Week 10: Discussion Section

Practice gridding and spatial interpolation

Author

Alessandra Vidal Meza

Published

December 4, 2024

Source Materials

The following materials are modified from Spatial Interpolation in R.

1. Background

Spatial interpolation is a method used to estimate values at unobserved locations based on observed data from nearby points. Two of the most popular spatial interpolation methods are Inverse Distance Weighting (IDW) and ordinary/universal Kriging. Today, you will perform an IDW interpolation, which assumes Tobler’s first law of geography, which states that everything is related to everything else, but near things are more related to each other.

2. Get Started

  • Create an .Rproj as your version controlled project for Week 10
  • Create a Quarto document inside your project
  • Download this data folder from Google Drive and move it inside your project
  • Load all necessary packages and read spatial objects
Warning

If you don’t have gstat installed, make sure to run: 

install.packages("gstat")
library(tidyverse)
library(sf)
library(terra)
library(stars)
library(gstat)
library(tmap)
library(kableExtra)
ca_ecoregions <- st_read(here::here("data", "CA_Level3_Ecoregions_EPA", "ca_eco_l3.shp"))

bioclim_dir <- here::here("data", "wc2.1_2.5m")
bioclim <- list.files(bioclim_dir, pattern = glob2rx("*.tif$"), full.names = TRUE)
bioclim_sort <- bioclim[
  # Sort filepaths based on numeric suffix
  order(
  # Extract numeric suffix of filenames and convert to numeric
  as.numeric(gsub(".*_(\\d+)\\.tif$", "\\1", bioclim)))]
bioclim_rast <- rast(bioclim_sort)

variables <- c("annualMeanTemp", "meanDiurnalRange", "isothermality", "tempSeasonality", "maxTempWarmMonth", "maxTempColdMonth", "tempAnnualRange", "meanTempWetQ", "meanTempDryQ", "meanTempWarmQ", "meanTempColdQ", "annualPrecip", "precipWetMonth", "precipDryMonth", "precipSeasonality", "precipWetQ", "precipDryQ", "precipWarmQ", "precipColdQ")
names(bioclim_rast) <- variables

3. Your Task

Now, to meet this week’s learning objectives, your task:

  1. Crop and mask bioclim_rast with ca_ecoregions
  2. Use terra::spatSample to generate random points in cropped and masked bioclim_rast
    • Hint: Set as.points = TRUE within spatSample and convert to sf object for Step #3
  3. Create a grid with sf::st_make_grid and convert grid to stars object with st_as_stars
  4. Use gstat::idw to perform an inverse distance weighted interpolation for a variable of interest (annualMeanTemp, meanDiurnalRange, isothermality, etc.)
  5. Use rast and mask on interpolated stars object from Step #4
  6. Find interpolated mean of the variable of interest for each ecoregion in California using zonal
    • Hint: Convert ca_ecoregions to rast object with rasterize