QTQt User Guide v4
Contents
1. 2009 A Bayesian approach to infer environmental parameters from stratigraphic data 1 Methodology Basin Research 21 5 25 Flowers R M Ketcham R A Shuster D L amp Farley K A 2009 Apatite U Th He thermochronometry using a radiation damage accumulation and annealing model Geochimica et Cosmochimica Acta 73 2347 2365 Gallagher K 2012 Transdimensional inverse thermal history modelling for quantitative thermochronology J Geophys Res 117 B02408 do1 10 1029 2011JB00882 Gallagher K Bodin T Sambridge M Weiss D Kylander M and Large D 2011 Inference of abrupt changes in noisy geochemical records using Bayesian transdimensional changepoint models Earth Planet Sci Letts 311 182 194 Gallagher K Charvin K Nielsen S Sambridge M and Stephenson J 2009 Markov chain Monte Carlo MCMC sampling methods to determine optimal models model resolution and model choice for Earth Science problems J Marine and Petroleum Geology 26 525 535 Gallagher K Stephenson J Brown R Holmes C and Fitzgerald P 2005 Low temperature thermochronology and modelling strategies for multiple samples 1 vertical profiles Earth Planet Sci Letts 237 193 208 Gautheron C Tassan got l Barbarand J amp Pagel M 2009 Effect of alpha damage annealing on apatite U Th He thermochronology Chemical Geology 266 157 170 Hopcroft P Gallagher K and Pain C C 2007 Inference of past c2. 13 242 672 0 06 3 154e 26 Enter c 0 0 d 0 0for EasyRo jai 14 251 04 0 05 3 154e4 26 c 0 00 15 259 408 0 05 3 1540 26 d 0 00 I I 16 267 776 0 04 3 154e4 26 17 276 144 0 03 3 154e 26 18 284 512 0 02 3 154e 26 Cancel OK 19 292 88 0 02 3 154e 26 P 20 301 248 0 01 3 154e 26 Here you enter a mean observed vitrinite reflectance VR obs with an error value Error as well selecting whether or not to use the observed value as a constraint Use for thermal history calibration for the inverse modelling if not then you can still plot the predicted vitrinite reflectance for various thermal history models once the inversion run has finished The table on the right shows the default kinetic parameters for EasyRo from Sweeney and Burnham 1990 You can enter up to 20 of your own values Eact Activation Energy F is the stoichiometric factors which need to total 1 and A 1s the frequency or pre exponential factor Similar to the He age data description above you can choose to Resample VR which will sample the observed vitrinite reflectance from a normal distribution centred on the input value with a standard deviation equal to the input error value Alternatively you can select Resample Error This option samples a scaling factor for the input error which is used in the calculation of the data fit The scaling factor is between 0 1 and 10 so the data can effectively be treated as being more precise low scaling factor or
3. He Data button Otherwise you can enter just the observed age and error or if you want to calculate the age you can enter the U Th Sm and He concentrations in the units indicated on the window and click the Calculate Age button Note this is the uncorrected for alpha ejection age The He age will be compared to the input age If these are not equal and the He ejection distance is not zero then by default the inut age is reset to the calculated age for consistency If you want to use Monte Carlo sampling of the observed He age using a normal distribution centred on the observed value with a standard deviation equal to the input error you can check the box Resample He age with MCMC Instead of trying to fit just the single input age this process samples the normal distribution allowing for the uncertainty in the observation In practice this can be thought of as a way of allowing for uncertianty in the predictive model e g the kinetics in that we do not try to fit the obsrved age exactly Alternatively you can select Resample He Error with MCMC This option samples a scaling factor for the input error which is used in the calculation of the data fit The scaling factor is between 0 1 and 10 so the data can effectively be treated as being more precise low scaling factor or less precise high scaling factor relative to the input error value This may be useful when you are not sure of how good you error estimates are When you set the M
4. Led KG ag gi H Time Mai To move an already created point place the mouse on the point click and hold down the mouse and drag the point to the new desired location You can move a point so its time value crosses becomes older or younger than an adjacent point To delete a point place the cursor on the point and double click Notes The time temperature points can be input in any order QTQt will sort them If you place the cursor at a time lt 0 0 QTQt will set the time to 0 If you do not input a point with time 0 QTQt will add one using the temperature of the closest time point When using a profile you can then click on the Edit Hot button Again you place the mouse in the window and when you press and hold the mouse you will see the time temperature and temperature offset written at the bottom left of the window The temperature offset is the difference in the temperature between the Hot and Cold thermal histories 1 e the top and bottom samples in a profile Note The temperature offset is not the temperature gradient You do not have to be too precise with the time coordinate as QTQt will just choose a point with the time closest to the mouse coordinate value Choose any time point then click and and move the mouse with the button held down to get the offset you want When you release the mouse you will see a red thermal history drawn below the blue one using a constant temperature offset equal to the value sele
5. Thermal history file Constraints menu you should see a window as shown to the cisestrragein pronte ony a right Click OK and the select Set MCMC parameters from nee riet Temperature CO the MCMC Run menu You will see the window below with as em ag BESSE Temp Offset CC 0 0 D Allow offset bo vary Over mg QTOt MCMC parameters default proposal scales aaa cor geg wa 20 0 10 0 temp Offset roi Ai MCMC chain Burn in Post burn in Thinning Tee ry Temperature CC Range Wear Range 1000 1000 1 ee Proposal Move _l Constrai Time Temperature Offset Constrai 5 0 5 0 2 0 Constral Geteste Temperature Offset Se 50 0 5 0 f Cancel z de Click OK and choose Run Ac tance rates gg ee ee from the MCMC Run menu 0 0 0 0 0 0 Once the run ios finished you should see a window similar to the one Birth Death below 0 0 0 0 Cancel OK Here you can see the acceptance rates around 20 30 for the Time and QTat MCMC parameters Temperature and typically lower values for the birth and death but the Burn in Post burn in Thinning two values birth and death are similar You can change proposal scales Legame Ss and look at how the acceptance rate changes If the scales are small then e ee the acceptance rates tend to be higher and vice versa This is not always 5 000 5 000 1 000 the case as there is some interaction between the different parameters and Proposal Birth Temperature Offset MCMC moves 50 000 1 000
6. apatite and those of Tagami et al 1998 and Yamada 2007 for zircon Currently a given sample can be modelled with a constant composition if appropriate The composition could be taken as the average of measured single grain compositions or alternatively a single real sample could be divided up into multiple samples based on different compositions for example amd then treated as mutliple sample for modelling purposes Similarly a sample with both apatite and zircon data should be treated as 2 samples For predicting He diffusion standard diffusion equations are used explicitly a spherical grain with the same surface area to volume ratio as the dimensions specified for a real grain You can input kinetic parameters to simulate He diffusion in any mineral zircon for example The He diffusion model also includes the recent developments on radiation damage trapping Flowers et al 2009 Gautheron et al 2010 using fission track annealing as a proxy to recalibrate the helium diffusion coefficient It is possible also to include Hei He degassing spectrum as part of the He data modelling process Finally vitrinite reflectance data can be incorporated being used either as a direct constraint on the inferred thermal histories or it 1s possible just to predict vitrinite reflectance and make a qualititative comparison to the observed values The inversion scheme is Bayesian transdimensional Markov chain Monte Carlo MCMC in which the number of time t
7. history 39 Index acceptance Reeg Dil Act energy Holu 13 Annealing Model 24 Bits 26 Build OTOt data file 10 Buri alleata 25 Calcule ASS 12 Calculate initial track length 18 Caluclate Initial Length 24 CHUCK Sani 11 18 24 ne E 10 11 Compositional Model 18 24 va EE 18 24 KE EE 18 24 Constrain Time temperature box 22 Constrained Point eia 17 Constrained Present day 18 23 Constrained Time Temperature point 23 Data Entering dS Taa s 14 Data Entering fission track counts 10 Data Entering fission track lengths 11 Data Entering U Th He CHei Hei purega 12 Data Entering Vitrinite Reflectance 15 DATA INPUT E E E 6 Deha 26 Do He VA EE 13 EE ppi 11 18 24 EXAMINE CHAIN iris 28 Expected Model 30 BILENMEND ass aaa e 8 Forward thermal history EE eebe Ee 19 Generate all plots 36 He ejection distance 13 Individual Sample E EE 22 25 INSTALLING OTTO 6 Introduction 2 length Giya lOr TE 13 Likelihood Chain 28 Max Likelihood Model 29 Max Mode Model 30 Max Posterior Model eee 30 MCMC RUN M
