|
|
| GeoCommunity Mailing List |
| |
| Mailing List Archives |
| Subject: | GISList: Spatial Statistics SUM |
| Date: |
11/09/2002 09:21:36 PM |
| From: |
DuBose Griffin |
|
|
Here is the SUM of the answers that I received concerning spatial
statistics (see below signature). Thank you for all your input. I am
going to exhaust all my resources and then probably send a clearer
description of my problem with some ways that I am going to tackle it so
that people can learn from me as well as tell me where I am going wrong.
Enjoy your weekend.
***********************************************************
Ms. DuBose Griffin
Endangered Species Biologist
GIS Analyst
Sea Turtle Desk
South Carolina Department of Natural Resources
Wildlife and Freshwater Fisheries Division
Wildlife Diversity Section
217 Fort Johnson Road
Post Office Box 12559
Charleston, South Carolina 29422-2559
U.S.A.
Email: griffind@mrd.dnr.state.sc.us
Voice: 843-953-9016
Fax: 843-953-9353
URL: http://water.dnr.state.sc.us/
***********************************************************
1. Short answer for technical problem.
I would use a structural analysis, also known as variography (element
of geostatistics) to determine patterns of occurrence. You will need to
generalize presence or non-presence of nests along the beach transect
for particular years. You can also assign value to transect grid over
entire dataset to look at temporal patterns as well as spatial. You can
perform variography on both.
Lastly, forget about spatial patterns and look at conditional
probabilities such as occurrence of nest within x miles of residences or
maybe beach slope or beach length. Really no end to what you can do
with a conditional analysis. I just did one on occurrence of recent
sinkhole given overburden thickness.
We did something similar for seagrass density along transects for St.
Johns River Water Management District (in Palatka, Florida). You just
need to develop some type of way to give value to turtle presence versus
point or grid location.
You may also want to check with Fla Marine Institute. I believe they
use GIS in innovative ways to track manatees.
Good luck,
2. Hi,
Talk with Lance Waller at Emory, he did very much the same thing in
Florida...
http://www.sph.emory.edu/bios/faculty/waller.html
3. Ripley, Spatial Statistics (Wiley & Sons, 1981) is a standard
resource for
these kinds of questions. He briefly discusses estimation based on line
transects in section 7.3. Your question is not directly answered there,
but enough of the statistical theory is developed that it could readily
be
answered.
I haven't read the following, but it looks promising:
McDonald, L.L. (1980), Line-intercept sampling for attributes other than
coverage and density, J. of Wildlife Management 44 530-533.
Steve Thompson references it in his book, Sampling (Wiley & Sons,
1992). He discusses estimating population means and variances from
line-intercept samples, but does not mention pattern estimation.
4. Hello,
One method you may want to try is spatial synchrony analysis, since you
have
time series data. It lets you look at how synchrony in time trends
falls
off over distance, and is useful in identifying broad-scale population
structure. See Koenig, W. D. 1999. Spatial autocorrelation of
ecological
phenomena. Trends in Ecology and Evolution 14: 22-26. If you wanted to
just look at spatial autocorrelation (of say, an average sighting
frequency
or something like that) you could use a program like GS+ (there's also
freeware available to do it, try
http://biol10.biol.umontreal.ca/mnorton/stats.html ) to calculate a
correlogram: this will tell you if autocorrelation exists and at what
distance it falls off.
hope that helps,
5. Dear Ms. Griffin:
Thank you for your note, it sounds like you have an interesting set of
data.
We've worked with similar data in Palm Beach County. I'll give brief
description of the project, and I'm attaching a .pdf file containing a
summary of the project that illustrates the type of work we've done.
Our primary interest was to analyze yearly sets of emergence locations
for
loggerheads and greens along Juno Beach. Locations are from GPS
readings
off the beach each morning during nesting season, but would be in a
similar
format to your transect data, I'm guessing. Basically, we have a long
line
with dots on it indicating emergence and nesting locations (we've
ignored
the location on the beach so far).
We estimate the "intensity" of dots along this line using something
called
kernel estimators. Basically, think of putting a scoop of ice cream
over
each dot on the line. If dots are close together, the ice cream piles
up so
after all dots have their scoops, we see a pile of ice cream along the
line
that is higher where there are more points close together. The top of
of
this pile defines a curve that summarizes the intesity of emergences
along
the beach (or along your transect). If there is a "dip" that is an area
with relatively few em
|
|
Sponsored by:
For information
regarding
advertising rates
Click Here!
|
