ARSreadme

Download as doc, pdf, or txt
Download as doc, pdf, or txt
You are on page 1of 14

Program ARSTAN

A Tree-Ring Standardization Program Based on Detrending and


Autoregressive Time Series Modeling, with Interactive
Graphics

Dr. Edward R. Cook


and
Paul J. Krusic

Tree-Ring Laboratory
Lamont Doherty Earth Observatory
of Columbia University
Palisades, NY

Introduction

Program ARSTAN is the personal research project of Dr. Edward R.


Cook, evolving out of his graduate work at the University of Arizona in
the 1980s when the science of dendrochronology was struggling with
extracting meaningful tree-ring information from closed canopy forest
trees. The ARSTAN methodology (AutoRegressive STANdardization)
was developed to deal with a certain class of problem that is common
to trees growing in closed-canopy forest environments: the
“endogenous disturbance” problem.

A detailed description of the ARSTAN methodology is contained in


Cook’s PhD Dissertation “A Time-Series Analysis Approach to Tree-Ring
Standardization and will not be repeated here. Rather, a copy of that
dissertation is enclosed in this folder for those who wish to learn more
about it.

What you do with the program is ultimately up to you, but be advised


that specialized software is only as valuable to the user as the user’s
understanding of what the software does. In an attempt to make the
program more user friendly ARSTAN can make a plot at almost every
point in the program where there is a significant transformation of the
raw data. Use this feature to help you see what is going on with your
data and help you interpret better the results.

Disclaimer
The authors assume no responsibility for the use and support of this
program. Of course we would want to know about any situation that
produces an incorrect computation or an impossible result. However,
what you do with the program and how you do it, is your business.

System Requirements and Installation

Program ARSTAN was developed entirely on Macintosh computers.


What this means is we know more about the behavior of the program
on Macs than on any other machine. The Windows port of the program
is new as of January 2005. We are constantly using the program and
updates will appear on our website from time to time. If you are
experiencing problems getting the program to run on your computer
and you are running one of the two operating systems mentioned
below, please let us know. Be advised the program can tax your
computers abilities resulting in what may appear as a crash. This is
not unusual when calculating bootstrapped confidence limits on a site
with 100’s of series covering the same time interval. Be patient before
reaching for the reset button or ctrl_alt_deleting. It is a good idea to
run the program on machines with at least 300mb of available RAM.
For larger data sets, 500mb or more is nice.

Macintosh OSX

1. It is advisable, but not necessary, for all Mac users to install the
“Apple Developer Tools” toolbox for OSX 10.1-10.2, or iTools for OSX
10.3 and greater. These may be downloaded for free from Apples
“developer” website.
2. Put the folder “Absoft” in your computers Applications folder.
3. Copy or move the application (executable) to any location on you
computer for use.
4. All input files must reside in the same folder as the application.

Windows XP Home/Professional

1. Copy the folder “Absoft80” (with its contents) to your c: drive.


2. Continue from step 3 above.

Note: The following steps may or may not be necessary depending


on the operating system you are using. Try the program first before
continuing.

Right click on "My Computer" and select properties. Locate the


"Environment Variables" section of the properties window (in the
"Advanced" tab in Windows 2000) and add a new variable for either
your user name or system wide. The variable name must be called
"PLPLOT_LIB" and the value must be:

C:\Absoft80\examples\PLPLOT\examples

Windows 98 users may need to add the environment variable to


your c:\autoexec.bat file by typing the following line:

set PLPLOT_LIB=C:\Absoft80\examples\PLPLOT\examples

Important note about input and output file types

Macintosh Users: The file type of all ARSTAN input and output files is
“unix”.
Windows Users: The file type is “dos”.

USERS GUIDE.

Program ARSTAN is a FORTRAN program without a lot of amenities like


drop and drag mouse movements, drop down windows, all the great
stuff we expect from software today. It’s a top down program, meaning
you set it up to do what you want first, then you let it run. There are
three principle modes for running the program, interactive, command
and batch. In “interactive” mode you enter the name of a file
containing ring-width data, you configure all the options to perform on
the data within the file, hit return for “no more options” and let it run.