8. is used for needto enter constraints for stratigraphic age modelling the vitrinite reflectance value for a given sample If you me am temperature and present day temperature have input vitrinite reflectance you will see the window to the left as a reminder Just click on OK to continue the input The Constrained Present day option can be used to set the present day temperature to be within given limits meantrange for any sample This is most likely to arise for well data If the Range parameter is set to zero the value will be fixed at the Mean value You can set the Annealing Model for a particular run and for the two Ketcham et al models you will need to set the Compositional Model parameter Value is required according to the type of kinetic parameter you want to use If you set the Uncertainty to 0 then the annealing model 1s chosen using this fixed value of the kinetic parameter If you set the Uncertainty to a non zero value then the compositional parameter is sampled from a normal distribution with a mean equal to Value and a standard deviation equal to Uncertainty This is one way to allow for uncertainty in the annealing models or their calibration You can set the Initial track length to a specified value Note that for the Ketcham et al annealing models the default is to Calculate initial track length using the compositional information You can uncheck this option if you do not want to do this You can modify the annealin
9. not click Save for Rerun the changes will not be saved Ae QTQt MCMC parameters MCMC chain Burn in Post burn in Thinning 10000 10000 1 Proposal Move Time Temperature Offset 15 000 15 000 1 000 FT Annealing He Diffusion Vitrinite Refl 1 000 1 000 1 000 Pri Birth SORA Temperature Offset 50 000 1 000 Acceptance rates Time Temperature Offset 05076 0 0000 FT Annealing He Diffusion Vitrinite Refl 0 1701 0 0000 0 0000 Birth Death Save for Rerun Cancel OK The acceptance rates for the time temperature and offset parameters should usually be around 0 2 0 5 with similar valuesfor the FT Annealing He Diffusion and Vitrinite Refl sampling values while the Birth and Death acceptance rates tend to be low e g 0 05 but in general should be more or less the same 28 PLOTTING MENU This menu allows you to examine various plots summarising the output from a given sampling run Each plot window has the menu options illustrated below By clicking on the appropriate image a plot can be saved in the generic Scaled Vector Graphics format this is the WWW standard and readable by Adobe Illustrator for example although different versions of Illustrator seem to import the graphics differently so sometimes the layout is not the same as the screen version printed directly and generally you can print to a pdf file 1f you prefer this format or you can change the axis scales from the o
10. 0 1 0000 1 0000 Convergence criterion for series sum Geometry flag 1 sphere 2 slab Number of domains Act energy kcal mol 4 184 kJ mol logio Do a prop of domain Number of heating steps Fraction released age Ma Error on age Ma 1 0 use for fitting or not l l l l l l l l l l l l l l l l l l l l The name of the Ar Ar datafile must be the same as the general data file with Ar added at the end before any For example if the general datafile with AFT and AHE data for example is called Mydata txt then the name of the Ar Ar datafile needs to be MydataAr txt This file must be in the same directory as the general data file 14 Vitrinite Reflectance data Once you click on OK the following window will appear to allow the input of vitrinite reflectance VR Data x Y Fa Sample O No Vitrinite Data Eaet die PD Amy 1 142 256 0 03 3 154e 26 V obs C Resample VR 2 150 624 0 03 3 154e4 26 3 158 992 0 04 3 154e4 26 En _ Resample Error 5 4 167 36 0 04 3 154e 26 5 175 728 0 05 3 154e 26 Use for thermal history calibration 6 184 096 0 05 3 154e 26 7 192 464 0 06 3 154e4 26 E 200 832 0 04 3 154e4 26 VR exp a bF o 209 2 0 04 3 154e 26 Enter a 1 3 b 3 7 for EasyRo I m 10 217 568 0 07 3 154e 26 a 1 30 11 225 936 0 06 3 154e4 26 b 3 70 12 234 304 0 06 3 154e8 26 Calculate Ln A c d Eact
11. 18 700 00 0 50000 4 6900 0 090000 1 3539 0 029976 19 900 00 0 50000 1 1700 0 050000 1 3339 0 069944 The name of the Hei He datafile must be the same as the general data file with 43 added at the end before any For example if the general datafile with AFT and AHE data for example is called Mydata txt then the name of the He He datafile needs to be Mydata43 txt This file must be in the same directory as the general data file 13 Ar Ar data To include such data in the modelling process you need to have the data in a separate file in the following format modified from a format provided by Peter Zeitler at Leigh University The modelling uses the MDD approach and Peter Zeitler provided the calculation routines based on the publications and code of Oscar Lovera University of California Los Angeles 0 00000000000001 l 6 33 12 6 3010 0 16667 33 12 5 6990 0 16667 33 12 5 3468 0 16667 33 12 5 0969 0 16667 33 12 4 9031 0 16667 33 12 4 7447 0 16667 20 0 08040 27 243 0 11170 45 861 0 13260 53 636 0 14550 58 686 0 15470 62 478 0 16700 65 519 0 17790 68 064 0 18950 70 321 0 20140 72 489 0 22590 74 760 0 25270 77 172 0 26820 79 549 0 28340 81 719 0 30450 83 621 0 32790 85 403 0 37820 87 227 0 56170 89 106 0 63300 90 944 0 81650 92 713 1 0000 94 657 1 0000 1 0000 1 0000 1 0000 1 0000 1 0000 1 0000 1 0000 1 0000 1 0000 1 0000 1 0000 1 0000 1 0000 1 0000 1 0000 1 0000 1 000
12. 301 248 0 010000 3 1536e 26 16 Once you have entered the data a final window will appear will provides a short summary and allows to you to choose certain modelling parameters e g the annealing model for fission tracks This window allows you to set the fission track annealing and compositional models as well as sample specific thermal history information for the current sample data file being created The Constrained point option is intended to be used with sedimentary samples for which you know the stratigraphic age and temperature at the time of deposition these can be input as a range If this option is selected then another time temperature point is used in the modelling to allow for the pre depositional thermal history It is assumed that the temperature of this pre depositional point is within the range specified for general prior of the total profile see the next section for discussion of this and the time is between the stratigraphic age and the maximum time specified for the general profile Even if the post depositional thermal history implies that the sample is totally annealed degassed this pre depositional time temperature point will still be incorporated it is not a significant computational overhead This stratigraphic age and present day temperature constraints Se are required for any sample with vitrinite reflectance and It iS you have entered VR data you will only the post depositional thermal history that
13. 95 credible range a little like the 2 standard deviation bounds as green lines calculated directly from the all post burn in samples in the MCMC chain Plot Offset Histograms This is enabled only when there is more than one sample in the profile and plots a histogram of the offset temperature parameter for either the present day or the palaeo offset temperature parameter only enabled if the offset parameter is set to be constant over time Plot Individual T t This plots the thermal history for a given sample selected from the drop down menu On this plot The maximum likelihood and maximum posterior models are shown as a yellow and magenta lines respectively The expected model is shown as a black line as are the 95 credible intervals The latter are calculated directly from the probability distribution of the model parameter i e the temperature at a given time and can be assymmetric if the distribution is not symmetric The black dashed lines are the credible intervals calculated as the expected model parameter 2 standard deviations and so are symmetrical Underlying these is a coloured plot showing the probability density of the thermal history effectively the probability that the thermal history passes through a 31 box of size 1 C x 1 million years The probability scale is shown on the right of the plot blue being low probability and red being higher probability Also shown on this plot are the maximum likelihood mod
14. CMC parameters described later this is controlled by the He Diffusion proposal scale parameter When you save the data file you will notice that the either He age or the He age error will be a negative value this indicates that you will use this sampling approach varying either the age of the error as described above 11 The U and Th ppm values are required if you want to use a radiation damage model to allow for preferential He retention For the diffusion calculation you need to enter the dimensions of the grain In QTQt we use the long axis length and the other 2 axes width thickness In terms of the the calculation a sphere is used and its radius is calculated from the three input dimensions under the assumption that the sphere has the same surface area to volume ratio as a rectangular grain If you enter the thickness as zero then it is assumed that the width and thickness are the same and we use the width value for both If you enter the width as zero then it 1s assumed that the length is the radius of a spherical grain For each grain you need to enter an activation energy Act Energy and diffusivity at infinite temperature Do and the helium ejection He ejection distance These will be initally set to standard default values for apatite with the Helium ejection distance being calculated following more or less Ketcham et al 2011 as a function of U Th and Sm contents if these are input In terms of the calculat
15. ENU iii 25 40 INO COUME daldara ann aN 10 lk E EEN 12 ING tte Datori 11 Offset Temperature History 32 Open Existing OTOt files S 8 Open Previous OTOt Run 9 Open Previous Summary File for plotting 10 Plot Individual FT sampling 34 Plot Individual HeVR sampling 35 Plot Individual PreDepositional sampling 35 Plot Individual predictions Expected MOE EE 33 Max Likelihood model 29 Plot Individual To 33 Plot Offset Histogranms 32 PLOTTING MENU 28 Post burn iN 25 Posterior Chain 28 Projected lengths 18 Proposal SCale EE 26 Ranges for General Prior SE 21 Resample He age with MCMC 12 Resample He Error with MCMC 12 Aisha 26 RUNNING OT OF egwgdieeeteeneguehedieeebite egen 6 Save LE WEE 18 Save data a e e DEE 24 Save for Rerun 27 Save Run Summary file as 9 Scaled Vector Graphics SV alain 28 Select Output Directory 9 Set MCMC parameters sees 25 Summary Model Predictions Maximum Likelihood model 30 Temp OFS CC sironnan 22 Thermal History EE teen teg 32 Maximum Dkelbood 29 THERMAL HISTORY CONSTRAINTS Ml 18 thickness RE rer Hoss
