Home

Details

1. 02 no_name 014 0 882000 0 932184 5 018437E 02 1 134 no_name o15 9 42500 9 90684 0 481840 0 1060 no_name ol6 0 601500 0 648450 4 694962E 02 1 663 no_name ol7 51 3300 53 3843 2 05426 1 9000E 02 no_name 018 5 89550 5 17830 0 717199 0 1700 no_name 019 47 3950 48 4369 1 04190 2 1000E 02 no_name 020 4 44250 4 62304 0 180537 0 2250 no_name 021 45 0650 43 9959 1 06909 2 2000E 02 no_name 022 3 95100 4 12910 0 178099 0 2530 no_name 023 23 1400 21 3140 1 82600 4 3000E 02 no_name 0o24 1 64700 1 70735 6 034768E 02 0 6070 no_name 025 10 8950 8 98367 1 91133 9 2000E 02 no_name 026 0 671000 0 544736 0 126264 1 490 no_name 027 3 13500 3 26320 0 128203 0 3190 no_name 028 0 152500 0 131795 2 070505E 02 6 557 no_name 029 1 44000 1 24259 0 197406 0 6940 no_name 030 3 050000E 02 3 866613E 02 8 166130E 03 32 79 no_name 031 0 450000 0 399623 5 037676E 02 2 222 no_name 032 0 00000 1 077987E 02 1 077987E 02 1 000 no_name 033 0 150000 0 162574 1 257429E 02 6 667 no_name 034 0 00000 4 019330E 03 4 019330E 03 1 000 no_name See file EXAMPLE RES for more details of residuals in graph ready format See file EXAMPLE n EO for composite observation sensitivities Objective function gt Sum of squared weighted residuals ie phi 0 4297 Correlation Coefficient gt Correlation coefficient 0 8997 Analysis of residuals gt All residuals Number of residuals with non zero wei
2. 3 Vector_4 Vector_5 fsne 1 4335E 02 0 9842 0 1744 2 2879E 02 1 1198E 02 crd 0 9999 1 4279E 02 1 5744E 03 1 9999E 04 6 4642E 05 dt50 9 4905E 04 0 1745 0 9453 0 2229 0 1618 masini 7 8350E 05 1 4702E 02 0 2006 0 9685 0 1467 met 7 0894E 05 1 9859E 02 0 1889 0 1084 0 9758 Eigenvalues gt 4 7821E 07 1 7663E 03 0 1313 3 753 8 837
3. Appendix Results last section of rec file These are the results of the default example provided with the package OPTIMISATION RESULTS Parameters Sss gt Parameter Estimated 95 percent confidence limits value lower limit upper limit fsne 0 600286 0 408681 0 791892 crd 1 226317E 02 9 886449E 03 1 463989E 02 dt 50 13 0062 11 5100 14 5024 masini 56 0664 52 1243 60 0084 met 108 755 102 806 114 705 Note confidence limits provide only an indication of parameter uncertainty They rely on a linearity assumption which may not extend as far in parameter space as the confidence limits themselves s PEST manual See file EXAMPLE n EN for parameter sensitivities Observations gt Observation Measured Calculated Residual Weight Group value value ol 51 6300 55 5316 3 90156 1 9000E 02 no_name o2 5 72850 5 39742 0 331082 0 1750 no_name o3 50 5900 53 4684 2 87844 2 0000E 02 no_name o4 5 05600 5 08081 2 480823E 02 0 1980 no_name o5 46 0200 46 2635 0 243511 2 2000E 02 no_name o6 3 66350 4 06320 0 399703 0 2730 no_name o7 38 6100 38 2669 0 343149 2 6000E 02 no_name 08 2 93200 3 2 0 93 90 0 161897 0 3410 no_name o9 32 8150 29 0458 3 76922 3 0000E 02 no_name 010 1 92800 2 16483 0 236828 0 5190 no_name oll 25 8700 20 9646 4 90536 3 9000E 02 no_name 012 1 46500 1 47675 1 175005E 02 0 6830 no_name 013 20 3150 13 8380 6 47702 4 9000E
4. K EXE gt The PEST optimisation software is available in the pest directory As PEST is now available freeware http www sspa com pest we included the latest version as of 29 10 2003 version 7 0 1 Separate installation of PEST is not necessary gt The Neq_Example directory contains results from an example study as described in the PEARL user manual page 51 54 You can now run the example to check if everything works and get experience with the system Go to the Neq_Example directory and run the example example bat gt The batch file will first call PEARLMK This pre processing program generates the input files for PEST i e example pst example tpl and example ins gt Than the optimisation starts PEST calls PEARLNEQ several times in the example 39 times gt If you get an error message after the first step PEARLMk type control break to stop the process and check the error messages available in the example err file After successful optimisation read the results from the file example rec The relevant results including parameter values 95 confidence intervals and correlation matrices can be found at the end of this file PEST also generates parameter sensitivity files etc Details can be found in the PEST manual which is available in the PEST subdirectory of the package If you encounter errors during the second step you can try running PEARLNEQ directly PEARLMK has created a file example neg which is t
