be counted from Mo MM Mo aa goes to Q 34 a the magnitude threshold will be counted from the activity goes to Q 35 y the displayed type will be used the displayed type will be used 33 MM aa The value of the constant magnitude threshold MM aa the displayed value of mm will be used 34 aa Mo MM ze The value of the difference aa between Mo and MM the displayed value of aa will be used 35 MM M aa where aa is the average The value of aa number of shocks per year aa the displayed value of aa will be used The value of the coefficient for the main 24 KREE eege 38 Is it correct 39 Values of percentile p for b pb and other functions p 40 Value of Nex Tex Nex dt 41 Values of G and H 42 Value of Nau Tau Nau dt 43 Do you want to print functions 44 Do you want to save data in a profile 45 Name the profile please shock weights in function s the displayed value of the coefficient will be used The value of the power for the number of main shocks summarized in function s the displayed value of the power will be used y goes to Q 3 n goes to Q 39 an integer from 1 to 5
9. thresholds 6 Running M8 The M8 program can be run either in interactive mode or in batch mode In interactive mode the parameters are entered through an interactive dialogue with the program In batch mode the parameters are entered through a special format profile prepared by the M8 program or with the aid of any DOS text editor We 17 recommend first running M8 program in interactive mode and create some profiles e g with the default values of the parameters Our experience in distributing the program at Workshops on earthquake prediction Lima Peru 1986 Trieste 1988 1991 and 1993 Caracas 1991 shows that it might be very useful to start with an example The participants of these Workshops performed postdictions of the Mexican 1985 Spitak Armenia 1988 Loma Prieta California 1989 Limon Costa Rica 1991 and Landers California 1992 earthquakes We suggest the user to begin with one example of the application of the M8 algorithm using the available global catalog of mainshocks MS4_NEIC DAT An Example Check that C M8 M8_EXE directory with M8 EXE is in your PATH and that your working directory has the catalog of mainshocks MS4_NEIC DAT from the IASPEI Disk X then execute M8 Opening information will appear on your monitor with the question Some profile y n name gt at the bottom By convention the program gives in brackets possible answers with a suggestion You can agree with the suggestion by ans
10. time interval Tb Te y goes to Q 3 n goes to Q 13 an integer from 1 to 5 see answers to Q 4 Comment Before this question the block No 3 concerning the space vicinities is displayed 1 the areas will be squares 2 the areas will be circles y the displayed value of R will be used the displayed value of R will be used The value of R which will be used If the answer is 0 the value of R is calculated by formula from comments to the description of the algorithm c goes to Q 16 d goes to Q 16 a goes to Q 17 n goes to Q 12 Comment Before this question the table of the areas considered is displayed For each area the coordinates of the center and the date of T are given 16 Area No The number of the area to be corrected or deleted according to the answer in Q 15 17 Latitude The value of latitude of the center the displayed value will be used 18 Longitude The value of longitude of the center the displayed value will be used 19 T year month day Values of year month and day for the date of T for the area the displayed value of T will be used 20 Is it correct y goes to Q 3 n goes to Q 21 An integer from 1 to 5 see answers to Q 4 Comment Before this question the block No 4 concerning functions of diagnosis is displayed 21 Correction deletion or addition c goes to 22 d goes to 22 a goes to 23
11. E directory should have the following files 1 M8 FOR Fortran source code of M8 main program with interactive interface for making counts on sequences of mainshocks and for identification of TIP s in areas of investigation 2 M8_BAT FOR Fortran source code of M8 main program to be run from command line for making counts on sequences of mainshocks and for identification of TIP s in areas of investigation 3 M8_SUBR FOR Fortran source codes of subroutines for M8 program 4 ASC2B20 FOR Fortran source code of 3 ASC2B20 program 12 5 CAT_M8 FOR Fortran source code of CAT MS program C M8 M8_TEST directory should contain 1 MS4_NEIC DAT the catalog of mainshocks prepared by P Shebalin UPDATE program from the National Earthquake Information Center Global Hypocenters Data Base System in 20 bytes binary format and 2 14 13 subdirectories with regional profiles to run the M8 algorithm test in Circum Pacific as on January 1 1994 and the results of such a run 15 Subdirectory TIMES with the files of dates of the M8 algorithm test updates 16 RUN_REG BAT the batch file to run the M8 algorithm test update in a region 17 RUN_TEST BAT the batch file to run the M8 algorithm test update in all regions of Circum Pacific Each of the 13 subdirectories with regional profiles should have a file of regional boundaries with extension CAT an input profile with extension REM see next Section and output
