CALCULATING DISSIMILARITY INDICES



The ANALOG program reads data from three different kinds of files (see the ANALOG Document)
The data and analog files can be in different formats as long as the Run Description file is written correctly.



Download the kirchner marsh file from the NGDC site and save it in the "napd" directory. (CLICK HERE to review the procedure.)
Instead of downloading "Tilia 2.0 Files" click on "Complete Data (Counts)" You will be presented with a long text file. Use your browser's "save as" function to save this file to the "napd" directory as "kirchner.txt"



Unfortunately, this file is not in exactly the right format for ANALOG, so you must edit "kirchner.txt" with WORDPAD and change the first line from 
# GPD ASCII Format
to 
# NAPD ASCII Format
Save "kirchner.txt"



Using WORDPAD, edit the file "run_description" in the "napd" subdirectory to produce the following Run Description file. Save it as "run.txt" (in the "napd" subdirectory.) 
basis {
	data modern.m50: napd-ascii
	meta modern.m50: napd-ascii
	transform napdrule.txt
	}

sample {
	data kirchner.txt: napd-ascii
	transform napdrule.txt
	}

distance squared_chord

report {
	name napd.out: tab
	closest 10
	meta all
	}

verbose


The values in both the analog "modern.m50" and fossil "kirchner.txt" files are raw counts. They must be converted to proportions. Using WORDPAD create a file named "napdrule.txt" containing these 3 lines 
divide by sum
#
# (end)
Save it in the "napd" directory.



Move a copy of analog.exe from the "bin" subdirectory to the "napd" subdirectory.
Open a DOS window, move to the "napd" subdirectory, and run the ANALOG program by typing at the DOS prompt 
analog run.txt


When the text stops scrolling, open the file "napd.out" with WORDPAD.
The first 10 rows (for the first Kirchner Marsh sample) should look like this 
Sample ID	Rank	Analog ID	Distance	Lat	Lon	Country	State or Province	PubIndx
20, , 182	0	0, MOD5_0454	0.101117	44.00.00	-84.10.48	USA	Michigan	25
20, , 182	1	0, MOD5_1660	0.129732	43.04.48	-89.52.12	USA	Wisconsin	167
20, , 182	2	0, MOD5_1742	0.170051	44.09.00	-91.27.00	USA	Wisconsin	171
20, , 182	3	0, MOD5_1441	0.170270	34.10.12	-80.46.48	USA	South Carolina	133
20, , 182	4	0, MOD5_1422	0.175288	32.34.48	-87.40.12	USA	Alabama	132
20, , 182	5	0, MOD5_0106	0.178156	47.16.12	-96.04.48	USA	Minnesota	11
20, , 182	6	0, MOD5_1414	0.178422	36.01.12	-87.58.48	USA	Tennessee	132
20, , 182	7	0, MOD5_1650	0.183215	43.57.00	-79.52.12	CAN	Ontario	160
20, , 182	8	0, MOD5_0103	0.202053	47.13.12	-95.46.48	USA	Minnesota	10
20, , 182	9	0, MOD5_1478	0.213885	36.06.36	-80.10.12	USA	North Carolina	132
Sample ID: the first number (20) is depth (cm), the second (182) is age (14C yr).


The squared-chord distances for sample 20 range from 0.101117 to 0.213885. A typical "cut off" for a "good" analog value is 0.14. Note that 8 of the 10 values exceed that limit.
Moreover, the distances are actually higher! ANALOG performs calculations ONLY ON TYPES THAT OCCUR IN BOTH THE "BASIS" (ANALOG) AND "SAMPLE" (FOSSIL) SAMPLES. The pollen names in the .M50 and .M70 files are not the same. For example, the Kirchner Marsh file uses the name "Asteraceae undiff." but the modern.m50 files uses "Asteraceae subfam. Asteroideae undiff." ANALOG interprets these as different types and includes neither in the calculation of dissimilarity.


Here's the Kirchner Marsh table from a run with correctly-matching
pollen names. The dissimilarities are larger because they are based on
more pollen types! 
Sample ID	Rank	Analog ID	Distance	Lat	Lon	Country	State or Province	PubIndx
20, , 182	0	0, MOD5_0454	0.140988	44.00.00	-84.10.48	USA	Michigan	25
20, , 182	1	0, MOD5_1660	0.205439	43.04.48	-89.52.12	USA	Wisconsin	167
20, , 182	2	0, MOD5_0106	0.221960	47.16.12	-96.04.48	USA	Minnesota	11
20, , 182	3	0, MOD5_0103	0.239915	47.13.12	-95.46.48	USA	Minnesota	10
20, , 182	4	0, MOD5_0329	0.241093	44.34.12	-92.01.48	USA	Wisconsin	21
20, , 182	5	0, MOD5_1742	0.243241	44.09.00	-91.27.00	USA	Wisconsin	171
20, , 182	6	0, MOD5_0104	0.257926	47.12.00	-95.49.48	USA	Minnesota	10
20, , 182	7	0, MOD5_1432	0.272571	32.18.00	-90.52.12	USA	Mississippi	132
20, , 182	8	0, MOD5_0105	0.274055	47.12.00	-95.52.12	USA	Minnesota	11
20, , 182	9	0, MOD5_0131	0.276106	45.51.00	-93.09.00	USA	Minnesota	10
When you download a fossil file from NAPD, be sure to edit it so the names
in the analog and fossil files match, before running ANALOG.

Next