Testing of new effective sampling size calculations and box-whisker plots

ncss-tech / soilReports

An R package that assists with the setup and operation of a collection of soil data summary, comparison, and evaluation reports. These reports are primarily used by USDA-NRCS soil scientists in both initial and update mapping.

15 stars 5 forks source link

Testing of new effective sampling size calculations and box-whisker plots #26

Open dylanbeaudette opened 8 years ago

dylanbeaudette commented 8 years ago

New process:

load rasters into memory if possible
perform cursory grid-based sampling to determine Moran's I of each raster
set sampling intensity based on I of each raster

This is all implemented in:

sharpshootR::sampleRasterStackByMU(..., estimateEffectiveSampleSize=TRUE)
sharpshootR::Moran_I_ByRaster()
sharpshootR::ESS_by_Moran_I()

These functions have not been extensively documented or tested.

dylanbeaudette commented 7 years ago

The Moran_I_ByRaster() function likely underestimates SA. A more appropriate measure would mask the input raster using the source polygons. This will not work when pixels are large relative to map unit polygons.

dylanbeaudette commented 7 years ago

Proposed, better approach:

mask current raster by input polygons
if cell size < 30m, resample to 30m (or some such value)
compute Morans' I within polygons

dylanbeaudette commented 6 years ago

Over-sampling will less of a concern after addressing this issue.