16. However if we continue let us look at the sampling in terms of the ee likelihood the fit to the data as a function of the MCMC iterations cia Choose the first option from the plot and you will see a plot similar to that Birth Death below on the left 0 0164 0 0174 The likelihood blue curve shows too much structure which indicates that SaveforRerun Cancel COK we need to do more iterations In the plot on the right we have the results of running 10000 iterations for the burn in and 10000 post burn in OTOtExample OTOtExample I I I Wal ss mi J d Woh II O Lag l s 3 i Finally if 3 d we run a d Foz 50000 j burn in and e 2s 50000 ee sco a a post burn in we have the results below Here we see less structure as well as more sampling between different dimensions or numbers of time temperature points the green curve This is what we expect to see OTOtExample INK i II GTE bc 70 000 80 000 90 000 100 000 iteration Likelihood Posterior Ho of Components For this example we can select the expected model thermal history choosing Plot Individual T t from the Plotting Expected Model and also the model predictions Summary Model Predictions and Plot Individual predictions we have plots similar to that below 38 ii Multiple samples As a second example we will consider a vertical profile the 8 files SynA1750HR txt SynA1500HR txt to SynAOHR tx
17. Note the age or error will be negative if you selected a resampling option for one of them you can only choose one or the other for resampling 12 He He data To include such data in the modelling process you need to have the data in a separate file in the following format modified from a format provided by David Shuster at University of California at Berkeley Sample name grain radius cm bulk U ppm bulk Th ppm agel delagel age2 delage2 0 00682 0 0060 148 0 235 0 1 0452 0 10 Istep Time step Temperature No of 3He atoms x10 6 error Rstep Rtotal error with R 4He 3He 1 260 00 0 38000 0 79000 0 040000 0 28477 0 059952 2 270 00 0 38000 0 38000 0 030000 0 50959 0 089927 3 290 00 0 51000 0 74000 0 040000 0 47961 0 059952 4 300 00 0 66000 0 85000 0 040000 0 50959 0 054956 5 310 00 0 66000 0 83000 0 040000 0 57454 0 059952 6 330 00 0 46000 0 95000 0 040000 0 56954 0 044964 fi 340 00 0 45000 0 96000 0 040000 0 62450 0 049960 8 350 00 0 48000 1 1300 0 050000 0 73940 0 049960 9 350 00 0 66000 1 1900 0 050000 0 70943 0 044964 10 370 00 0 53000 1 4800 0 050000 0 75439 0 039968 11 400 00 0 48000 2 4500 0 070000 0 79436 0 029976 12 410 00 0 50000 2 2400 0 060000 0 90927 0 039968 13 420 00 0 56000 2 1000 0 060000 0 95423 0 034972 14 440 00 0 63000 2 9400 0 070000 1 1341 0 039968 15 475 00 0 50000 3 8600 0 080000 1 2190 0 029976 16 500 00 0 50000 2 5800 0 070000 1 3289 0 039968 17 550 00 0 50000 1 4100 0 050000 1 2440 0 049960
18. QTQt User Guide Kerry Gallagher kerry gallagher univ rennes 1 fr Introduction QTQt is a program to infer thermal histories from low temperature thermochronology data using multiple samples The name comes from QT being Quantitative Thermochronology and Qt pronounced as cute or cutie being the software used to develop the user interface QTQt is currently implemented for apatite and zircon fission track apatite U Th He data and vitrinite reflectance although future versions should include Argon data too You can enter you own kinetics for any mineral isotope system combination e g to simulate zircon U Th He or mica argon data subject to the requirement that the diffusion domain can be treated as a single sphere whose size or equivalent dimesions of a rectangular crystal need to be sepcified also Also the current version allows for multiple sample modelling only if the samples under consideration have the same form of thermal history 1 e can be treated as a vertical profile see Gallagher et al 2005 You can still model a single sample but for generality we will still refer to a profile even if there 1s just one sample A future version will include the 3D partition model vertical profile approach developed by Stephenson et al 2006 The program uses the multicompositional algorithms of Ketcham et al 1999 2007 and the original Durango apatite based algorithm of Laslett et al 1987 for predicting fission track annealing in
19. arised below for each He age in terms of the proposed distribution left and the accepted right On the latter you will see the input He age and its uncertainty the black bars and the distribution of the predicted He ages the continuous black line He Age o 100 000 If Hierachical sampling of the input data errors has been selected then similar plot will be produced but the variable plotted will be the estimated error scaling parameter for each datum that has been sampled 8 000 6 000 4 000 2 000 45 50 55 200 000 300 000 Iteration He Age Sampling Synthetic SyntheticData txt 400 000 500 000 600 000 He Acc 0 He Acc 1 He Acc 2 He Sampled 0 He Sampled 2 10 000 8 000 6 000 4 000 Strat Age Plot Individual PreDepositional sampling This lets you examine the sampling of the stratigraphic ages temperature and pre depositional time temperature points The top two panels summarise these parameters as a function of post burn in iteration and the lower two the joint sampling of time and temperaure of Te ee which is useful to assess any correlation On l Lo KE SES these last two plots the maximum likelihood KEE tee Kee ML and the expected AV model values are fo Sr i ni ene also given i 5 Summary Model Predictions pe Hh This lets you plot the observed data circles and SAI 132 5 lig predicted values for the expected model cross for a si
20. ata file names where appropriate the individual data predictions and individual thermal histories for example The final part of the file name indicates in a reasonably obvious fashion which plot it contains The plot files will either be in Svg Scaled Vector Graphics or JPEG format 35 Appendix 1 Example of QTQt input file format OCO 04 07 Sample ID 0 1 0 2 198 X Y Z co ordinates Z in metres 210121 338 0 1 106e6 6131 No of time temp points Niengths Neounts zeta Paos Naos 105 code for annealing model 0 2 04 0 0 code for composition Value error on value 0 16 300000 code for initial track length code for projected track data 9 code for Cf tracks code for etchant 100 10 100 10 Time Otime Temp temp 20 20 Present Temp temp 61 600 3 0 FT age error on FT age Ma 12 80 0 1270 MTL error on MTL microns 1 275 0 1275 Std Dev error on Std Dev microns 61 135 Ns Ni must have Nounts values 16 264987 30 0 Individual track lengths must have Niengins values 15 561141 67 16 and angle to C axis if available see note 5 and compositional parameter if available No of AHe ages for the sample Code for Radiation Damage Model 11 35 32 61 106 51 0 0 51 4 3 160 2 45 6 38 2 He ncc gm or atoms U Th Sm ppm or atoms 20 0 005 138000 0 Age Ma error on age Grain length width 6 02 16 24 45 56 0 0 54 1 3 2 197 4 50 2 40 5 height microns The line below is a ejection 20 0 005 138000 0
21. ation itself The figures in this documentation are all taken from a version of QTQt running on a Macintosh so may differ visually on a PC DATA INPUT FORMAT The current input format required for QTQt is summarised in appendix and a series of example data files are provided with the installation package QTQt allows the user to input the data via the screen and this is the recommended method as there will be no problems with the format of the data files saved from the program which may not be the case if you create the files using just a text editor you do use a text editor a word of caution make sure that the values entered on the same line are separated by a single space or tab only MAC USERS On a Macintosh there have been problems with the default end of line character which 1s invisible Qt the environment used to develop the user interface apparently does not yet recognise the Macintosh end of line This can be a problem with Excel and TextEdit for example If the program crashes when you open a file created with a text editor on a Macintosh this is a likely explanation You can use TextWrangler the old BBEdit and use save as with Unix end of line characters The example files included are 1 QTQtexample txt this has AFT data and 3 AHe ages from a single sample with a simple reheating cooling history 11 A series of 8 samples in a vertical profile SynA1750HR txt to SynAOHR txt with AFT data and AHe ag
22. ax Likelihood Model is active and can be used see the later section on plotting Note when using sedimentary samples the stratigraphic age is used as a constraint in the thermal history models such that a given sample needs to be a surface temperatures at the time of deposition This may then lead to the thermal history being different to that input using the graphical approach above To confirm the actual thermal history used in the calculation use the Plotting Max Likelihood Model Thermal History menu option 22 Total profile This allows us to set up information used in the MCMC sampling for the thermal history for one or mutliple samples The window below will appear The window shows the oldest central age from all samples in the current profile as a guide to the time constraints we might want to use To set the Ranges for General Prior we need to enter a box in time temperature space defined by the time and temperature point in the middle of the box and the half width for time and temperature These are best set to be fairly broad However the values you do use will affect the solution you obtain as the broader the range the more likely we will end up with a fairly simple thermal history solution unless there is a lot of thermal history information in the data That is how transdimensional MCMC works If you are using multiple samples in a profile then the thermal history parameters you enter are relevant to the the hi