12. User Manual for M8 V G Kossobokov Contents l Introduction dc ieciesensuccnrveneyinessicedtvertemacssceeinesoanem 2 2 The M8 Algorithm EE 4 3 Hardware Requirements ege deg e 11 4 Software Installation cccccssoecscsedeeceveveeneheaenatsedbanaa tes 11 5 Input f r M8 so onconcnasonayextatssaacixnaseneseassaaeiieinomeass 14 6 tere M8 eege besen been 17 Te gt E 27 8 Error Messages and Trouble Shooting 28 Acknowledgements iecccncdncereccaianorcrciadertartiextmmecs 28 IRE TCPOHCES ccrveascsceieayenndeeotanssanvnrssevemnesnandedaewerteasevx 29 APPENA E EE 31 APpEndIK geegegeg engkeier 32 1 Introduction Earthquake prediction is an uncertain profession Following the Good Friday Alaskan earthquake the U S president s science advisor called a panel which recommended a ten year program of research on earthquake prediction 7 Stimulated by this recommendation the U S Geological Survey began a major program of research in 1966 Japan China and the Soviet Union also began at about the same time large research programs on earthquake prediction After almost thirty years of serious research however a reliable method to predict strong earthquakes does not exist Many methods for earthquake prediction have been proposed and some of these methods may be reliable The problem is that most of these methods cannot be tested and evaluated either in lieu of their precise definition or due to small sample statistics
13. When h t gt H and g t gt G at two successive determinations t dt and t the program reports of a TIP from t to t 1 Here T Tex H and G are given constants specified by the user M8 classifies a TIP as e c an earthquake caused activation when an earthquake of magnitude Mo or above occurred in the same area of investigation during the time from t 2dt to t 1 e just before a TIP The program considers such a case as no alarm STIP a successful alarm when an earthquake of magnitude Mop or above occurred in its spatial and temporal limits FTIP a false alarm when the alarm expired before the end of analysis Te and no earthquakes of magnitude Mo or above occurred in its spatio temporal limits CTIP a current alarm when it is not expired before T and no earthquakes of magnitude Mo or above occurred in its spatial and temporal limits In sum M8 allows the user to investigate the seismicity of a region in a multidimensional space of integral measures In such an investigation the user may get far away from the original M8 algorithm which passed through a number of tests at least in retroactive applications in numerous regions worldwide and thus loose much of its reliability The user is advised to supplement earthquake prediction results with a number of stability tests described earlier e g retroactive simulation of forward prediction etc To apply the M8 algorithm the user must use the default va
14. am then starts the calculations At the end of a run the program reminds that Outputs are in files m8 l The contents of these files are described below By inspection of the M8 PRI file you can check that the Guam 1993 earthquake did occur in a STIP If you do not want to lose results during the next run rename output files using the MS DOS ren command e g execute ren m8 guam we assume that M8 EXE is not in the same working directory Changing the algorithm The program permits variation of all parameters including the list of functions so that the user may create his own algorithm which has nothing to do with the M8 algorithm The M8 algorithm is neither optimal nor unique algorithm for the intermediate term prediction of earthquakes The fact that several large earthquakes were predicted and more than fifty retroactive case histories of successful predictions are established speaks in its favor However after almost ten years of study of premonitory activation we are still in the field of small sample statistics 20 Thus there is the need for additional testing and control experiments in each particular case Description of the dialog Following is the table of questions of the M8 program dialog and possible answers together with the program responses Answer and response of the program 1 Some profile y goes to Q 2 n the default is used goes to Q 3 the name of profile goes to Q 3 2 Na
15. d its Manual 28 References 1 Earthquake prediction A Proposal for a Ten Year Program of Research A Report of the Ad Hoc Panel on Earthquake Prediction to the Office of Science and Technology F Press Chairman Washington D C September 1965 2 The work of this Institute has been extensively published in the literature See for example Phys Earth Planet Inter 61 1990 3 V I Keilis Borok V G Kossobokov Phys Earth Planet Inter 61 73 1990 4 Proceedings of the National Earthquake Prediction Evaluation Council R G Updike Ed U S Geol Surv Open File Rep 89 114 1989 5 V I Keilis Borok V G Kossobokov in Proc of 27 Geol Congress 61 56 66 1984 6 J H Healy V G Kossobokov J W Dewey U S Geol Surv Open File Rep 92 401 1992 See also V I Keilis Borok V G Kossobokov in Comput Seismol 19 48 58 1986 7 J B Minster N P Williams in EOS Transactions AGU 1992 Fall Meeting 73 No 43 366 1992 8 R E Habermann Tectonophysics 193 277 1991 9 Yu V Riznichenko in Research on Physics of Earthquakes Yu V Riznichenko Ed Nauka Publishing House Moscow 1976 pp 9 27 in Russian I P Dobrovolsky et al Pure Apple Geophys 117 No 5 1025 1979 10 V I Keilis Borok et al J Geophys Res 85 803 1980 11 R S Olson The Politics of Earthquake Prediction Princeton University Press Princeton NJ 1989 Supplementary literature on the M8 algorithm 1
