Autocorrelation for 2D structure targeting

Metal Detecting and Prospecting trips around Tasmania

Moderator: Philski

User avatar
Posts: 2
Joined: Tue Sep 20, 2016 1:10 pm

Autocorrelation for 2D structure targeting

Postby 4nines_ » Sun Oct 16, 2016 11:30 am

Here's an example of a very useful field tool for pre-work that a very clever structural Geo taught me .
I've used the lake Cethana area (5mile, sunday creek, union mine) as an example.
Most of us know the power of research, but this method will often highlight structures of importance beyond what is recorded in data, and coupled with the wealth of modern imaging available, offers a competitive edge for the keen scrubber.
The method is based on Autocorrelation (or Fry analysis), a powerful 3D method with the right data, but very useful for 2D (surface) analysis of 'known' mineralised areas.
It can be done with software packages, but I prefer the paper and pencil, or Xcel spreadsheet using drawing tools.
First, select the area of interest. (works best with between 10 to 100 data - in this case known hard rock gold occurrences)
Make sure they are ONLY hard rock (no alluvial) or the results will look like creeks and random noise.
This will work with most structure or vein or rock type controlled mineralisation, but have to be done separately.
this example is gold (I only used the HR gold occurrences in the MRT map even though all types are showing in the map)
here is the 5 mile area from mrt map:

Next, select a mine occurrence in roughly the centre and place an overly transparent sheet over the entire area
and mark a unique symbol on the chosen centre mine occurrence.
I used xcel in this example, so I didn't have to scan paper to post - and xcel works quite well as you can just move the image around "underneath" the symbols as placed over the map.
Once the chosen centre is marked with say a star or X, being careful to keep the overlay in the same orientation (north to top)
with the X over the centre occurrence - carefully mark every other occurrences relative position using another symbol (a dot or oval etc). Once every occurrence in the chosen area is marked, the overlay is then shifted to place the X over the next occurrence, and then every other occurence is again marked relative to this new position.
this is what this area will look like once this is done (by hand takes a bit of patience)

You might notice the result is a symmetric scatter plot.
this is where the fun starts.
next you take out your magic pen (or mouse) and carefully squinting your bad eye, you mark the obvious trends.
(they might not be obvious at first but make a copy and go through several goes until it starts to come together).
here is what I came up with for the central sunday creek area:

Now. what can I infer/interpret from this?
I can see that possibly a strong first order control is the strong NW trending structures
(possibly the deeper lodes the old timers didn't/couldn't or wouldn't chase)
Second and third order controls are likely to be the NE and ENE trending conjugate arrays with the "shorter" NW trending structures.
These AND especially perhaps junctions of the above are what a prospector can target on the surface.
I use these orientations to plan my traverse orientations if detecting etc. (i.e. walk to cross them as often as possible - not to run alongside unless on a patch)

Now then - where are these structures ?
remembering that the results only show the important trends - not exactly 'where' they are (except for the general area).
One must use their own devices to determine where these might 1. exist and 2. be accessible on the surface unless you are drilling.
Here's where MRT has kindly provided a high res LIDAR dtm (luckily in this area - but not all areas)
Using my scatterplot with trends drawn on I can underlay imagery at the same or similar scale and see what pops up.
here's the MRT (partial) lidar dtm as an example:

we all probably have our favourite imagery and mapping sources etc..
I usually try to find sneaky structures on the fringes or those potentially not evident to the bloke on the ground back in the 1800's etc. In this area I like the look of any junctions.
next time you're stuck at home dreaming of dusky maidens and fairer weather try it out on your own place of interest.
you might be pleasantly surprised by the result.
Remember this is trends only NOT actual locations (other than general area) - you have to ID the likely locations from other imagery.
try overlaying the results over Magnetics as often the invisible (at surface) trends show up here. (eg in this case the NW trend first order structures) like here - If one was deeper drilling these would be useful.

stereo pair airphoto's and google earth etc are quite useful but LIDAR and/or SRTM (shuttle radar) penetrate scrub and some soils/sands etc if one can get hold of them they are useful.
please note - I am not endorsing this area (it may well be no-go) but just using it as a visual example.
and don't forget the Geology - one can see the central area here is clearly more rigid (eg the Moina sandstone responds to strain differently.. No point taking the local interp in the wrong place - but always mindful that not all geology maps are accurate..

happy hunting

Posts: 81
Joined: Sat Dec 29, 2012 9:30 pm

Re: Autocorrelation for 2D structure targeting

Postby drystone » Sun Oct 16, 2016 8:05 pm

Hi 4nines,
that looks like a handy technique for picking up structures and possible target areas.
With all the wet weather about, it might allay "cabin fever" for a bit!
Now, how to get the MRT images onto Excel...hmmm...rusty

Who is online

Users browsing this forum: No registered users and 4 guests