## vrt.create_warped_vrt()


Create a warped VRT reprojecting a raster onto a fixed target grid.


Usage

``` python
vrt.create_warped_vrt(
    src_path,
    output_vrt,
    *,
    crs,
    resolution,
    bounds=None,
    resampling=Resampling.average,
    snap=True,
    dtype=None,
    nodata=None
)
```


With `snap`, the grid origin grows outward to a multiple of `resolution`, so warps of several sources land on one common grid. Heights are shifted between vertical datums when both CRSs are compound or 3D.


## Parameters


`src_path: str | Path | UPath`  
Source raster; may be a VRT.

`output_vrt: str | Path | UPath`  
Output VRT, written with an absolute source path so it can live anywhere.

`crs: rasterio.crs.CRS or str`  
Target CRS.

`resolution: float or tuple of float or None`  
Target pixel size, or `(x, y)` sizes, in units of `crs`. `None` takes the grid `rasterio.warp.calculate_default_transform` derives.

`bounds: rasterio.coords.BoundingBox or tuple of float or None = None`  
Target extent as `(left, bottom, right, top)`, in `crs`. Defaults to the reprojected extent of the source.

`resampling: rasterio.enums.Resampling = Resampling.average`  
Warp kernel; `average` suits downsampling to a coarser target.

`snap: bool = ``True`  
Snap the derived extent onto the target pixel grid. Ignored when `bounds` is given.

`dtype: str or None = None`  
Data type of the warped band. Defaults to the source's; promote to float so an averaging warp yields a mean, not a rounded count.

`nodata: float or None = None`  
NODATA value of the warped band. Defaults to the source's. Pass alongside `dtype` when the promoted type can't hold the source's sentinel, e.g. a 0 count becoming NaN.


## Raises


`ValueError`  
If `resolution` is not positive, `bounds` is given and does not lie on the `resolution` grid, no exact transformation is available (e.g. a datum grid is missing), only one of the source CRS and `crs` has a vertical datum, or the nodata (`nodata`, else the source's) does not fit in the warped dtype.