16. d to a period of about 10 years in most regions when used to predict a large earthquake Some of the functions in the algorithm may not scale with time and some changes in the method of selecting the starting time may be required for regions where there are longer periods of good data and in the future when more decades of good data are available In preparation of the data we recommend setting the initial cutoff depending on the completeness of the catalog For example in the global test of the M8 algorithm 6 we initially cut off the global catalog at magnitude 4 0 because the catalog is seriously incomplete below this threshold The Functions To count the number of earthquakes we must choose a magnitude cutoff The M8 algorithm does not choose a fixed magnitude cutoff rather it chooses the number of earthquakes that are desired for the calculation It selects in each CI two populations of mainshocks The first one CAT20 is defined by a magnitude cutoff that yields an average of 20 earthquakes per year Thus to analyze a 30 year interval the algorithm selects a magnitude cutoff that provides 600 earthquakes Usually the selected cutoff 5 will yield somewhat more than 600 earthquakes If there are less than 20 earthquakes per year in the CI we must make a decision to continue or to terminate the analysis In the test of the M8 algorithm 6 the analysis is not performed in a CI that contains less than 16 earthquakes per year A second p
17. e catalog the thresholds which cause the functions to vote may change so a TIP may be terminated early or extended for more than five years The M8 program The ultimate description of the M8 algorithm is the M8 program with a set of fixed values of parameters These are provided as default values in the program However the program permits one to define and run on seismic catalogs other algorithms based on the general concept of activation According to this concept predicted phenomena are likely to occur in a system during or following without much delay a period of activation when the majority of integral measures rise above their normal values A prediction strategy in this case could be to declare an alarm or time of increased probability TIP each time when such an activation is identified In addition to the M8 algorithm the M8 program may consider square areas of investigation and up to 6 types of functions described below M8 computes several integral measures on a sequence of earthquakes from an area of investigation and identifies TIP s The user must have a catalog of mainshocks in appropriate binary format see Input for M8 In one run of M8 the user has a choice to consider up to 18 areas either of circular or of square shape The dimension of areas i e radius or half side can be set to a constant value The user has a choice to use an arbitrary constant or to compute it using the above mentioned equation for R Mo T
18. erefore the user has to rebuild the catalog of mainshocks to make changes to M and or to e For each measure the user can select either of the two magnitude thresholds m or MI 1 as an arbitrarily chosen constant con 2 as a value shifted by constant from Mo shf and 3 as a magnitude cutoff that provides on average in an area of investigation constant annual number of events act The user also defines the percentile to be used in determining whether a value of a measure is abnormally high or not The program permits different percentile levels for B type and other measures For determination of percentiles and if any magnitude cutoffs defined by constant annual numbers of events in areas of investigation there is a choice between the interval Ts Te or Tp T where Te is the beginning of the analysis Te is the end of the analysis and T is a specific time in a given area of investigation For example during the M8 algorithm real time test in Circum Pacific 6 we chose intervals Ty T with T set to the date six months before of the update or the origin time of the magnitude M7 5 earthquake predicted in the circle of investigation To declare times of increased probability TIP s M8 computes two additional numbers h t and g t h t counts how many measures were above their normal values sometime in the preceding Tex years while g t counts how many different types of measures give positive contribution into h t
19. es of several integral counts based on the transient seismicity in a region These counts might be useful for an independent analysis of seismicity The first version of M8 dates from 1986 when the algorithm was distributed among participants of the Workshop on Earthquake Prediction held in Lima Peru Its parent programs were in operation in 1984 when Keilis Borok and Kossobokov presented the first results related to retroactive prediction of magnitude 8 0 earthquakes worldwide 5 Here we provide the 1991 versions for IBM compatible personal computers PC XT AT 286 PC 386 PC 486 PC 586 PC PS 1 and PS 2 A version for batch runs is used in updates of the test of the M8 algorithm 6 Another version permits interactive change Both versions have the same computational core and therefore give identical output for the same set of input parameters In 1992 J B Minster and N P Williams 7 reprogrammed M8 using the C language in a form which might be more suitable for statistical evaluation Final results obtained from their program M8C and from M8 may differ due to different roundoffs in the evaluation of functions The user should understand that catalog errors may impact all of the quantities determined by M8 since they may change the population of events which are counted especially when these errors have long term systematic nature 8 Therefore we strongly recommend the user to start his application with a thorough analyzi