23. ch you know the stratigraphic age and temperature at the time of deposition these can be input as a range If this option is selected then another time temperature point is used in the modelling to allow for the pre depositional thermal history It is assumed that the temperature of this pre depositional point is within the range specified for general prior of the total profile see the next section for discussion of this and the time is between the stratigraphic age and the maximum time specified for the general profile Even 1f the post depositional thermal history implies that the sample is totally annealed degassed this pre 24 depositional time temperature point will still be incorporated it is not a significant computational overhead The Constrained Present day option can be used to set the present day temperature to be within given limits meantrange for any sample This 1s most likely to arise for well data If the Range parameter is set to zero the value will be fixed at the Mean value You can set the Annealing Model for a particular run and for the two Ketcham et al models you will need to set the Compositional Model parameter Value is required according to the type of kinetic parameter you want to use If you set the Uncertainty to 0 then the annealing model 1s chosen using this fixed value of the kinetic parameter If you set the Uncertainty to a non zero value then the compositional parameter is sampled from a normal dist
24. cted for the point in the Hot thermal history see figure below left You can choose any point to set this constant offset value To edit points in the Hot thermal history to change from a constant temperature offset again select and drag the point vertically and the point will change see figure below right 21 7 E zeg e u 7 Dp a eg ou Gees pe Save TRioSle EsxCold Pm aun Ser Anes Scales Opes Tiri file Save me Sie Edit Cold EdrHor Ran Set Anes Scales Single Thermal History Single Thermal History o EI i EI A Ee 8 Re a 3 a e l Ba z Bio 5 A too f 125 Le Ka we BR 100 G 120 Ta Ta 140 ro E A BLA BOALAR AOLE I RA E 5 20 o 140 120 100 a SC a a TI 120 300 x a Time Ma ii Cursor Position Cick mouse button in plot region Cursor Position Click mouse button in piot region You can not change the time values not can you delete a point while editing the Hot thermal history To do this you need to click on the Edit Cold option and move or delete the relevant time point for the cold blue thermal history Once you have created the forward thermal history for one or more samples you can select the Run option and a new window will appear summarising the results either for a single sample or the predictions as a function of elevation depth for a vertical profile For plotting the forward model predicted values for individual samples in a profile the plotting menu M
25. distance microns Do m 2 s and activation energy Jmol K Apart from sediments the temperature history can be ignored set the number of time temperature points to 0 If you want to fix a stratigraphic temperature or time point within a specified 0 range set the Otemp to a negative value If you set the errors to zero there will be a fixed time temperature point Remember these 0 values are the half width of the range used to sample the time temperature points You will also have to set the present day temperature value to a value and range otherwise the program may crash Also if the length data are in binned format enter a negative number for the number of track lengths Enter zero if there are no track length measurements code for annealing models 1 Laslett 1987 100 Ketcham multikinetic 1999 102 Laslett 1987 recoded by Ketcham 105 Ketcham multikinetic 2007 code for composition 0 Dpar 1 Cl apfu 2 OH apfu 3 Cl wt Value is that appropriate to the composition code If the error 1s non zero the value will be sample from a normal distribution centred on the valu with a standard deviation equal to the error code 1 calculate compositionally dependent initial length 0 use input value You need to give a value of the initial length whether it is used or not code 1 for inputting c axis projected lengths 1 means use the projected length model 1 just to read the angles but not use the
26. e for each sample In appendix 2 there are examples of input and output for these 2 sets of data so you can test these out quickly RUNNING QTQT Having launched QTQt you will see a menu bar over a main window as shown below Currently the number of samples in a vertical profile is limited to 20 The various menus are described as they appear on the menu bar Note that some menu options are disabled until certain parameters have been set If you try to run a different profile having already run 6 another one then many menu options enabled for the previous run will be disabled for the current one until you set the appropriate parameters for the current run FILE MENU Open existing QTQt data file s 30 Open previous QTQt run Select output directory Save Run summary file as Build QTQt data file Review existing QTQt data file Open previous summary file for plotting Open Existing OTOt files s This will produce a dialog window as below Clicking on Open file will produce a general open file dialog with a list of files in the current directory The usual options will apply to selecting multiple files at once e g on a Macintosh using shift will allow you to select consecutive files in the file list while shift command lets you select mutliple files that are not consecutive in the the file list Once you have selected one or more files from the first file list another dialog box will appear allowing you to select another
27. e paste option int the edit menu to do this The 3rd column labelled Comp allows you to input compositional data for individual grains in the same data file and this can be actual compostional data e g Cl Wt or proxy data such as Dpar In practice the average compositional value taken as the mean of all input values for both age grains and track length measurements will be used for all calculations for a given data file Finally you need to input Zeta the analyst specific calibration factor used in the External Detector Method Rho_d the dosimeter track density and Nd number of tracks counted in the dosimeter Clicking on the button will calculate the central age Having input the age data you then need to input track length data actually the order in which you do this is not important but you do need to do both even if there 1s no data as the file format expected when reading files will not be correct Fission track length data The next windon allows you to input track lengths and other length relevant data for a given sample The window below appears If you have no length data click on the No Length Data button Otherwise if you have already input information concerning the sample identifier and location this will appear automatically As with age data you can cut and paste values for the length and angle and the same caveat applies about generally having a single space between each value You can input just length meas
28. e samples which are further apart in the chain and therefore less likely to be correlated In general provided enough iterations are used the thinning parameter is not too important There are no hard and fast rules about how to choose these except that they should be large enough that the inference is reliable see Gallagher et al 2009 While the values can be too small to have achieved satisfactory sampling they can not really be too big although larger values will increase the computation time accordingly A qualitative check on whether the chain is sampled appropriately is to examine the likelihood or posterior values as a function of iteration We will return to this later The other parameters to set are the proposal scales for moving Time Temperature and if relevant Offset for sampling the FT annealing kinetic parameters He Diffusion data errors and Vitrinite 26 Refl data errors if these options have has been previously selected when building a data file and for Birth and Death that is creating deleting referred to as birth death new time temperature points including offset if the offset is allowed to vary with time For birth the time 1s selected randomly between two existing time temperature points and a temperature is calculated using linear interpolation between the two existing temperatures The birth proposal scale Temperature is used to perturb this interpolated value The same applies to the offset parameter if this i
29. ed files This provides a short cut to open a single file You can not open multiple files using this option Build OTOt data file This allows you to input FT single grain age counts length data U Th He age and vitrinite reflectancedata for a given sample You can optionally input track count and track length data as well as some other specific information concerning a given sample and U Th He and vitrinite reflectance data before being able to save the data to a file Fission track count data The window below appears first If you have no count data then click on the No Count Data button Otherwise you need to enter a sample identifier name and the location information The X and Y coordinates represent grid locations or longitide latitude but are not actually used in the current version Z is the elevation in metres enter negative Z for depth You can manually enter the spontaneous and induced track count pairs Ns Ni for a given sample or these can be cut from an standard application such as Excel by placing the cursor in the first row of Ns and pasting them into the table If you find that the values do not line up in two columns then it 1s because the character s between the Ns and Ni values in the original file are not a single space or a tab The simple solution to this 1s to replace what ever that character is with a single space PC USERS using the control v option does not seem to work for pasting but you can use th
