The problem with that is buffer size - if you look at assessing the
location of historic sites in relation to roads, there's a serious snag.
If you put any kind of a buffer - say 100 meters - on roads in the
state, you will provide coverage for the majority of the state. You will
also put a serious kink in your CPU. The end result is a model that is
not statistically valid. 
Apparently, we've spent a lot of time and money constructing roads -
even ones that no longer appear in our GIS databases; there are
abandoned roads (and railroads) all over the state. 

Daniel B. Davis
Archaeologist Coordinator
Kentucky Transportation Cabinet
Division of Environmental Analysis
200 Mero Street
Frankfort, KY 40622
(502) 564-7250

-----Original Message-----
From: HISTORICAL ARCHAEOLOGY [mailto:[log in to unmask]] On Behalf Of Al
Tonetti
Sent: Thursday, February 05, 2009 11:00 AM
To: [log in to unmask]
Subject: Re: Predictive modelling of burials

I don't have the quantitative data to show it, but for historic
cemeteries
distance to ground transportation facilities such as roads, paths,
trails,
etc., may be predictive.  I know of few historic cemeteries that
aren't/weren't "near" a road, path, trail, etc.  This is confirmed by
viewing the location of cemeteries recorded by the Ohio Genealogical
Society
contained in the Ohio Historic Preservation Office online GIS.

Al Tonetti
Cultural Resource Specialist
ASC Group, Inc.
800 Freeway Drive North, Suite 101
Columbus, OH 43229-5439
(614) 268-2514 x3547
(614) 268-7881 fax
[log in to unmask]
www.ascgroup.net

-----Original Message-----
From: HISTORICAL ARCHAEOLOGY [mailto:[log in to unmask]] On Behalf Of
Davis,
Daniel (KYTC)
Sent: Thursday, February 05, 2009 10:41 AM
To: [log in to unmask]
Subject: Re: Predictive modelling of burials

Slope? Softness of ground? pH?

I'd say you'd need to alter your variables based on the location you are
attempting to predict. I've recorded a number of small cemeteries in
eastern Kentucky and some of them are on ground so steep (and rocky)
you'd think the graves were dug into the mountain like a tunnel. Looks
like the best predictive variable (on anti-predictive, I suppose) in
that neck of the woods is prime farmland. If it's flat and fertile (or
relatively so), don't plant people - plant crops. 

That being said, the place to start is with a clear understanding of
historic land use patterns in your area of interest. In the flat lands
(the rest of the state) historic period burials usually associate
closely with residential land use patterns. The earliest burial clusters
appear to be found in areas suitable for building a house, or later, a
church. I've not completed any statistical modeling studies by any
stretch, but I've a long and abiding interest in DPS, as evidenced by
the morbid decorum (and library) of my office. 

Daniel B. Davis
Archaeologist Coordinator
Kentucky Transportation Cabinet
Division of Environmental Analysis
200 Mero Street
Frankfort, KY 40622
(502) 564-7250

-----Original Message-----
From: HISTORICAL ARCHAEOLOGY [mailto:[log in to unmask]] On Behalf Of
Richard Wright
Sent: Wednesday, February 04, 2009 6:43 PM
To: [log in to unmask]
Subject: Predictive modelling of burials

Predictive modelling of settlement sites used to be all the rage. 

The independent variables were slope of ground, direction of view,
elevation in landscape, closeness to water, closeness to edible
resources, etc..

The dependent variable was the probability of sites, or estimates of the
number of sites.

My question is whether any attempt has been made to apply predictive
modelling to burials.

I'm not talking about pie in the sky stuff, but a study where real data
were fed into the 'equation', and predictions emerged that were
subsequently tested by ground truthing.

Any advice will be welcome, but please don't tell me that I am naive to
think it can be done. The point is that I don't think it can be done. 

Of course we might use the independent variables of softness of ground
and pH of soil to put a circle around some hectares of terrain, but what
about predictive modelling within those circled areas?


Richard Wright

No virus found in this incoming message.
Checked by AVG - www.avg.com 
Version: 8.0.233 / Virus Database: 270.10.17/1934 - Release Date:
02/04/09
08:24:00