20. fied below If you are not an experienced PC user please consult you DOS Manual We assume here that 1 you have the required hardware specified in Section 3 and 2 your PC boots up on a hard disk C and has a floppy and or CD disk drive A 11 Step 1 Turn on your PC After the DOS prompt appears on the monitor go to the root directory by issuing the DOS ed command Step 2 Check once again for a free space on your hard disk by issuing the DOS dir command Step 3 Insert IASPEI Disk X in the A drive and type a M8_inst If your floppy disk drive is B device in your PC then substitute b for a in the above command After installation your PC should have C M8 directory with subdirectories M8_EXE M8_CODE M8_TEST C M8 M8_EXE directory should have the following files 1 M8 EXE Executable program with interactive interface for making counts on sequences of mainshocks and for identification of TIP s in areas of investigation 2 M8_BAT EXE Executable program to be run from command line for making counts on sequences of mainshocks and for identification of TIP s in areas of investigation 3 ASC2B20 EXE Executable program to convert a catalog of mainshocks in an ASCII format to the 20 bytes binary format used by M8 program 4 CAT_M8 EXE Executable program to select subcatalog of events from specified ranges of latitudes and longitudes out of a given catalog in the 20 bytes binary format C M8 M8_COD
21. files from the January 1 1994 update with extensions PRI ERS RES TIP and STR Following is the list of the geographical names for the 13 regions TO Tonga Kermadek GU New Guinea JV Java Trench PH Philippines TJ Taiwan and Ryukyu JB Japan Idzu Bonin Trenches BM Mariana Trench KK Kurils and Kamchatka AA Aleutians and Alaska WA Western United States CA Central America SA South America SS South Sandwich Islands The complete contents of these directories are given in the INDEX file of C M8 directory Step 4 Check the presence of the above files on the hard disk If these files exist remove theNLDEP Disk from the A drive and return it to the binder 13 5 Input for M8 To run M8 the user must specify the following parameters 1 Catalog file name You must have a catalog of events in the 20 bytes binary format a standard for earthquake data analysis used in the International Institute for Earthquake Prediction Theory and Mathematical Geophysics Russian Academy of Sciences Remark The catalog file name should not exceed 14 bytes 2 Magnitude cutoff Mo This parameter defines events to be predicted as those with magnitude M gt Mp 3 Interval of analysis Ch Te The dates year month day of the beginning and ending of determinations of measures e g 1975 1 1 1994 1 1 4 Choice of interval for determination of percentiles Two choices to consider values either 1 fro
22. gt G at two successive determinations t dt and t the program reports of a TIP starting at t 23 Interval t This parameter sets the duration of a TIP 24 Choice of printout file This parameter allows the user to chose between full size output and short one default In the later tables of function counts are omitted 25 Choice to save the profile It is convenient to store in a file all the parameters for one run of M8 Such profiles permit the user to reproduce the results achieved in the past They are very helpful when the user makes numerous runs with few changes of parameters If needed the program can output all the parameters into a profile of special format with the name specified by the user 16 Because of the limited number of typeface characters used in the M8 program there is a need in the following table which clarifies equivalence of the constants Mo Magnitude cutoff which defines large events M Lower magnitude cutoff for the count Upper magnitude cutoff for the count Magnitude cutoff for the count of aftershocks The start point of the analysis The beginning of determinations of measures The ending of determinations of measures A specific time in a given area The step between determinations of measures Intervals of time for determination of measures The duration of a TIP The time span for counting votes The constant for exponential weights in s t The power for denominator in s t The voting
23. he formula values are used as default values The user can consider up to 9 integral measures of the following types n count of the number of mainshocks with magnitude M gt m in the preceding s years n tlm s 1 deviation of count n from a longer term average Im e tel n tlm s n tlm t s to s t s to where to is the beginning of the analysis k difference between two consecutive independent determinations of count n Kim s n thm s n t slm s v variation of count n in preceding u years v tim s u n Clm s n C dtlm s where 6 belongs to t t u and dt is a uniform step between consecutive determinations s magnitude weighted sum of mainshocks divided by the number of these events to the y power s tlm M s B y 10 IC 1 here the summations are taken over the preceding s years on a population of mainshocks j with magnitudes Mj from the interval m lt Mj lt MI Note n type is formally embedded into s type for B y 0 b maximal number of aftershocks in the preceding s years b thm M M s e max Bj here Bj is the number of aftershocks with magnitudes above the threshold M in the first e days following each mainshock j and the maximum is determined over the preceding s years on a population of mainshocks j with magnitudes Mj from the interval m lt Mj lt MI Note M8 does not calculate B but uses the counts of aftershocks provided in the catalog of mainshocks Th
