obrl-soil

library(sp)
library(sf)
library(tidyverse)
library(rosm)
library(ggspatial)

nc <- st_read(system.file("shape/nc.shp", package="sf")) 
ncsp <- as(nc, 'Spatial')

# only for sf < 0.5 I think...?

obrl-soil

library(sf)
library(tidyverse)

setwd('C:/data/prepair_test')

# a file with known problems, also at https://github.com/obrl-soil/bits-n-pieces/tree/master/r2p
tester <- read_sf('C:/data/r2p/cat_v_pretty.gpkg')

# I used the QGIS topology checker tool to find invalid geometries. Row 130 is no good:
plot(st_geometry(tester[130, ]))

obrl-soil

# Inset method based on https://gis.stackexchange.com/questions/222799/create-an-inset-map-in-r/222877#222877

# libraries
library(sf)
library(raster)
library(tidyverse)
library(rosm)
library(ggspatial)
library(ggsn)
library(rmapshaper)

obrl-soil

# method: raster stack to uniquely attributed polygons
# using GRASS 7.4 to avoid rasterToPolygons
library(sp)
library(sf)
library(raster)
library(rgrass7)
library(dsmartr) # just for the demo dataset 
# https://github.com/obrl-soil/dsmartr/blob/master/data/heronvale_covariates.rda
data("heronvale_covariates")

obrl-soil

# 2018-07-03
# An example of doing some slightly fancy stuff with GDAL on the command line 
# from R using OSGeo4W GDAL, which is on system PATH already

# Task: rescale all input covariate rasters to the last BM run to [0, 1], see 
# if it makes any difference to model outputs (spoiler: no)

library(raster)
library(tidyverse)
options(stringsAsFactors = FALSE)

obrl-soil

library(sf)
options(stringsAsFactors = FALSE)

# 94720 points - 
# src: http://qldspatial.information.qld.gov.au/catalogue/custom/search.page?q=SALI_SITE
many_points <- st_read('D:/Spatial_data/soils/soil_survey_sites.shp') %>%
  st_transform(., crs = 3577) # always dist planar, kids

# divide in 2
many_points$nrow <- as.numeric(1:nrow(many_points))

obrl-soil

# from https://twitter.com/RallidaeRule/status/1083102769169514497

library(sf)
options(stringsAsFactors = FALSE)

nc <- st_read(system.file("shape/nc.shp", package="sf"))
pts <- st_centroid(nc[1:3, ]) 
pts <- pts[ , c('NAME')]
pts <- st_transform(pts, 32617)
buff <- st_buffer(pts, 10000)

obrl-soil

# find the nearest vertex on a polygon to a point
# re: https://twitter.com/EmmaVitz/status/1095574327855001600 

library(sf)
nc <- st_read(system.file("shape/nc.shp", package="sf"))

# pick a random state:
nc1 <- nc[sample(seq(nrow(nc)), 1), ]

# always project to a plane before using geos funs!

obrl-soil

# faffing about with methods to approximate a "lag vertex polygon" (Poore, 1986)
# or "best-fit regular geometric feature" (Stoddard, 1965) for polygon shape
# metric

# References:
# Poore 1986 - MSc Thesis on soil map quantification at 
# https://ir.library.oregonstate.edu/downloads/zk51vk04q
# Stoddart, D. (1965). The shape of atolls. Marine Geology, 3(5), 369–383.
# doi:10.1016/0025-3227(65)90025-3 (its on scihub)
# also apparently Fridland 1972 "The pattern of the soil cover" but that's 

obrl-soil

# datapasta'd http://worldpopulationreview.com/us-counties/nc/
#library(datapasta)
#df_paste()
nc_pop <- data.frame(stringsAsFactors=FALSE,
          V1 = c("Mecklenburg County", "Wake County", "Guilford County",
                 "Forsyth County", "Cumberland County", "Durham County",
                 "Buncombe County", "Union County", "New Hanover County",
                 "Gaston County", "Cabarrus County", "Johnston County", "Onslow County",
                 "Pitt County", "Iredell County", "Davidson County",
                 "Alamance County", "Catawba County", "Orange County", "Randolph County",