First Look at a Raster File: Dimensions, Bands, and NoData

Upload a GeoTIFF and get an instant summary of dimensions, resolution, band count, CRS, NoData, and per-band min/max/mean, in Python and R on the same map.

A vector file’s first look answers “how many features, what shape, what CRS.” A raster’s is a different set of numbers entirely: how big is the grid, how many bands does each pixel carry, what does one pixel cover on the ground, and which value means “no data here” rather than a real measurement.

Get those wrong and everything downstream is wrong too: a slope calculation that treats -9999 as a very steep cliff, a mean elevation dragged down by a border of NoData pixels, a band mistaken for degrees Celsius when it’s really reflectance times 10,000.

Just want the summary, without the rest of the reading? Use the standalone tool — same engine, less prose.

1. Upload your file

A single .tif or .tiff, no sidecars, unlike a shapefile.

2. Why start with a summary?

Eight numbers, each answering a question that’s otherwise easy to get wrong by assumption:

• Dimensions: width × height in pixels. Determines how much work any pixel-by-pixel operation actually is, and whether a “small” file is small in bytes but enormous in pixel count (or the reverse).
• Bands: one grid of numbers, or several stacked together (RGB, a multispectral stack, a time series). Some checks and most visualizations only make sense once you know how many there are.
• CRS: same coordinate reference system concept as a vector file. A missing CRS is reported as genuinely unknown, not silently relabeled as WGS84. The map still needs some projection to place the footprint, so that assumption is shown separately, labeled as a display choice rather than folded into the CRS field itself.
• Resolution: the ground distance one pixel actually covers, in the file’s own CRS units. A 30 (Landsat-scale) versus a 0.001 (sub-meter) changes what questions the data can even answer.
• Data type: uint8, int16, float32, … . The single most common source of a “why are all my values 0 or wildly wrong” surprise: an integer type storing values that were meant to be divided by a scale factor, or a float column silently truncated by being written back as an integer type.
• NoData: the sentinel value marking “no measurement here” (often -9999, 0, or a type’s max value), as fundamental to a raster as a null is to a table column, and just as easy to accidentally treat as a real number if it isn’t checked first.
• Bounds: the same bounding-box sanity check as a vector file: numbers wildly outside -180..180 / -90..90 are the CRS-mismatch tell, here too.
• Value range: the per-band min/max/mean, with NoData pixels correctly excluded. The fastest way to catch a unit mistake (a “temperature in Celsius” band whose values run from 200 to 320 is actually Kelvin) before it reaches an analysis.

3. Shared map and summary

4. Inspect: summarize your file

Run the same summary in Python or R on the current data (your file uploaded above, or the built-in example below).

Python

R

5. Same spatial question, two languages

Python and R read the same file through the same underlying GDAL, but the two engines disagree in ways a vector file’s sf/GeoPandas split never surfaces:

Python (rasterio)

src.width, src.height     # dimensions
src.count                 # band count
src.crs                   # CRS
src.res                   # pixel resolution
src.dtypes[0]              # data type, as declared in the file
src.nodata                 # the NoData sentinel value itself
src.read(i, masked=True)  # one band, NoData pixels masked out

R (terra)

ncol(r), nrow(r), nlyr(r)   # dimensions + band count
crs(r, describe = TRUE)     # CRS, resolved to an EPSG code when GDAL recognizes one
res(r)                      # pixel resolution
datatype(r)[1]               # terra's own type code, one per band -- [1] mirrors dtypes[0]
# no direct equivalent — see below
global(r, "mean", na.rm = TRUE)  # NoData pixels already excluded

Three genuine divergences, not just naming differences:

• CRS formatting: same issue as the vector article’s st_crs() vs GeoPandas footnote. When GDAL can resolve the file’s CRS to a known EPSG code, crs(r, describe = TRUE) gives exactly that ("EPSG:4326", matching Python here), but for anything it can’t match to a registry entry, the fallback is a +proj=... string, a different format entirely from str(src.crs).¹
• Data type: closer to Python than stars was. datatype(r) does reflect the file’s actual storage type, just under terra’s own naming convention ("FLT4S" for a 4-byte signed float, "INT2U" for an unsigned 16-bit integer, …) rather than GDAL’s own descriptive names ("float32", "uint16") the way rasterio’s dtypes[0] reports them. A real format difference, not a loss of information the way it was with stars, though datatype(r) itself returns one type per band (terra allows different bands to carry different types), so both sides here report only the first band’s type, dtypes[0] and datatype(r)[1] respectively.
• NoData retention: rast() reads the file’s declared NoData value and excludes those pixels automatically wherever a reduction is computed (confirmed: a -9999 sentinel pixel never pollutes global(..., "min") above), the same way rasterio’s masked=True does, but the sentinel value itself isn’t exposed anywhere afterward. rasterio keeps it directly on src.nodata; terra doesn’t have an equally direct equivalent for a file already open. That’s why the R card’s “NoData” row never shows anything but “None declared,” even on a file that clearly has one: it isn’t wrong, the exclusion already happened, there’s just no simple accessor left to report the raw number by.

¹ A GeoTIFF’s own embedded CRS often resolves to a human-readable name in GDAL’s metadata (e.g. "WGS 84"), which crs(r, describe = TRUE) still maps back to its EPSG code when one exists, while an unusual or locally-defined CRS with no registry match falls back to a PROJ string on the R side, with no equivalent fallback needed on Python’s, since str(src.crs) already prints whatever GDAL gave it either way. Worth checking which form you’re looking at before assuming a mismatch means an actual problem.

6. A raster with no declared CRS

The reusable cell pair above already builds a small synthetic hill with one NoData pixel whenever nothing has been uploaded. Click ▶ Run on both languages in section 4 to see it. This button swaps in the same raster, minus the CRS, to see the CRS row switch to “Unknown” and a separate “Display assumption” row appear below it.²

² An unreferenced raster isn’t rare in practice: a sensor’s raw output, or a file exported by a tool that never wrote a CRS tag at all. Nothing about the pixel values is wrong in that case; only where they sit on the earth is unknown until someone supplies it, which is exactly the situation the warning exists to flag before a map silently plots it in the wrong place (or, for sf/GeoPandas, refuses to reproject it at all).

7. Where to next

A summary is a starting point:

• Just needed the numbers? The standalone Raster Inspector has the same summary, no reading required.
• Working with vector data instead? Start with First Look at a Geospatial File, the same idea, for GeoJSON and Shapefile.
• Need to combine two bands into a result (a difference map, a Normalized Difference like NDVI)? See Raster Band Math, or the standalone Raster Calculator.