Geostatistics Workshop

Examples & Downloads

USING SNAGIT:

Snagit is incredibly useful.  I like to have the output to clipboard option selected.  I also like to define my hotkey as Alt-z because it’s easy to remember and fast on the keyboard.  I also like to have the “comment” option available. (It puts a date, time, and a comment on the output (you can just erase the time when prompted if you don’t like it).  I like the “capture region” option, too---this allows you to capture a user-defined area. Tip:  If I think I want a graphic to be publication quality, I will blow it up on the screen as big as it will get, then snagit, and then shrink it down in powerpoint.  If you snag a small object, especially one with text, you will find a loss of resolution. If you snag a giant object, and shrink it down, you will find that the resolution loss is negligible.  Remember:  Whatever is within the box you define will be captured, including grid markers on graphics screens.  You should turn off the grids if you don’t want to see it in output image!

• sgsim.par -- make sure that the data columns are correct! -- model with the original data not the nscored data.  Sgsim performs the nscoring internally, and if you feed it the original data, it will generate output in the units of the original data
• nscore.par (included with gslib program download, below) -- input data file, variable columns, # columns, file for output, file for transform table output (not used with sgsimm; only used again if you were going to backtransform the data using the GSLIB routine backtr.par, such as if you were going to do OK with the normal-scored data and want to backtransform the results.)
• example parameter entry file:  here's a screen shot of 2 surfer variograms (short and long ranges of autocorrelation), and an associated parameter file.

• Example: .sql files (ME and RI, and output for RI)-- for extracting ba/acre values from the Eastwide Database in Oracle for a given species, as well as for the plot as a whole (so you can arrive at %ba/acre as an "importance value" if you like)
• Alternatives to Arcview for the projection step (both free and commercial)
  
• Surfer-- free demo, relatively cheap to buy, for viewing and modeling the variogram/correlogram, kriging, final map display
  
• SnagIt -- for capturing and copying screen images; is Shareware, and, if you decide to buy it after using it for quite awhile, it's just 45$

Snagit is incredibly useful.  I like to have the output to clipboard option selected.  I also like to define my hotkey as Alt-z because it’s easy to remember and fast on the keyboard.  I also like to have the “comment” option available. (It puts a date, time, and a comment on the output (you can just erase the time when prompted if you don’t like it).  I like the “capture region” option, too---this allows you to capture a user-defined area. Tip:  If I think I want a graphic to be publication quality, I will blow it up on the screen as big as it will get, then snagit, and then shrink it down in powerpoint.  If you snag a small object, especially one with text, you will find a loss of resolution. If you snag a giant object, and shrink it down, you will find that the resolution loss is negligible.  Remember:  Whatever is within the box you define will be captured, including grid markers on graphics screens.  You should turn off the grids if you don’t want to see it in output image!

• gslib (full program or selected ones (recommended) used in workshop (selected ones have example parameter file we use, as well as some templates for the "rankdupe", as well as a Word2000 template for sgsim's parameter file).  We really recommend getting the GSLIB book (it's reasonably priced).  It will have descriptions of all of the programs.  If you don't get it, you can get some somewhat useful help files from here.  Note:  to get a clean parameter file, just run the programs without any parameter file (hit enter when prompted).  Also, you need to compile postme99.f using a fortran compiler like G77 (below) to run it.  You don't necessarily have to use postme; you can use postsim.exe from the gslib download page, but it's not automated like postme.  An example parameter file for gslib can be seen here; it will give you ideas on how to enter parameters for all other gslib programs.
• g77 --a free fortran compiler for compiling postme99     Note!  You need to read the instructions for installing carefully.  If you get stuck, feel free to contact Andy.  See the step by step for this.  You use this program to turn the fortran program postme99.f into an .exe program to run on the pc.  Otherwise, you'll have to compile and run postme99 on the unix
• -ArcView scripts and extensions &-Arc aml's
• -ArcView
• -Excel -- for manipulating data files, checking for duplicate locations, adding a tiny random number to eliminate duplicate values, for comparing county summaries with
• -Imapper -- for creating image maps for the web--this one's free, but there are other ones at the arcscripts page -Other software:  Variowin, SAGE2000, Gstat Variowin is great free software for variogram modelling and exploratory data analysis.  The book is really useful, and pretty necessary for figuring out how to use it.  Once you get the idea, however, it's easy to use, and offers more features than Surfer.  It does not perform kriging, however.  Many statisticians prefer the use of Variowin. Sage2000 is interesting commercial software, and offers a lot of features that the other variogram modelling features don't offer.  Not only that, but it's by Ed Isaaks of the famous An introduction to applied geostatistics. Gstat is free and offers a lot of features, but I haven't used it a lot.  It appears to be similar to gslib in that you need to enter parameters in a somewhat awkward way, but the documentation is excellent, and it offers much more compatibility with other software.

References

Keywords – geostats, spatial stats, techniques and examples…  (download .rtf version here)
Keywords - plant ecology, spatial pattern, spatial variability…

Good web pages:

• Utilizing the subplot information available on FIA plots to determine a better (more realistic) nugget value
• Incorporating ancillary information into the simulation to improve the estimates
• Incorporating joint distribution information (i.e. the relationship between the individual species) into the simulation to improve the relationships between individual species output maps
• Incorporating TM-derived information and increasing the spatial resolution of the output and hopefully the uncertainty too.
• Better use of the 'multigrid' capability of sgsimm -- to improve maintenance of both the long and short distance structures of the original data  (in particular, how does a search radius limit or affect the multigrid parameter...)
• Use of sequential indicator conditional simulation (SICS) instead of IK for creating the 'probability of occurrence' maps with a little more spatial variability.

Previous: Step by Step | Next: Videos