30. ein 13 ROT 25 RTE TEE 21 Use Projected Eet 24 Use projected ack 2 50006s uniassadehoassnquaiaesedes 11 41 Resample Brror iis cciscsexieinadcaietses use for thermal hisory calibration width Crystal for A es ges thud ah ele oes
31. el red the maximum posterior model magneta and the mode model white Plot Individual Predictions This lets you select a single sample from the profile using a drop down menu and produce a summary plot of the observed data and the predictions from the expected thermal history for that sample This plot includes the track length distribution observed and predicted with the 95 credible intervals on the predicted values This plot also summarises the output of any sampling on the kinetic parameter and He ages as described below Each value is indicated by the following codes O observed P Predicted SP sampled values of the Predicted value SO sampled values of the observed value 32 Ne of tracks mi S S Synl Syn txt 1 000m Expected LLa 1560 01 FTA 008358 Abel MTL O 11 78 Pi TL78 EP 11 09040074 la ijeri back 101 0 j Ha 150101 PS 85 SPs 1S ed ae h Lem He 26 76 P 237 65 GP 26 75 0 351 He 33 85 P 34 60 SP 31 5340408 el i Plot Individual FT sampling This allows you to examine the sampling of the FT compositional parameter if this has been selected as a variable by adding an uncertianty to the input value Each sample can be examined individally A typical plot is shown below The top left panel shows the sampling of the kinetic parameter as a function of post burn in iteration the proposed and accepted values although often these are the same as in the example bel
32. emperature points or the complexity of the thermal history solutions are inferred from the data rather than being specified in advance The development of the method for thermal history modelling is given in Gallagher 2012 and some other relevant publications are Gallagher et al 2009 Charvin et al 2009 Hopcroft et al 2007 and Sambridge et al 2006 The approach as implemented in QTQt allows the user to specify one general time temperature box from which time temperature points are sampled to construct a continuous thermal history by linear interpolation between the sampled points It also allows for up to 5 additional time temperature boxes to be specified to allow the user to add more specific constraints on the thermal history It 1s possible to select resampling schemes for the observed data or the errors on the observations it it is considered they are not well known or could be more noisy than the default values may suggest This is sometimes referred to as Empirical or Hierachcial Bayes and examples of this are given in Gallagher et al 2011 When the program is actually running there is little to see just a progress bar in the top level window The results of the run are saved to a file which is deleted when you quit the program unless you choose to save it and this summary file is used for generating a series of plots to examine the results once the run has finished QTQt was written in C and C by Kerry Gallagher ker
33. enu which we will come to below Finally you can add in some constraints to the general thermal history that are applied directly to the highest elevation or shallowest in a well sample and these are then imposed on the other samples by using the offset temperature To do this check the Constrain box and add in the parameters defining a time temperature box as described above as shown below You can add up to 5 constraints although this is a little against the spirit of transdimensional MCMC Note that you need to add these points is order of decreasing time to the present day However these points are independent to the general thermal history prior desscribed above so can be older or younger or within the range defined by the general prior If you want to save the thermal history information to a file for later runs you can do that by clicking in the button You will be prompted for a file name If you want open a saved thermal history file click on the Again you will be prompted for a file 23 The parameters are checked automatically to make sure they are valid you can not enter negative values for examples and the OK button will not be enabled untill all the relevant parameters have valid values Default values are loaded when the window first appears and if you want to use them you may have to just edit one value to enable the OK box Once you have entered at least the minimum information required for the thermal his
34. file using Open another file You can keep doing this and when you have selected all files for a particular run you can select Finish When you click on the finish button or if you have loaded a file from the previously opened file list at the bottom of this menu the run title window will appear This lets you give a name or identifier to the current modelling run This name is used for all output files and will appear on all of the plots subsequently When multiple samples are used in a profile sample specific results will also have the sample s file name This name 1s also used for a file with the appendix run which will contain the filenames and thermal history modelling parameters the last model sampled and the best model found during the current run This can be reloaded as described below so you do not have to enter all the parameters again Open Previous OTOt Run This allows you to open a file containing the filenames and thermal history modelling parameters the last model sampled and the best model found during a previous run This file will have the name of the modelling run see Run Title above and the suffix run It 1s saved into either the directory where the QTQt application is located or a user specified output directory see Select Output Directory below If no name is enter then the default is to save the information for the current run to a file called QTQt run in the current directory Only files with the suffi
35. fset parameters By changing the scaling parameters the acceptance rates for the thermal history parameters can be improved to lie within this range In general if you need to increase the acceptance rate you should decrease the scale parameter However as the thermal history parameters are not independent in terms of how they affect the better data fitting models and also the prior ranges set on the time temperature and offset parameters will influence the acceptance rate then this not always true Note that if you are modelling a single sample the offset parameter is not used and the acceptance rate is zero Run This runs the MCMC sampler and you will see a standard progress bar to indicate where you are as below You can cancel the run at anytime and any subsequent inference or plotting will be made using the number of post burn in samples at the time you cancel the run Once the run is finished you will see a window similar to the Set MCMC parameters window but this time the acceptance values will be for the run just completed This time however you can review the acceptance rates and change the proposal scale parameters and or the number of iterations 1f desired This window has the option Save for Rerun to save these new parameters for a subsequent run Normally you will do this next run immediately You need to click on the OK button and then 27 use the Run menu option again Note that if you change the parameters and do
36. g models parameters for Use Projected lengths Cf tracks or Etchant if desired and the appropriate adjustment will be made to the annealing model predictions Note if you use the Laslett et al 1987 the options for Cf tracks projected lengths etchant and the compositional parameters will not be used The same 17 applies to the 3 zircon models and for these the appropriate default value for the initial track length will be inserted If the sample has He analyses then the number of ages 1s summarised together with the choice of alpha radiation damage model one of none Gautheron Flowers Note this will be set to none if there are no U Th and He concentration data enter for any analysis from a given sample as the U and Th concentrations are required for the alpha radiation damage models Finally if you are building a QTQt file for the first time or modifying an existing file you can Save data for reload and you will be prompted for a file name If you do not select this the changes will not be saved It is recommended to build all the data files you might need for example samples from a vertical profile or borehole suite and save them as you build them When you want to proceed to running thermal histories models it is probably best to quit QTQt and restart it 18 Review existing OTOt data file This allows you to open an existing data file and check or edit the values This process follows the same progression as desc
37. ghest elevation or shallowest in a well sample For multiple samples the thermal history for lower elevation or great depth in a well samples has the same form as that for the shallowest sample but is offset by a value to be determined during the modelling The offset parameters are the temperature difference between the lowest and highest elevation or shallowest and deepest samples over time The thermal history for samples between these two is obtained by linear interpolation based on the difference in elevation or depth If you are running multiple samples you will also need to set a prior range on the offset If you have been running a single sample profile then decide to use multiple samples you might see the following dialog box This is just to remind you that you need to see the Temp offset parameter prior range If you do not as an additional reminder the OK button in the General Thermal History window will not be enabled By default the offset values for each time temperature point are the same for all points except the present day but currently the same prior range in offset is used for the present day offset the option to use a different range has not yet been enabled If you want to let the offset vary between all time temperature points the you can check this option in the window If you want to fix the present day temperature for various samples in a well for example this is possible by using the Individual Sample m
