[R-sig-Geo] Newbie questions

Howard, Tim G (DEC) tim.howard at dec.ny.gov
Fri Sep 9 21:39:05 CEST 2016

In terms of differential growth of islands it seems like the key variables are:

1. Change in island size over time
2. Distance of island center and island edge to the mainland (and change of this distance over time)

So that would require repeated images of each culture and, at each time step, newly digitizing the mainland and island boundaries. If you continue to approach this conceptually from the GIS perspective, you actually may want to manually register and digitize your petri dishes/ cultures in a GIS such as GRASS GIS or QGIS (or from the non-GIS perspective, there is also ImageJ), then bring those data into R to calculate changes in areas over time and model the relationships with distance to mainland. Which might utilize some of the standard statistics packages. 


From: tkeitt at gmail.com [mailto:tkeitt at gmail.com] On Behalf Of Tim Keitt
Sent: Friday, September 09, 2016 1:06 PM
To: quantrum75 <quantrum75 at yahoo.com>
Cc: Howard, Tim G (DEC) <tim.howard at dec.ny.gov>; r-sig-geo at r-project.org
Subject: Re: [R-sig-Geo] Newbie questions

ATTENTION: This email came from an external source. Do not open attachments or click on links from unknown senders or unexpected emails.
You might also have a look at: 


I did island-to-island distances in R or PostGIS (don't recall which) and the modeling in JAGS.



On Fri, Sep 9, 2016 at 10:39 AM, quantrum75 via R-sig-Geo <r-sig-geo at r-project.org> wrote:
Hello Tim,Thank you so much for the reply! I shall surely check out gDistance in rgeos for the distance distribution. That is definitely a question for consideration since we think the boundary influences the location of the growths.
For the issue related to differential growth of islands (as polygons), what would your advice be? Any hints, tips or suggestions? Would a weighted point process distribution analysis be appropriate? What kind of a regression model would be appropriate?
I hugely appreciate any insights.RegardsQuant

      From: "Howard, Tim G (DEC)" <tim.howard at dec.ny.gov>
 To: "r-sig-geo at r-project.org" <r-sig-geo at r-project.org>
Cc: quantrum75 <quantrum75 at yahoo.com>
 Sent: Friday, September 9, 2016 5:45 AM
 Subject: RE: Newbie questions

If we assume that the islands are growing in all directions equally, then for the 'distance from boundary edge' type of question you are most interested in island initiation, which might be a point-based question. Well, points to polygon, as your boundary edge is the edge of a polygon.

On the other hand, for questions about differential growth of islands I would think you then are dealing with questions about the islands as polygons.

Certainly using many of the GIS/spatial tools can get you quite far down this path. For example, getting a distribution of island (point) distances to the boundary (with perhaps, gDistance in rgeos) would give you something you could compare among cultures.  Of course you'd be using sp to structure the data (in spatial data frames).

Good luck!

> ------------------------------
> Message: 4
> Date: Fri, 9 Sep 2016 07:53:31 +0000 (UTC)
> From: quantrum75 <quantrum75 at yahoo.com>
> To: "r-sig-geo at r-project.org" <r-sig-geo at r-project.org>
> Subject: [R-sig-Geo] Newbie questions
> Message-ID: <571527464.1514114.1473407611621 at mail.yahoo.com>
> Content-Type: text/plain; charset="utf-8"
> Hello all,I am an absolute newbie to spatial statistics, but I am quite
> thoroughly obsessed by it currently. I am an engineer by training, so I can find
> my way around math and programming. I have begun to look at the Spatial
> CRAN view and am absolutely blown away by the enormous body of work in
> there. So thank you all for that. But I am equally confused regarding the right
> choice of software and statistical approach as well.
> With that introduction, I will try to explain (one of) the problem(s) I am trying
> to solve. Of note, we absolutely do not deal with GIS issues.
> We have a setup where we are trying to grow some bacterial cultures on a
> polygonal irregular surface. I am able to obtain 2-D projection views/images
> similar to the illustration attached to this mail. Some of the things I want to do
> are
> 1) Establish a spatial statistical analytic pipeline to answer questions like
> a) Are the patterns of bacterial island growth in different cultures statistically
> random or correlated?b) Are the distances from the boundary edge to a
> bacterial island random or have a pattern?
> Now, the issue I have is, the bacterial islands have an "area", so I cannot
> really consider them as "point" processes. Yet, these "areas" are sufficiently
> far away from each other (relative to the polygonal boundary) that I can (to
> an approximation) consider them as being "point" processes (maybe?)...
> Some of my questions are
> 1) Is it appropriate to use a "point processes" method or an "areal processes"
> method for analysis of this problem?
> 2) If point processes are ok, is it ok to weight the area and use the centroid at
> the location of the point?
> 3) Where do I start? I currently have 5 books - The Baddeley book (spatstat,
> for point processes), the Cressie book (for theory), the R-INLA book
> (Bayesian spatial statistics), Applied Spatial Analysis in R (Dr Bivand's book)
> and the Gelfand book (Hierarchical modelling)
> 4) Packages - Spatstat? sp? spdep? splancs? spatial? DCluster? AMOEBA?
> Which spatial regression package? Arghhhh.....
> Now, I am not asking anyone to provide me with a ready made solution. I
> would simply appreciate if someone could point me in the right direction on
> how to start and go far from there. I am willing to learn and put in the
> necessary elbow grease to understand it well. Experts in spatial statistics are
> unavailable at my local college, so an immediate collaboration is bit far off.
> But something we can look into in the future.
> Thank you for your kind consideration. Any and all advice is eagerly
> welcomed and I will be highly grateful for the same.RegardsQuant.
> PS - Long term, I would like to contribute in some way to this group. If
> anyone is in need of any assistants for any packages, I am willing to volunteer
> my time.
> -------------- next part --------------
> An HTML attachment was scrubbed...
> URL: <https://stat.ethz.ch/pipermail/r-sig-
> geo/attachments/20160909/d0d9b4f5/attachment-0001.html>
> -------------- next part --------------
> A non-text attachment was scrubbed...
> Name: Bacterial_colonies(1).jpg
> Type: image/jpeg
> Size: 22890 bytes
> Desc: not available
> URL: <https://stat.ethz.ch/pipermail/r-sig-
> geo/attachments/20160909/d0d9b4f5/attachment-0001.jpg>

        [[alternative HTML version deleted]]

R-sig-Geo mailing list
R-sig-Geo at r-project.org

More information about the R-sig-Geo mailing list