Change the coordinate reference system of raster data with various resampling methods
Powered by GDAL WASM in your browser
Files processed in browser, never uploaded
Supports thousands of coordinate reference systems
Upload the raster file to reproject
Set target CRS and resampling options
Enter EPSG code (e.g. EPSG:4326) or WKT
Set output cell size, leave empty for auto
Raster Reprojection transforms a raster from its current coordinate reference system into any target CRS you specify - an EPSG code such as EPSG:3857, a PROJ string, or a WKT definition. Six resampling methods (nearest neighbour, bilinear, cubic, cubic spline, Lanczos, and average) control how pixel values are recalculated on the new grid, and you can set the output resolution in the same pass.
It is the essential bridge step when data must arrive in a mandated projection: converting global EPSG:4326 rasters to Web Mercator for web maps, moving survey data onto a national grid, or aligning one raster with another layer before a merge or overlay analysis.
Load the raster to reproject - the tool reads and displays the source CRS so you can confirm the starting point.
Type the target CRS (for example EPSG:32633 for UTM zone 33N), pick a resampling method - nearest for land-cover classes, bilinear or cubic for continuous imagery - and optionally set the output pixel size; leaving it blank keeps the source resolution.
Run the transform and download the reprojected GeoTIFF, with every cell recalculated locally in your browser.
ANearest neighbour never invents new values, which is why it is required for categorical data such as land-use classes. Bilinear and cubic produce smoother surfaces for continuous data like elevation or temperature, and Lanczos gives the sharpest imagery at the cost of slower processing.
ASearch epsg.io or spatialreference.org by region or grid name. Thousands of codes from the EPSG registry are supported here, because the full PROJ database ships inside the browser-side GDAL build - no lookup happens on any server.
AThe output grid must cover the same footprint in the new projection, where it is usually tilted relative to the grid axes. That tilt adds boundary pixels and typically a slightly larger file; choosing a coarser output resolution has the opposite effect.