关于terra包地理坐标（lon/lat）下面积计算的技术问询

Let’s break down your questions about expanse() and cellSize() in the terra package clearly:

1. Are area calculations based on a sphere or ellipsoid? #

By default, both functions use the WGS84 ellipsoid (the standard geographic reference for EPSG:4326 coordinates) for highly accurate area calculations. You can toggle this behavior with the ellipsoid parameter:

• Keep ellipsoid = TRUE (the default) to stick with ellipsoid-based measurements.
• Set ellipsoid = FALSE to switch to a simplified spherical calculation if needed.

Example code snippets:

# Calculate total raster area using ellipsoid (default)
total_expanse <- expanse(my_lonlat_raster)

# Generate a raster of cell areas using spherical geometry
cell_areas_sphere <- cellSize(my_lonlat_raster, ellipsoid = FALSE)

2. How does the algorithm compare to the fixed cell width (median width) method in the raster package? #

The raster package’s area calculation (via area()) relies on estimating cell area using the median longitude of each cell to assign a fixed width. This can introduce noticeable errors, especially at high latitudes where the distortion of geographic coordinates makes cell shapes less uniform.

terra’s cellSize() and expanse() use a far more precise approach:

• For every cell, they calculate the area of the polygon formed by the cell’s four corner coordinates.
• This polygon area is computed using geodesic measurements tailored to your chosen ellipsoid (or sphere), which accounts for the actual curved shape of the cell on Earth’s surface.
• expanse() then sums these individual, precise cell areas (either for the full raster or a subset) instead of relying on aggregated median width estimates.

This makes terra’s area calculations more accurate across all latitudes compared to the raster package’s method.

内容的提问来源于stack exchange，提问作者Jean-Luc Dupouey