24. hour minute latitude longitude both in degrees depth in km magnitude number of 26 aftershocks e g 1993 8 8 8 34 12 96 144 78 60 8 2 75 By convention positive values of coordinates are N and E and negative S and W To convert an ASCII catalog named ASCII to binary catalog named BINARY execute ASC2B20 ASCII BINARY 7 Output of M The M8 program outputs several files M8 PRI M8 ERS M8 TIP M8 STR M8 RES and optional profile REM M8 PRI is an easy to read output with the results of the M8 program run Its heading lists all the values of input parameters For each area of investigation i e Region the program reports the annual average rate of all the events from the catalog including records for the first and the last quakes used in this calculation and for every function the magnitude range of the count Next for each Region the program compares the annual average rate of all events to the maximal requested rate from the definitions of functions If the later is greater the program will give a warning The program will not perform computations for the region if the rate of activity in it is less than 80 of the requested If the rate is sufficient for calculations the program reports if any the TIP s and strong quakes mainshocks with magnitude greater or equal to Mo in the region Each TIP is specified by its identification Region number coordinates of center and time span If requested these aree
25. hows the current state of alarm in all the areas of investigation at T 1 for an alarm O for no alarm 1 for insufficient data 8 Error Messages and Trouble Shooting M8 returns the error messages when it can not find a file of mainshocks or a profile specified by the user In these cases the program displays input parameters and asks the user to correct them the batch version of M8 program will terminate the run As mentioned above the program gives warnings when the catalog might be or is insufficient for the analysis The author apologizes for not protecting most of the numerical input from mistyping and range inconsistency The author strongly recommends the book by R S Olson for the user who does not want to get into trouble issuing earthquake predictions Acknowledgements The author of the M8 program is indebted to his coauthor of the M8 algorithm Prof Vladimir Isaacovich Keilis Borok The author is grateful to Profs L Knopoff I M Rotwain G M Molchan A V Gabrielov M G Shnirman J H Healy J W Dewey and J B Minster for criticism and fruitful discussions at different stages of design and testing of the M8 algorithm The comments of Drs I V Kusntsov I M Primakov and A A Soloviev related to the program per se are gratefully appreciated The author thanks Profs J H Healy and W H K Lee for the opportunity of this publication and Profs D Vere Jones Ma Li and R E Habermann for the reviewing of the program an
26. ing each mainshock F t is set equal to the largest Bj in CATMS in the preceding year The functions are evaluated every six months and it is convenient to describe them in terms of six months intervals ending at ti t i half years Let X be the number of earthquakes in the six months interval ending at ti and S be 10 where Mij are magnitudes of all mainshocks from the i th half year interval In this notation the functions are Fit Kags Xion EE CAT20 Pa a Xt Ren eX CAT10 R Fitt kk ER Aen CAT20 Falt Pat Rat X 12 12 G 12 CAT10 Fs t Suir 8 40 Fut SOIC i Xote t XD CAT20a Fst Sart Sco Hut SOX art Sea tc CAT10a Filz max Bj in the preceding year CATMS In we indicate the catalog used to calculate the function Votes The functions vote for declaring of a warning or time of increased probability TIP for an earthquake with magnitude gt Mo in the CI The first six functions cast their vote for a TIP when they are in the top ten percent of their historic range The seventh function votes for a TIP when it is in the top 25 of its historic range TIP s A time of increased probability is declared for the CI when five of the first six functions and the seventh function have voted sometime in the preceding three years and when this condition is met in two successive six month evaluations Once a TIP is declared it lasts for five years As new data is added to th