During an interactive mode run ARSTAN produces a logfile, or


command file. This text file (“<inputfilename>_log”) records all the
options and files used in the last run of the program. This file may be
used as input for any subsequent runs of the program, thus repeating
the exact same instructions that were used when the log file was
created. This is called command mode. Command mode is useful to
one desiring to repeat the same chronology building options on
different data files. Simply changing the input file name in the _log file
will repeat a previous option set on a new data file

ARSTAN can also open a text file containing a list of ring-width file
names (one name/line) then perform all the selected detrending and
chronology building options on the contents of each file in the list. This
is called “batch” mode and useful for building chronology networks.
We will now demonstrate the program in interactive mode to learn the
software by using a test data file <tst2.raw> and a few screen shots to
get you going.
1. Loading file: tst2.raw. (an increment file in decade format).
2. Setup option set for the current run in interactive mode.

Here we have chosen to change the first detrending method (option


#4 in the Main Program Menu) from its default (negative exponential
curve 1) to a Hugershoff growth curve fitting procedure (option #7, in
the First Detrending Sub-Menu). At the same time we request the
program to plot each individual series curve fit.
3. To continue customizing our run, the program displays the Main
Program Option window again.

Now we have chosen to view various chronology summery plots first


by selecting option #19 from the Main Program Menu, then option #2
from the Summery Plot Sub-Menu.
4. Each time options are changed the new changes are displayed in
the Main Program Options Menu.

Here, in the final set-up screen shot, we see all the changes we have
made to the default settings. Hitting return at this point launches the
program and saves this command set to the log file “tst2.raw_log”.
Here is the first series plotted with its Hugershoff curve fit.
5. The program will now prompt us to either SAVE the plot just
created, CONTINUE with the next plot, STOP viewing more detrending
plots, or allow us to customize the plot we just made.

We have chosen to change the (z) axis maximum value from 3.5 to 4.0
by selecting option 18 from the Plot Options Menu. To see the change
we now select option zer[0] (or replot) from the same menu.
Looping within the Plot Options Menu we can change a number of
attributes of any plot the program makes saving our plots in either eps
format, using the programs Plot Options Menu “save” selection. Mac
users can save plots in jpeg format using the ARSTAN program
command Bar-PlPlot-Save. There is also an choice here to copy the
plot to the computers clipboard then paste directly into a document,
as was done when writing this MSword document.
6. After all the individual series detrending is done the program, as
we have configured this run, displays various summery plots before
completing.
7. Finally, after choosing not to display more summery plots the
program terminates normally with a salutation from a great Venetian
poet, (Venice, CA that is).

Acknowledgments

This version of program ARSTAN represents an evolution spanning


nearly three decades of development. The number of people and
organizations who contributed to this effort are many. Some have
contributed ideas, others actual code. One significant contributor was
Richard Holmes who worked briefly at Lamont in the 1980’s and
passed away in 2003. Richard’s contribution of the many subroutines
to read and write data, organize files, and perform basic curve fitting
procedures are sincerely appreciated.

We sincerely acknowledge the many contribution by Ken Peters who


provided invaluable code for optimizing the cubic splining routines and
calculating their confidence limits. Ken has spent years working out
the mechanics behind many of ARSTAN’s mathematical gymnastics in
order to better serve tree-ring analysis. Though he still claims he
doesn’t understand what dendrochronologists do with his inventions,
and some publishers probably wished they never printed certain
textbooks on time series statistics, the tree-ring community is better
off now from his efforts.

We must also acknowledge the value of institutions like IMSL, Source


Forge, GNU and the worldwide community of FORTRAN programmers.
Many of the 18,441 lines of code in ARSTAN are either variants or
direct inclusions of public access code. The FORTRAN programming
language remains the language of computational science. Without this
resource, programs like ARSTAN would not be free. In using this
software you are directly benefiting from the spirit of intellectual
sharing and open source programming, a philosophy we consider
paramount to the advancement of all science.

Finally, we thank Absoft for providing a decent FORTRAN compiler that


can accept legacy FORTRAN77 code, include C++ code, and produce
executables on both Macs and PC’s. Without this ability there would be
no Windows version of the program. We want to especially thank Mr.
Scott Campbell at Absoft for helping us with the intricacies of the
Absoft programming environment and the compiler’s Pgplot port
(called Plplot) that Absoft has so thoughtfully provided with their
product. Consequently, with Absoft’s FORTRAN compiler installed,
anyone requesting the program source code may produce his or her
own version of the program.

Good luck!

P.J. Krusic
TRL-LDEO
Palisades, NY
February 3, 2005

You might also like