如何在R中提取RasterStack的三维子数据集而非矩阵？

Got it, let's break down why your slicing is returning a plain matrix instead of a 3D RasterStack, and how to get the result you want.

Why Your Current Approach Isn't Working #

When you use square bracket indexing like r[1:100, 1:100, ] on a RasterStack, the raster package defaults to extracting the raw numeric values as an array/matrix—this is designed for quick access to the underlying data, not for preserving the RasterStack structure.

Solution 1: Use crop() (Recommended for Large TIFFs) #

The crop() function is the go-to tool for this task, especially with large files, because it efficiently reads only the portion you need without loading the entire TIFF into memory. Here's how to use it:

library(raster)

# Read your large TIFF as a RasterStack
r <- stack("tiffile.tif")

# Define the crop extent using row and column indices
# Syntax: extent(raster_object, row_start, row_end, col_start, col_end)
crop_extent <- extent(r, 1, 100, 1, 100)

# Crop the stack to get your subset as a RasterStack
r_part <- crop(r, crop_extent)

This will give you a RasterStack with the exact 100x100 subset across all layers, preserving all the original raster metadata (projection, extent, etc.).

Solution 2: Convert Extracted Matrix Back to RasterStack (If You Must Use Indexing) #

If you already have the matrix from your original indexing and need to convert it back, you can rebuild the RasterStack by leveraging the original stack's metadata. This is less efficient for large files, but here's how:

# Extract the matrix as you did before
mat_part <- r[1:100, 1:100, ]

# Initialize an empty RasterStack
r_part <- stack()

# Iterate over each layer to create cropped raster layers and add to the stack
for (layer_idx in 1:nlayers(r)) {
  # Get the original layer's metadata
  original_layer <- raster(r, layer = layer_idx)
  # Crop the layer to the desired extent
  cropped_layer <- crop(original_layer, extent(r, 1, 100, 1, 100))
  # Assign the extracted matrix values to the cropped layer
  values(cropped_layer) <- mat_part[, , layer_idx]
  # Add the layer to the stack
  r_part <- addLayer(r_part, cropped_layer)
}

Key Note for Large TIFFs #

Always prefer crop() for large files—it's memory-efficient and avoids loading the entire dataset into RAM, which can cause crashes or slowdowns with massive TIFFs.

内容的提问来源于stack exchange，提问作者Daan