5. Manual of PEARLNEQ VERSION COMPATIBLE WITH FOCUS PEARL 3 3 3 Aalarik Tiktak Jos Boesten 1 MNP PO BOX 1 3720 AH BILTHOVEN the Netherlands 2 Alterra PO BOX 47 6700 AA WAGENINGEN the Netherlands Tip Start with the default example provided with the package Release notes Values in table 6 of the PEARL manual should be replaced by the values in this example see appendix 2 The previous version contained a series of small bugs These bugs have been fixed in this version The weighing factors for the measurements have changed all measurements are now given a weighing factor that equals the inverse of each measured value thus giving equal weight to all measurements According to recommendations of the FOCUS Kinetics Workgroup the initial mass in the system should be optimised This recommendation has been implemented in the current version z A second example example2 mkn has been added This document describes the PEARLNEQ PEST combination which can be used to obtain the half life at reference temperature DT50Ref the desorption rate coefficient CofRatDes and the multiplication factor for the non equilibrium sorption coefficient FacSorNeqEql in the case of sorption desorption kinetics If the incubation experiment has been carried out at multiple temperatures the molar activation energy MolEntTra can be optimised simultaneously Precautionary remark This PEARLNEQ software tool sh
6. g The initial total mass of pesticide in each jar In contrast to the previous version of PEARLNEQ this parameter will be optimised There is no default value for this parameter gt FacSorNegEal factor describing the ratio KFne Kreg This parameter will be optimised but you have to specify an initial guess here The default value is 0 5 gt CofRatDes d the desorption rate coefficient This parameter will be optimised but you have to specify an initial guess here The default value is 0 01 d gt DT50Ref d the transformation half life under reference conditions applying to the equilibrium domain This parameter will be optimised but you have to specify an initial guess here As a default value you can use the classical half life which applies to the total soil system i e the equilibrium domain the non equilibrium domain gt TemRefTra C The reference temperature at which the half life should be known The default values is 20 C gt MolEntTra kJ mol the molar enthalpy of transformation This parameter will be optimised if you have carried out the experiment at multiple temperatures otherwise it is a model input f optimised you have to specify an initial guess here The default value is 54 kJ mol gt table Tem C List of temperatures at which the incubation experiment has been carried out At least two temperatures must be specified gt table Observations List of observations The fir
7. ght 34 Mean value of non zero weighted residuals 1 3161E 02 Maximum weighted residual observation o013 0 3174 Minimum weighted residual observation 030 0 2677 Standard variance of weighted residuals 1 4816E 02 Standard error of weighted residuals 0 1217 Note the above variance was obtained by dividing the objective function by the number of system degrees of freedom ie number of observations with non zero weight plus number of prior information articles with non zero weight minus the number of adjustable parameters If the degrees of freedom is negative the divisor becomes the number of observations with non zero weight plus the number of prior information items with non zero weight Parameter covariance matrix gt fsne crd dt50 masini met fsne 8 7787E 03 8 4444F 05 5 6502E 02 6 4073E 02 0 1016 crd 8 4444E 05 1 3507E 06 4 5078E 04 6 0200E 04 6 0023E 04 dt50 5 6502E 02 4 5078E 04 0 5353 0 5756 1 463 masini 6 4073E 02 6 0200E 04 0 5756 3 716 0 8663 met 0 1016 6 0023E 04 1 463 0 8663 8 463 Parameter correlation coefficient matrix gt fsne crd dt50 masini met fsne 1 000 0 7755 0 8242 0 3548 0 3727 crd 0 7755 1 000 0453 01 0 2687 0 1775 dt50 0 8242 0 5301 1 000 0 4081 0 6872 masini 0 3548 0 2687 0 4081 1 000 0 1545 met 0 3727 0 1775 0 6872 0 1545 1 000 Normalized eigenvectors of parameter covariance matrix gt Vector_l Vector_2 Vector_
8. he input file for PEARLNEQ Type Neq_Bin PEARLNEQ EXAMPLE PEARLNEQ will create an output file example out and a log file example log The output files are self explaining You can use the XyWin program to make a graph XyWin is available in the neq_bin xypearl directory of PEARLNEQ the example job does this call for your Neq_Bin xypearl xywin j example job W 1 Run PEARL_Neg with your own data 1 Be sure that an appropriate incubation experiment has been carried out see the FOCUS PEARL user manual The first step of optimising your own data consists of editing the file example mkn which can be found in the example subdirectory of the PEARLNEQ directory Please make a copy of this file before editing An example of this input file is listed at the end of this appendix The following parameters must be provided gt TimEnd d The duration of the incubation experiment gt MasSol g The mass of dry soil incubated in each jar gt VolLigSol mL Volume of liquid in the moist soil during incubation gt VolLiqgAdd mL Volume of liquid added to the soil after incubation i e the amount of liquid added to perform a conventional desorption equilibrium experiment CntOm kg kg Mass fraction of organic matter in the soil ConLigRef mg L Reference concentration in the liquid phase ExpF re Freundlich exponent KomEa L kg coefficient of equilibrium sorption on organic matter Masini u