38. h R Yamada R and Laslett G 1998 Revised annealing kinetics of fission tracks in zircon and geological implications In Van den Haute P and De Corte Eds Advances in Fission Track Geochronology Kluwer 99 112 Yamada R Murakami M and Tagami T 2007 Statistical modelling of annealing kinetics of fission tracks in zircon Reassessment of laboratory experiments Chemical Geology 236 75 91 INSTALLING QTQT If you are reading this you have probably already installed QTQt MAC USERS On a Macintosh the installer by default puts the application QTQt some example files and this documentation file into a directory called QTQt the location depends on what you choose for the install directory and installs 3 libraries into the system directory Library Frameworks or into your home directory under Library Frameworks You can move the location of the QTQt application but the libraries QtCore QtGui QtSvg must not be moved from the default location If you wanted to do a manual install by copying files then you just need to copy the 3 library files to Library Frameworks and you can put the QTQt application wherever you want Note that if you change from one version to another you will probably need to remove the framework libraries that are already installed to make sure the versions are compatible PC USERS On a PC I have set up QTQt such that all the required libraries need to be in the same directory as the applic
39. he second case you will initially see the Edit Cold option in colour and Edit Hot greyed out Here Cold refers to the thermal history for the top or shallowest sample in a vertical profile and Hot refers to the lowest or deepest sample These two thermal histories Cold and Hot will have the same number of time temperature points and the same time points They only differ in the temperature values you assign The thermal histories for all samples between the top and bottom samples are determined by using the elevation depth differences and linear interpolation of the temperature offset between the top and bottom thermal histories You have the option to save an input thermal history to a file Save T t to file or open a previously created thermal history file Open T t file Set Axes Scales allows you to change the range on the time and temperature axes 20 We start the input the top thermal history by placing the cursor in the white part of the window If you click the mouse and hold it pressed down you will see cross hairs in the window and the time temperature points given by the position of the cursor is written at the bottom left of the window You select a point by releasing the mouse click The thermal history will be drawn in blue adding in each point as a as you create them de RM e Q im File awe Tihi np bie ast Cold ada Single Thermal History Iemmeme mae Ge Ge Ga ERR ER a a a mi 140 130 Dog
40. ice of maximum posterior model is then sensitive to the range of the prior specified for the general thermal history model The larger the range on the priors the more we encourage the posterior models to be simple 1 e the more we penalise models with a larger number of time temperature points Max Mode Model This model is obtained from the distribution of all models sampled In 1 D the mode is the peak of a 1 D probability distribution If we divide the time temperature space into squares with a resolution of 1 million year and 1 C then for all temperature paths we can count how many temperature paths pass through each square The mode model is the temperature value at each 1 million year step that has the most number of paths passing through it The plot options for this model are described below for the Expected Model Expected Model In Bayesian formulations as adopted in QTQt the preferred single model is the Expected Model This is effectively a weighted mean model where the weighting is provided by the posterior 30 probability for each model This model contains features of all the models sampled in the post burn in sampling and in terms of complexity will generally lie between the maximum likelihood model more complex and the maximum posterior model less complex Also we can use the MCMC sampling to calculate the uncertainty for the expected model and so draw meaningful credible intervals more or less the Bayesian equiva
41. ion the input age should be the uncorrected 1 e no FT correction age as the He ejection effect is calculated as part of the thermal modelling If you want to use the corrected age then set the helium ejection distance to zero If the He ejection distance is not zero and you have input an age not equal to that calculated from the input U Th and He then the input age will be replaced by the calculated age To avoid this set the He ejection distance to zero Finally you can select one of two published models to allow for the possible effect of alpha damage on the diffusivity of He Note this will not be available if there are no U Th and He concentration data enter for any analysis from a given sample You can add another age Add more and you can have up to 25 separate He ages for a given sample As mentioned at the beginning of this section you can input any single domain diffusion data e g U Pb ages from apatite as well as ZHe ages You just need to put in the ages grain dimensions and the appropriate diffusion kinetics Do and E Set the alpha ejection distance to zero if appropriate The He age data and parameters in a datfile can look as below No of He ages 0 No radiation damage 1 Gautheron 2 Flowers 51326 23 72 0 0 0 15 97 0 64 50 0 0 0 0 0 He U Th Sm Age Error length width thickness 20 66 0 0032 138000 0 He ejection Do Act Energy 0 0 0 0 0 0 0 0 29 0 0 5 100 0 0 0 0 oe 20 0 0 0032 138000 0 E
42. lent of confidence intervals These intervals represent a 95 probability range for a given parameter calculated so that 2 5 of the parameter values lie below and above the limits defined by the range The plot options for the Expected Model are a little different than the other two models The drop down menu 1s below Thermal History The thermal history here represents the average of all the models sampled If the profile contains more than one sample then the uppermost sample thermal history will be plotted in blue and the lowermost sample thermal history will be plotted in red The thermal histories for all samples in between are drawn in grey For the uppermost sample the 95 credible intervals are draw in cyan light blue and these reflect the uncertainty in the inferred thermal history alone For the lowermost sample the 95 credible intervals are drawn in magenta and these reflect the combined uncertainty in the inferred thermal history and also the offset parameters Any constraints will be drawn as black boxes Offset Temperature History This is enabled only when there is more than one sample in the profile and plots the offset temperature between the upper and lower samples in a profile as a function of time During the MCMC run the mean and standard deviation of the offset temperature is calculated as a function of time The mean is plotted as a red line and the 1 standard deviation bounds as magenta lines Also shown is the
43. less precise high scaling factor relative to the input error value This may be useful when you are not sure of how good you error estimates are Note in the datafile the input value will be negative if you choose to Resample VR or the error will be negative if you choose Resample Error 15 There is a conversion of the calculated time temperature integral to equivalent VR VR exp a bF and again the default values for a and b are for EasyRo Finally you can also enter a function Calculate Ln A c d Eact to calculate the frequency factor A as a function of the activation energy Eact This needs to be a log linear relationship and the default parameters c d 0 0 are equivalent to EasyRo In a data file the Vitrinite Reflectance data and parameters typically look as below No of VR observations resampling has been selected 1 or 0 indicates used as a constraint or not 1 3 3 700 142 256 0 030000 3 1536 e 26 150 624 0 030000 3 1536e 26 158 992 0 040000 3 1536e 26 167 360 0 040000 3 1536e 26 175 728 0 050000 3 1536e 26 184 096 0 050000 3 1536e 26 192 464 0 060000 3 1536e 26 200 832 0 040000 3 1536e 26 209 200 0 040000 3 1536e 26 217 568 0 070000 3 1536e 26 225 936 0 060000 3 1536e 26 234 304 0 060000 3 1536e 26 242 672 0 060000 3 1536e 26 251 040 0 050000 3 1536e 26 259 408 0 050000 3 1536e 26 267 776 0 040000 3 1536e 26 276 144 0 030000 3 1536e 26 284 512 0 020000 3 1536e 26 292 880 0 020000 3 1536e 26
44. limate from borehole temperature data using Bayesian Reversible Jump Markov chain Monte Carlo Geophys J Int 171 1430 1439 Ketcham RA Carter A Donelick RA Barbarand J Hurford A J 2007 Improvedmodeling of fission track annealing in apatite American Mineralogist 92 799 810 Ketcham R A Donelick R A Carlson W D 1999 Variability of apatite fission track annealing kinetics III Extrapolation to geological timescales American Mineralogist 84 1235 1255 Ketcham R Gautheron C and Tassan_Got L 2011 Accounting for long alpha particle stopping distances in U Th Sm He geochronology Refinement of the baseline case Geochim Cosmochim Acta 75 7779 7791 Laslett G M Green P F Duddy I R and Gleadow A J W 1987 Thermal annealing of fission tracks in apatite 2 A quantitative analysis Chem Geol 65 1 13 Sambridge M Gallagher K Jackson A and Rickwood P 2006 Trans dimensional inverse problems Model Comparison and the Evidence Geophysical Journal International 167 528 542 Stephenson J Gallagher K and Holmes C 2006 Low temperature thermochronology and modelling strategies for multiple samples 2 Partition modeling for 2D and 3D distributions with discontinuities Earth Planet Sci Letts 241 557 570 4 Sweeney J J and Burnham A K 1990 Evaluation of a simple model of vitrinite reflectance based on chemical kinetics AAPG Bulletin 74 1559 70 Tagami T Galbrait
