## -----------------------------------------------------------------------------
library(terra)
landsat5 <- rast('data/rs/centralvalley-2011LT5.tif')
names(landsat5) <- c('blue', 'green', 'red', 'NIR', 'SWIR1', 'SWIR2')


## -----------------------------------------------------------------------------
ndvi <- (landsat5[['NIR']] - landsat5[['red']]) / (landsat5[['NIR']] + landsat5[['red']])


## ----kmeans, fig.width = 8, fig.height = 4------------------------------------
# SpatExtent to crop ndvi layer
e <- ext(-121.807, -121.725, 38.004, 38.072)

# crop landsat by the extent
ndvi <- crop(ndvi, e)
ndvi

# convert the raster to a data.frme
nr <- as.data.frame(ndvi, cell=TRUE)
str(nr)


## ----kmeansobject-------------------------------------------------------------
# It is important to set the seed generator because `kmeans` initiates the centers in random locations
set.seed(99)

# Create 10 clusters, allow 500 iterations, start with 5 random sets using "Lloyd" method. 
# Do not use the first column (cell number).
kmncluster <- kmeans(nr[,-1], centers=10, iter.max = 500, nstart = 5, algorithm="Lloyd")

# kmeans returns an object of class "kmeans"
str(kmncluster)


## ----kmeansraster-------------------------------------------------------------
# Use the ndvi object to set the cluster values to a new raster
knr <- rast(ndvi, nlyr=1)
knr[nr$cell] <- kmncluster$cluster
knr


## ----kmeansplot, fig.height=4, fig.width=9------------------------------------

# Define a color vector for 10 clusters (learn more about setting the color later)
mycolor <- c("#fef65b","#ff0000", "#daa520","#0000ff","#0000ff","#00ff00","#cbbeb5",
             "#c3ff5b", "#ff7373", "#00ff00", "#808080")

par(mfrow = c(1,2))
plot(ndvi, col = rev(terrain.colors(10)), main = "Landsat-NDVI")
plot(knr, main = 'Unsupervised classification', col = mycolor, type="classes")