9. ogram to fit the half life pesticide that show kinetic sorption behaviour This file is intended for use with the PEST program Doherty et al 1991 Copyright MNP RIVM Alterra 2003 2005 2006 Model control Yes ScreenOutput 0 0 TimStart d 500 0 TimEnd d 0 01 DelTim d Time step of output System characterization 45 36 MasSol g 54 64 MasIni ug 6 64 VolLiqSol mL 0 0 VolLigAdd mL 0 047 CntOm kg kg 1 Sorption parameter 1 0 ConLiqRef mg L 1 0 87 ExpFre Freundlich exponent 2 KomEql L kg 1 ee FacSorNeqgEql 0 01 CofRatDes d 1 Transformation parameters 10 00 DT50Ref d 20 0 TemRefTra C 54 0 MolEntTra kJ mol 1 Initial guess of molar activation energy Temperature at which the incubation experiments are being carried out table Tem C Led 2ra A850 end_table Measured mass and concentration of pesticide as a function of time and temperature Tim Tem Mas Con a C ug ug L table Observations 2 9 51 6300 5ei285 10 5 50 5900 5 0560 42 5 46 0200 3 6635 87 5 38 6100 2 9320 57 5 3228150 1 9280 244 5 25 8700 1 4650 358 5 20 3150 0 8820 451 5 9 4250 0 6015 2 15 51 3300 548959 6 TS 47 3950 4 4425 0 T5 45 0650 379510 42 15 23 1400 1 6470 87 T5 10 8950 0 6710 57 Ls 32 E350 0 2525 244 15 1 4400 0 0305 358 15 0 4500 0 0000 451 T3 0 1500 0 0000 end_table OBS OBS OBS OBS OBS OBS OBS OBS OBS OBS OBS OBS OBS OBS OBS OBS OBS
10. ould be seen as an introduction to fitting results experiments on long term sorption kinetics to the PEARL sorption submodel PEARLNEQ shows you how PEST can be coupled to a fortran programme that contains this PEARL sorption submodel i e PEARLNEQ EXE PEARLNEQ should not be seen as a ready to use tool PEARLNEQ provides you with example input files to the PEST optimisation package and shows you how to organise this optimisation We had to make a number of assumptions for generating these PEST input files e g upper and lower bounds of parameters weighing factors for each measurement etc etc We do not claim that these assumptions are defensible for your problem they are our best guesses at this moment While using PEARLNEQ we noticed that very regularly the results depend on the initial guesses of the parameters Therefore we advise you to perform always a number of runs with different initial guesses We advise you also to analyse the results very carefully especially the confidence intervals of the estimated parameters We do not accept any responsibility for use of PEARLNEQ Running the example The following steps must be followed 1 Installation PEARLNEQ is distributed in a self extracting zip file Run unpack_pearlneq exe and specify a path e g c pearlneq The package contains three directories i e Neq_Bin pest and Neq_Example gt The Neq_Bin directory contains the PEARLNEQ executables PEARLNEQ EXE and PEARLM
11. st column contains the time d the second column the temperature column 3 contains the total mass of pesticide in the system ug column 4 contains the concentration of pesticide VV VV V ug L in moist soil then VolLigAdd 0 or in the water phase after a desorption experiment in which case VolLigAdd is not zero and column 5 contains the characters OBS Repeat step 2 5 in the example Please replace example everywhere it occurs by the name of the copied input file Ifthe optimization is not succesfull you can try re running PEARLNEQ with different initial guesses of MasIni DT50Ref FacSorNeqEq and CofRatDes Runid mkn PearlMk Runld ins Runld tpl Runld pst If Converged Runld rec Runld neq PearlNeq Runld out Figure 1 Dataflow diagram for the PEARLNEQ PEST combination Example input file activation energy and desorption rate in the case of Start time of experiment Duration of the incubation experiment Mass of soil in the icubation jars Initial guess of the initial mass of pesticide Volume of liquid in the moist soil Volume of liquid ADDED for equilibrium experiment Organic matter content Reference liquid content Coefficient of equilibrium sorption on org matter Initial guess of ratio Kfneq Kfeql Initial guess of desorption rate coefficient Initial guess of half life at reference temperature Reference temperature STANDARD FILE for PEARLMK version 3 3 3 Pr

Details

Contents

Download Pdf Manuals

Related Search

Related Contents