45. load and OK buttons will not be enabled until valid input parameters have been set for the particular annealing model selected This applies also for the thermal history constraints which must have positive temperature values 25 MCMC RUN MENU This menu lets you set parameters controlling the MCMC sampling and run the sampler for thermal histories Set MCMC parameters When you select this menu option the window below will appear QTQt MCMC parameters MCMC chain Burn in Post burn in Thinning 10000 10000 E Proposal Move Time Temperature Offset 15 0 15 0 5 0 FT Annealing He Diffusion Vitrinite Refl un 10 10 deenen Temperature Offset son en Acceptance rates Time Temperature Offset 0 0 0 0 0 0 FT Annealing He Diffusion Vitrinite Refl 0 0 0 0 0 0 Birth Death 0 0 0 0 Cancel OK The sampler for MCMC has 3 parameters to set controlling the number of iterations used in the sampling chain The total number of iterations is equal to the Burn in Post burn in The Burn in is the number of iterations which will be discarded while the Post burn in is the number of iterations that will be used in subsequent inference of the thermal history Thinning is a parameter that controls how many of the number of Post burn in iterations are used If the parameter is set to 1 then all sample will be used if it is set to 5 then every 5th sample will be used if it is set to 10 then every 10th sample etc Larger values us
46. m 0 for no angles and 2 for already projected track lengths with no angle data code 1 Cf tracks 0 no Cf tracks 7 code 0 for 5 5 Molar Donelick 1 for 5 Molar everyone else For binned length data enter 20 bin interval values 1 20 and the frequency for each interval code 0 no Radiation Damage Gautheron et al 2009 2 Flowers et al 2009 Age error set age error to a negative value to sampling from a distribution centred on the observed age input error The AHe data need to be in this format i e the concentrations or number of atoms but at the moment I use just the uncorrected age and the error Alpha ejection is dealt with when running the diffusion model as in Meesters and Dunai For the grain geometry you need 1 2 or 3 values I do the calculation for a spherical grain which just requires the radius as the length parameter set the width height 0 0 If you input the last height 0 0 I assume this is equal to the width I convert this to a sphere with the same surface are to volume ratio To sample the He age set the error to a negative value 36 Appendix 2 Examples of typical input and output from runs with the example files provided note these results may differ from a run you make using these files i Single sample QTQtExample txt QTQtExample txt If you open this file then select the Thermal History Open Existing Thermal history Abe _ Save Current
47. ngle value or continuous line for FT age Th Test MT 2360 131 5 Strat Age e Stratigraphic Age Temp AV ML 60 Time Temperature EM 130 5 8 000 6 000 4 000 2 000 0 8 000 6 000 4 000 2 000 10 000 8 000 6 000 4 000 E wi funiadual pn L era qa ee LT 100 000 120 000 140 000 160 000 180 500 200 000 130 He Accepted 0 He Accepted 1 FTE RE SE EE RE pere 45 50 55 60 He Accepted 2 Predeposition Test MT 2360 Predep Time E i Gi Loi fe mi La TTT TTT TETTE a 160 000 120 000 140 000 160 000 180 000 200 000 iteration Time Temperature PreDeposition Test MT 2360 3 Jtt d Predep Time Predep Time Temperature AN ML blue FT MTL red He age green as a function of elevation for all the samples in the profile If the kinetic parameter has been varied this will also be shown yellow curve relative to the axis for the MTL 34 njesacduey dapaig Summary Information This option is not currently enabled Generate all plots This option automatically generates and saves to files the available plots You may be prompted to save the run summary file although this is not required so you can select cancel The files are saved to the directory where the QTQt application is running or to a previously selected directory The file names will start with the run name you selected and use the individual d
48. ow The top right show the variation in the predicted FT age and MTL note this also is a function of the thermal history The lower panels show the distribution of the proposed green and accepted magenta values the latter have the mut value with the 10 range and the distributions of the predicted FT Age and MTL with the black line indicating the input value Kinetic parameter Synthetic SyntheticData txt Age Mtl Synthetic SyntheticData txt 112 12 8 12 6 X pa a rm 224 106 Si EA m lt 12 2 2 2 104 2 102 1 8 100 11 8 0 100 000 200 000 300 000 400 000 500 000 600 000 0 100 000 200 000 300 000 400 000 500 000 600 000 Iteration Iteration Kin Sampled Kin Accepted Age MTL Kin Sampled Kin Accepted FT Age predicted MTL predicted i i 8 000 8 000 ege 8 000 8 000 6 000 6 000 6 000 6 000 4 000 4 000 4 000 4 000 2 000 2 000 2 000 2 000 0 0 0 0 100 102 104 106 108 110 12 12 2 12 4 12 6 12 8 33 Plot Individual He VR sampling This allows you to examine the sampling of the He ages for a given sample The top panel summarises the sampling as a function of iteration of each He age for a given sample there will be on such plot for each group of 5 He ages The coloured curves for each sample are the proposed and accepted He ages the proposed are drawn as a dotted line and black curve represents the predicted He age a function of the thermal history The distributions are summ
49. ribed above for creating a new QTQt data file 19 THERMAL HISTORY CONSTRAINTS MENU Draw forward thermal history Before running the modelling we need to set up the constrants on the thermal history relevant to all samples and if required desired any sample specific information Thus the run option is not available until we have at least chosen to set up a forward model or set the general thermal history constraints for an inverse modeling run Also to activate these menus we need to have already opened one of more data files as these contain data and information concerning the data we need for the modelling Draw forward thermal history This allows us to use a graphics window to set up a forward model for one or more samples in a vertical profile After selecting this option a window will appear in the form of the one of the two possibilities shown below D a 0006 Save Tito file Cd Colg EditHot Bun Set Awes Sale D O a Open Ti file Save Tito bie Run Set Anes Scales Single Thermal History Single Thermal History quer rr LEES ELE EELER EELER ELE EE DEL ERR E ER BO Ed Time Ma BO Ed Time Mai Cursor Position Click mouse button in plot region e button in plot region If you have opened only one data file you will see the window on the left while if you have opened more than one for a vertical profile you will see the window on the right In t
50. ribution with a mean equal to Value and a standard deviation equal to Uncertainty This is one way to allow for uncertainty in the annealing models You can set the Initial track length to a specified value Note that for the Ketcham et al annealing models the default is to Calculate initial track length using the compositional information You can uncheck this option if you do not want to do this You can modify the annealing models parameters for Use Projected lengths Cf tracks or Etchant if desired and the appropriate adjustment will be made to the annealing model predictions Note if you use the Laslett et al 1987 the options for Cf tracks projected lengths etchant and the compositional parameters will not be used The same applies to the 3 zircon models and for these the appropriate default value for the initial track length will be inserted If the sample has He analyses then the number of ages 1s summarised together with the choice of alpha radiation damage model one of none Gautheron Flowers Note this will be set to none if there are no U Th and He concentration data enter for any analysis from a given sample as these are required for the alpha radiation damage modelling Finally if you are building a QTQt file for the first time or modifying an existing file you can Save data for reload and you will be prompted for a file name If you do not select this the changes will not be saved Note that the Save data for re
51. riginal default values Examine Chain These plots let you examine the performance of the chain in terms of the log likelihood Likelihood Chain or log posterior Posterior Chain as a function of post burn in sampling blue curve and use the left hand axis On the same plot you will see the number of time temperature points green curve and use the right hand axis There should be no obvious trend in the likelihood posterior i e the mean value should be pretty much flat and the values should change almost every iteration De it should not get stuck on the same value for too many iterations Belwo the first panel shows a chain with poor convergence while the other two are well mixed Not the right hand panel is the posterior and this tends to be lower as the number of model parameters increase Max Likelihood Model This option lets you examine details of the maximum likelihood model that is the best data fitting model The philosophy adopted here is that generally this model is likely to be too complex that is it may have features which are not really justified from the data However it can be useful to examine the model Thermal History This plots the maximum likelihood thermal history There are no uncertainties associated with this single model 29 The figure shows an example of a thermal history for 3 samples in a profile The blue curve is the reference thermal history highest elevation sample and the red curve is thi
