📄 unflat2.m,v
字号:
head 3.0;access;symbols;locks; strict;comment @// @;3.0date 2000.06.13.19.22.36; author gilles; state Exp;branches;next 2.0;2.0date 99.05.21.18.47.02; author mah; state Exp;branches;next 1.1;1.1date 99.02.09.21.47.55; author mah; state Exp;branches;next ;desc@unflatten data@3.0log@Release 3@text@function [dataout]=unflat2(datain,datum_time,word)%%function [dataout]=unflat2(datain,datum_time,word)%%The funcion unflat2 unflattens the times stored in a header word from a desired datum time%%Input arguments are:% datain= input data in official DSI format% datum_time= datum time at which to unflatten the data from in seconds% word= header word containing the picked times of the horizon%%DSI customized VSP processing software%Written by Marko Mah February 1999%$Id: unflat2.m,v 2.0 1999/05/21 18:47:02 mah Exp gilles $%$Log: unflat2.m,v $%Revision 2.0 1999/05/21 18:47:02 mah%Release 2%%Revision 1.1 1999/02/09 21:47:55 mah%Initial revision%%%%Copyright (C) 1998 Seismology and Electromagnetic Section/%Continental Geosciences Division/Geological Survey of Canada%%This library is free software; you can redistribute it and/or%modify it under the terms of the GNU Library General Public%License as published by the Free Software Foundation; either%version 2 of the License, or (at your option) any later version.%%This library is distributed in the hope that it will be useful,%but WITHOUT ANY WARRANTY; without even the implied warranty of%MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the GNU%Library General Public License for more details.%%You should have received a copy of the GNU Library General Public%License along with this library; if not, write to the%Free Software Foundation, Inc., 59 Temple Place - Suite 330,%Boston, MA 02111-1307, USA.%%DSI Consortium%Continental Geosciences Division%Geological Survey of Canada%615 Booth St.%Ottawa, Ontario%K1A 0E9%%email: dsi@@cg.nrcan.gc.cadisp('[dataout]=unflat2(datain,datum_time,word)');dataout=datain; %initializes the datanrec=datain.fh{12}; %number of recordsint=datain.fh{8}; %sampling interval (sec)tstart=datain.fh{9}; %start time (sec)samples=datain.fh{7}; %number of points per trace for COUNT1=1:nrec [a,ntraces]=size(datain.dat{COUNT1}); %ntraces is the number of traces in the record dataout.dat{COUNT1}(:,1:ntraces)=0; %initialize traces for COUNT2=1:ntraces %calculates shift time from datum time and from trace header word 'word' shift=datain.th{COUNT1}(word,COUNT2)-datum_time; index=round(abs(shift)/int); %converts shift from time to samples if shift==0 %no shift dataout.dat{COUNT1}(:,COUNT2)=datain.dat{COUNT1}(:,COUNT2); elseif shift<0 %shift upwards dataout.dat{COUNT1}(1:(samples-index),COUNT2)=datain.dat{COUNT1}(index+1:samples,COUNT2); else %shift downwards dataout.dat{COUNT1}(index+1:samples,COUNT2)=datain.dat{COUNT1}(1:(samples-index),COUNT2); end %if index end %for COUNT2end %for COUNT1@2.0log@Release 2@text@d15 1a15 1%$Id: unflat2.m,v 1.1 1999/02/09 21:47:55 mah Exp mah $d17 3@1.1log@Initial revision@text@d15 4a18 2%$Id: Exp $%$Log: $d21 1@
⌨️ 快捷键说明
复制代码
Ctrl + C
搜索代码
Ctrl + F
全屏模式
F11
切换主题
Ctrl + Shift + D
显示快捷键
?
增大字号
Ctrl + =
减小字号
Ctrl + -