27. lues of the program 1 e to consider circular areas of investigation with radius R Mp the seven functions F and F of type n F and F of type 1 F5 t and F of type s and F of type B etc see above 10 3 Hardware Requirements To run M8 you need 1 An IBM compatible personal computer PC XT AT 286 PC 386 PC 486 PC 586 PC PS 1 and PS 2 running the IBM PC DOS or MircroSoft DOS operating system version 3 1 or later 2 At least 250 kilobytes of free RAM memory 3 A floppy or CD disk drive 4 A hard disk is optional but necessary for any serious work Floppy disk drives are just too slow and limited in storage capacity For complete installation of the programs auxiliary files and data for the M8 algorithm test in Circum Pacific you should have at least three megabytes of free space At run time you must have free room to hold output files generated by M8 5 A math coprocessor compatible with your PC s CPU is desirable but not necessary Without it floating points calculations will be emulated by software and will slower execution 6 A printer if you need to print the results 4 Software Installation Although it is possible to run M8 from a floppy we will describe the software installation for a hard disk only Software installation for M8 simply consists of copying the M8 program and its associated files from NLDEP Disk To proceed first find the NLDEP Disk and then follow the steps speci
28. m interval Ch Te or 2 from Oh T where T is a specific time in a given area of investigation e g the origin time of a large earthquake 5 Increment of determinations dt Interval in years e g 0 5 6 Number of functions Total number of counts in one determination i e dimension of description 7 Number of areas of investigation Total number of areas to be considered in arun of the program 8 Shape of areas Two options circular or squared shape 9 Dimension of areas R Radius of circle or half side of square in kilometers If the value of this parameter is set to 0 the program calculates R R Mo 55 5 exp Mo 5 6 1 For each area of investigation the user has to define 14 10 Center of the area The latitude and the longitude of the center for each area positive values for N and E and negative for S and W 11 Time T The date which is specific for the area For each function the user have to define 12 Type of measure One of n 1 k v s b 13 Type of lower magnitude cutoff This parameter can be either an arbitrarily chosen constant con or a value shifted by constant from Mp shf or a magnitude cutoff that provides on average constant annual number of events act It is used to specify the lowest limit of magnitude range for the count 14 Type of upper magnitude cutoff This parameter can be either an arbitrarily chosen constant con or a value shifted by constant
29. me of profile please the name of profile goes to Q 3 3 Any changes y goes to Q 4 n goes to Q 43 Comment Before this question the table with all parameters considered is displayed 4 Number of block to be corrected 1 goes to Q 5 2 goes to Q 8 3 goes to Q 12 4 goes to Q 20 5 goes to Q 38 5 Is it correct y goes to Q 3 n goes to Q 6 an integer from 1 to 5 see answers to Q 4 Comment Before this question block No 1 concerning the catalog is displayed 6 Name of file with main shocks the file with the displayed name will be used the name of the file with main shocks which will be used 7 Mo the displayed value of Mo will be used the value of Mo which will be used 8 Is it correct 9 Tb Feet er S 10 dt 11 Time interval for activity estimation 12 Is it correct 13 Shape of area 14 R km 15 Correction deletion or addition y goes to Q 3 n goes to Q 9 an integer from 1 to 5 see answers to Q 4 Comment Before this question the block No 2 concerning common time constants is displayed the displayed values of Tb and Te will be used the values of Tb and Te which will be used the displayed value of dt will be used The value of dt which will be used 1 magnitude thresholds will be defined for time interval Tb T 2 magnitude thresholds will be defined for
30. n goes to 20 Comment Before this question the table of functions of diagnosis is displayed 22 Function No The number of function to be corrected or deleted according to the answer in Q 15 23 Function x y the displayed type x of function remains the displayed type x of function remains the type of function to be used Comment Before this question the list of general notations of functions which can be used in diagnosis is displayed 24 Value of ns s ns dt The value of the number ns of steps dt in s the displayed value of ns will be used 25 Value of nu u nu dt The value of the number nu of steps dt in u the displayed value of nu will be used 26 Beginning of the catalog To The year to start estimation of long range trend of seismic activity the displayed year will be used 27 Type of M c the magnitude threshold will be constant M a goes to Q 28 s the magnitude threshold will be counted using Mo M Mo a goes to Q 29 a the magnitude threshold will be counted from the activity goes to Q 30 y the displayed type will be used the displayed type will be used The value of the constant magnitude threshold M a the displayed value of M will be used 29 a Mo M The value of the difference a between Mo and M the displayed value of a will be used 30 M M a where a is the average number The value of a
31. ok V I and Kossobokov V G 1990 Premonitory activation of seismic flow algorithm M8 Phys Earth and Planet Inter 61 73 83 7 Keilis Borok V I and Kossobokov V G 1990 Times of Increased Probability of Strong Earthquakes M 7 5 Diagnosed by Algorithm M8 in Japan and Adjacent Territories J Geophys Res 95 No B8 12413 12422 30 Appendix 1 Format of the profile for M8 REM The first line contains 1 14 the name of the catalog of events a14 15 17 the magnitude Mo multiplied by 100 13 The second line contains 1 8 the value of dt F8 5 9 12 the year of Tb 14 13 16 the year of Te 14 17 18 the number of functions 12 19 20 the number of areas of investigation 12 21 22 the index which defines the shape of areas of investigation 1 square shape and 2 for circular shape i2 23 26 the value of R Mo in km 0 for the default formula 14 27 28 the index which defines the interval of analysis 1 for Tb T and 2 for Pb Te G2 Each of the following lines their number must be equal to the number of areas of investigation defined in the second line contains 1 6 the latitude of the center of an area of investigation F6 2 7 13 the longitude of the center of an area of investigation F7 2 14 21 the date of T year month day 14 12 12 Each of the following lines their number must be equal to the numbe