52. ry gallagher univ rennes1 fr although some of the implementation of the fission track annealing algorithms is based substantially on subroutines provided by Richard Ketcham and Rich s co operation is gratefully acknowledged We also acknowledge QT for providing such a professional programming environment for free This software would not exist without them QTQt has been developed on a Macintosh using OS X 10 6 Snow Leopard The 64 bit version built with Qt 4 7 0 will run on machines with OS X 105 Panther or higher and requires an Intel processor 32 or 64 bit The 32bit version built with Qt 4 5 2 will run on at least OS 10 4 Tiger and also run on the older PPC processors although is likely to be pretty slow The PC version has been tested on various operating systems Windows XP Vista but not extensively If you have problems with any version i e it crashes then feel free to contact Kerry Gallagher Try and detail as much as you can concerning what you did and how it crashed Also send the data files you were using as that seems to be the most common fault Some platform specific issues are highlighted in this documentation as PC USERS or MAC USERS However read this documentation closely before trying to run the software on your own data if it 1s not clear please contact Kerry Gallagher who 1s happy to change things to make them clearer REFERENCES Charvin K Gallagher K Hampson G amp Labourdette R
53. s allowed to vary over time using the Offset birth proposal scale For moving the parameters the scale parameters in the same units as each parameter type are in fact the standard deviation of a normal distribution used to generate new values of the time temperature or offset during the sampling chain For the birth parameters these represent the width of a uniform distribution between 0 5 and 0 5 In the terminology of MCMC we take the current model choose a parameter from this model then peturb that parameter using a random sample from a normal or uniform distribution centred on the parameter value of the current model and a standard deviation equal to the value input here This concept 1s explained further in Gallagher et al 2009 The remaining boxes in the windows are not enabled but will summarise the acceptance rates for the sampling from the most recent sampling run These rateswill be between 0 and 1 and are useful to assess the performance of the sampler The first 3 summarise the acceptance rates for the time temperature and offset parameters The birth and death terms reflect the acceptance rate for adding a new time temperature point or deleting an existing time temperature point respectively As with the number of iterations there are no hard and fast rules about what values are optimal for the acceptance rates However a general rule of thumb is a value between 0 1 and 0 7 is probably OK for the time temperature and of
54. s plus the offset parameter for the lowest elevation sample Intermediate sample thermal histories are shon in grey Plot Individual predictions This lets you select a single sample from the profile using a drop down menu and produce a summary plot of the observed data and the predictions from the maximum likelihood thermal history for that sample This plot includes the track length distribution Summary Model Predictions This lets you plot the observed data circles and predicted values cross for a single value or continuous line for FT age FT MTL He age as a function of elevation for all the samples m the profile Summary Information At the moment this option does nothing but has been included for future developments Max Posterior Model This has the same options as described above for the maximum likelihood model but for the maximum posterior model The posterior probability is proportional to the likelihood multiplied by the prior For models of constant dimension 1 e the number of time temperature points and for uniform prior distributions used in QTQt the maximum likelihood and posterior models will be the same However QTQt implements a transdimensional MCMC sampler the number of time temperature points is not fixed Then the prior acts as a penalty against making the model too complex and the maximum posterior model will generally be simpler less time temperature points that the maximum likelihood model The cho
55. t represent samples at intervals of 250 m and each sample has AFT data and one AHe age These data were generated with a thermal history equivalent to that used in Gallagher et al 2005 their figure 3 Using the default values for the Total profile thermal history constraints the likelihood plot should resemble that on the left again showing too much structure and not much transdimensional sampling Running for 20000 iterations for the burn in and 60000 post burn in gives us the plot on the right and much better sampling Overall we can have too few iterations but never too many except for the amount of time it may take In this case the data are high quality and the solution is well constrained so the burn in period 1s relatively short Below are the plots for the Expected Model Thermal History and Summary Model Predictions The thermal history starting around 110 Ma at temperatures below the total annealing temperature for the shallower samples implies rapid cooling from above these temperatures immediately before this time If we add a constraint on the total profile thermal history as shown in the dialog to the right this constraint is possible as we know the solution Running the profile data again we have the results shown below The fit to the data does not change much and the form of the thermal histories after 100 Ma are similar This shows there is little information in the data concerning the early part of the thermal
56. tory e g the general thermal history prior range and offset if required the OK button will be enabled Individual Sample If you are building a QTQt data file from scratch or want to change some individual sample settings such as the annealing model parameters you need to use this menu option A drop down menu will appear with the current loaded files or the current run if you are building a new QTQt file When you select a file a window will appear as below OTOt Indiviudal Sample information Sample Bra827 txt Central age Ma MTL microns 5D microns 284 89 12 95 1 599 Constrained point ven Range Time m y Temperature C _ Constrained Present day Temperature C range Annealing Model Compositional Model Ketcham et al 2007 v Dpar microns Value _ Ketcham et al 1999 Cl Mw 2 0 _ Laslett et al 1987 Cl tapfu Uncertainty J Tagami et al 1998 1 OH apfu 0 1 _ Yamada et al 2007 Long term _ Yamada et al 2007 Short term Initial track length 16 30 v Calculate Initial track length Wi Use Projected lengths Cf tracks Etchant V5 0 Molar C 5 5 Molar No of Helium 2 No of Argon Alpha Damage Gautheron i Save data for reload Cancel m im This window summarises the information concerning the thermochronological data for the sample and lets you add sample specific thermal history information The Constrained point option is intended to be used with sedimentary samples for whi
57. urements or lengths and angle to c axis data and in this case you can choose to use a project track length annealing model or not see Ketcham et al 1999 2007 As with the age data the Comp field allows for individual compositional data for each track length measurement and the mean value will be calculated and output to the final data file Select Use projected tracks to use a projected track length model obviously this makes sense only if you have put in angle information or you have previously transformed length data into c axis equivalent lengths not using projected tracks is the default Select the appropriate Etchant 5 M is the default Select Cf tracks if your sample used Cf irradiation to increase confined track revelation the default is no Cf tracks IF you want to return to the Counts data window click on that button 10 U Th He data and He He data To incorporate U Th He data or any single domain diffusion type data the window below will appear Helium Data x Y z Sample No He Data Aaa EE Resample He age with MCMC SO m Resample He Error with MCMC Calculate Age uncorrected U ppm Th ppm Sm ppm He ncc gm Grain dimensions length microns width microns thickness microns Act Energy kj mol K Do m 2 sec He ejection microns 138000 0 0032 20 0 No of He Ages 1 a Add more Cancel OK Return to Lengths If you have no U Th He data click on the No
58. x run will be available to open When you have selected to open a previous run file you will see the following dialog window This allows you to start the model run from the last model or best data fitting maximum likelihood model of the previous run or start from a completely random thermal history model The run title window described above will appear after this window disappears Select Output Directory This allows you to select or create a directory for all output from QTQt for a given run Save Run Summary file as A text file 1s created during a model run and all the results are output to this file By default it is deleted when you quit QTQt If you want to save the file select this option and give you preferred name to the file This file is created once you run the thermal modelling in QTQt so if you are running several individual samples and want to save the file for each you will need set the name to different values between sample run If you have selected the output directory above this file will be saved to that directory This also saves some other files to that directory which need to be kept it you want to produce plots later as described directly below Open Previous Summary File for plotting This lets you open a Run Summary file as described above to generate plots if you decide you did not save ones you wanted previously If you have opened files previously with QTQt you will have a list of previously open
Download Pdf Manuals
Related Search