Repository for Dave White's DSM scripts and related tools
DSM Tools 1.0.zip
An ArcGIS toolbox containing common tools used during preprocessing of raster data. Included are quite a few batch processes that work by adding the input and output folders for your raster layers.
Random Forest Model
r script utilizing the caret package to run Random Forest. This script utilizes parallel processing to speed up calculations
Under active development - Still need to finish code for producing probability maps, and some of the code for utilizing RF output to ArcSIE
cLHS
A tutorial providing several methods for implementing cLHS.
http://ncss-tech.github.io/soil-pit/sandbox/dave/clhs.html
Will be updating soon to include a section on how to use the Gower's similarity index, developed by Colby Brungard to select replacement points.
CovariateReduction_v0.1.R
Several functions to reduce the number of covariates by statistical means.
Correlation reduction - Creates a correlation matrix and then removes highly correlated covariates.
Near Zero Variance - Calculates the variance of each covariate and removes the ones that are at or near zero.
Iterative Principal Component Analysis (iPCA) - Method of iterating through PCAs to reduce the number of covariates down to the ones that account for the majority of the overall population variance.
polypedon_points_v0.3.R
A method I developed to generate model training points based on the concept of the polypedon.
landsat_indicies.R
Basic R script for calculating landsat8 indicies.
Stats_Project_DW My final project for Stat for SS part 1 http://ncss-tech.github.io/soil-pit/sandbox/dave/Stats_Project_DW.html
cLHS_replacement_pts.R Utilizing the cLHS package to generate replacement cLHS points.
DSM_Project_DW My final project for the Intro to DSM class http://ncss-tech.github.io/soil-pit/sandbox/dave/DSM_Project_DW.html
Questions or Comments contact:
Dave White USDA-Natural Resources Conservation Service Las Cruces Soil Survey Office 760 Stern Drive, Suite 124 Las Cruces, NM 88005 575-522-8775x136 (Phone) [email protected](email)