32. opulation CAT10 is selected from the CI in the same way using the magnitude cutoff that produces 10 earthquakes per year These two populations of earthquakes are selected for each CI so that the corresponding magnitude cutoffs may vary from one CI to another in response to varying levels of seismicity The algorithm is based on seven functions of the seismicity data F t is the count of the number of earthquakes from the first population CAT20 in the preceding six years F t is the same count as F t for the second population CAT10 F t F t Rutt where Rut is the average number of earthquakes per six years in the interval from the beginning of the analysis to to t 6 years The measure is taken on CAT20 F t is the same as F3 t measured on CAT10 F t is a magnitude weighted sum of mainshocks divided by the number of these events to the 2 3 power amp jo JC 1 The summations are taken over the preceding six years from a third population CAT20a CAT20 events with magnitudes Mj above M 0 5 F t is the same as F t calculated on a fourth population CAT10a CAT10 events with Mj above Mg 0 5 F t is based on a count of the number of aftershocks To calculate the function we define a fifth population CATMS including mainshocks in the magnitude range Mo 2 0 Mo 0 2 within a given CI The catalog of this population includes the count of aftershocks B in the first two weeks follow
33. plays parameters of the first block and asks Is it correct y n 1 5 gt No we want to change the catalog name and set Mo 8 0 After n the program will ask for Name of file with main shocks mWestUS dat_ gt Let us type in ms4_neic dat we assume that this file is in your working directory For request Mo 7 00 gt type 8 0 The program will display the first block with your corrections and will ask again Is it correct y n 1 5 gt If the corrections are right type 2 to change T The program displays parameters of the second block and asks ls it correct y n 1 5 gt Here we have to change Te 1990 01 01 for 1994 01 01 i e the end of the NEIC catalog available After n the program brings on the screen Tb Te 1975 1 1 1990 1 1 gt Answer 1975 1 1 1994 1 1 It is possible to give shorter answer 1994 because of convention of numeric input Answer to the further requests in this block with s After the program displays the second block with your corrections and the message Is it correct y n 1 5 gt check if the corrections are OK and type 3 to change the settings of areas of investigation Now the program displays parameters of the third block and the message Is it correct y n 1 5 gt Note that one change in the block was already made when we set Mo 8 0 The value of R now is 668 km but not 281 km as it was for Mo 7 0 After n the program brings on the screen Shape of area 1 squares or 2
34. r in an aftershock zone are removed from the catalog to produce a catalog of mainshocks This definition of aftershocks and mainshocks oversimplifies a complicated issue but seems rather adequate for our purposes Selecting a starting time and magnitude cutoff Following the methods of pattern recognition the data is transformed by a number of functions The most important part of these transformations is the averaging over time and space The spatial averaging is accomplished by selecting all of the earthquakes in a given CI The time averaging is accomplished by summing the values over the preceding s years To count earthquakes we must deal with two important questions the time interval over which the earthquakes are to be counted and the magnitude of the earthquakes The time interval selected for analysis is largely determined by the quality and completeness of the seismic catalogs In the 1960 s the quality and completeness of the global catalogs were improved with installation of the Worldwide Seismic Network Since 1963 all the earthquakes of magnitude five or greater are located and in many areas catalogs are complete to magnitude four The beginning of the analysis is usually selected as January 1 1963 although the time may be earlier or later depending on the quality of the local catalogs It takes about 20 years of data for the algorithm to set the thresholds that define anomalous seismicity so application of the algorithm is limite
35. r of functions defined in the second line contains 1 the type of measure al 3 5 the type of lower magnitude cutoff a3 8 10 the type of upper magnitude cutoff a3 13 16 the value consistent with the type of lower magnitude cutoff F4 2 18 21 the value consistent with the type of upper magnitude cutoff F4 2 22 24 the number of steps dt in the interval s 13 25 26 the number of steps dt in the interval u 12 27 30 the year of To 14 31 35 the value of beta F5 2 36 40 the value of d F5 2 The following line contains 1 8 the value of percentiles p and pb f4 2 f4 2 9 11 the number of steps dt in Tau i3 12 14 the number of steps dt in Tex 13 15 16 the value of H i2 17 18 the value of G 12 31 19 20 21 34 the index which defines whether to print tables of functions or not 1 for yes and 0 for no il reserved il reserved a14 The following last line contains 1 8 9 16 the date of Tb year month day 14 12 12 the date of Te year month day 14 12 12 Appendix 2 The format of an ASCII output file STR Each record of such a file contains 32 1 4 5 6 7 8 9 10 11 12 13 14 15 19 20 25 26 28 29 31 32 34 36 41 the year of the origin time of an event 14 the month of the origin time of an event i2 the day of the origin time of an even
36. s of the input catalog for completeness and for presence of duplicates The authors emphasize that any specific prediction can be made only on experimental basis and users are advised to disclaim any liability for consequences of actions based on their predictions Accordingly the authors disclaim liability for any consequences of the use of this program Let us remind also that each prediction of a large earthquake must be released by a qualified scientific body to proper local authorities It should be mentioned that the M8 program allows broad variation of parameters including a set of measures two different shapes of an area of investigation size of areas and other constants of the M8 algorithm Therefore M8 program should be considered as an exploaratory tool which permits when set to default values to run the M8 algorithm In the next section Section 2 we start with a description of the M8 algorithm along with definitions of the measures which are traditionally used and conclude it with a more general scheme which is employed in M8 The following two sections describe hardware requirements and software installation Section 5 lists all input parameters of M8 In Section 6 we provide an example of 3 application of the M8 algorithm to predict the Great Guam August 8 1993 M 8 2 earthquake This section contains also the table with the description of the dialog and an explanation of how to run the batch version of M8 Then we describe
37. t i2 the hour of the origin time of an event 12 the minute of the origin time of an event 12 reserved i2 the latitude of epicenter of an event multiplied by 100 15 the longitude of epicenter of an event multiplied by 100 16 the depth in km 13 the magnitude of an event multiplied by 100 13 the count of aftershocks of an event multiplied by 100 43 reserved 13 13 11
38. the output of M8 Section 7 and possible error messages Section 8 2 The M8 Algorithm The circles of investigation CI s The MS algorithm examines seismicity in circular regions In early testing of the algorithm on magnitude 8 earthquakes the best results were obtained using a radius of six degrees of the Earth meridian i e 668 km Previous work had established a rough scaling law between magnitude and source dimensions 9 After some testing the developers set the relationship between the radius of the CI and the magnitude of the predicted earthquake Mo as R Mo 55 5 exp Mo 5 6 1 which gives 668 km for magnitude 8 0 427 km for 7 5 281 km for 7 0 and 192 km for 6 5 Aftershocks The strongest statistical effect in earthquake catalogs is aftershock phenomenon the clustering of smaller earthquakes following the occurrence of a larger earthquake Aftershocks can mask weaker phenomena unless something is done to suppress or remove them In the M8 algorithm aftershocks are defined as all of the smaller earthquakes that follow a larger earthquake within a time and distance that are functions of the magnitude of the mainshock For purposes of the algorithm the time and distance that define the aftershock zone are arbitrarily set to an exact value in a table from 10 Magnitude M 40 45 50 55 60 65 7 0 7 5 8 0 r M km 40 40 50 50 50 100 100 150 200 T M days 23 46 91 183 183 365 730 913 1096 All the earthquakes that occu
39. wering if not you should type in your own answer Let us start with the default parameters by answering n to the opening question The program will show the default parameters of the M8 algorithm which were used for the prediction of the Loma Prieta 1989 earthquake as a magnitude 7 0 event in circle centered at 36 N 120 W and will ask for Any changes n y gt In the first run we recommend to go for changes answer y and correct the file name of the catalog and some other parameters As an example let us consider that we want to know whether or not the Great Guam August 8 1993 magnitude 8 2 earthquake occurred within a TIP diagnosed by the M8 algorithm After answering y the program will ask the user to specify Number of block to be corrected 1 5 gt The M8 program parameters are arranged into 5 blocks specifying 1 catalog of mainshocks and magnitude Mo that defines earthquakes to be predicted 2 temporal constants of analysis 3 areas of investigation 4 functions and 5 constants of 18 a TIP diagnostics In our example we should not change anything but the catalog name the constant Mp the ending of analysis and location and specific time of the CI Thus we have to make corrections of blocks 1 3 Let us start with 1 Note For a larger set of areas of investigation the list of all parameters may not fit one screen In such a case the user can go for changes in a block he wants to check The program dis
Download Pdf Manuals
Related Search