# writers.gdal¶

The GDAL writer creates a raster from a point cloud using an interpolation algorithm. Output is produced using GDAL and can use any driver that supports creation of rasters. A data_type can be specified for the raster (double, float, int32, etc.). If no data type is specified, the data type with the largest range supported by the driver is used.

The technique used to create the raster is a simple interpolation where each point that falls within a given radius of a raster cell center potentially contributes to the raster’s value. If no radius is provided, it is set to the product of the resolution and the square root of two. This is consistent with the original Points2Grid application from which this algorithm has its roots. If a circle with the provided radius doesn’t encompass the entire cell, it is possible that some points will not be considered at all, including those that may be within the bounds of the raster cell.

The GDAL writer creates rasters using the data specified in the dimension option (defaults to Z). The writer creates up to six rasters based on different statistics in the output dataset. The order of the layers in the dataset is as follows:

min
Give the cell the minimum value of all points within the given radius.
max
Give the cell the maximum value of all points within the given radius.
mean
Give the cell the mean value of all points within the given radius.
idw
Cells are assigned a value based on Shepard’s inverse distance weighting algorithm, considering all points within the given radius.
count
Give the cell the number of points that lie within the given radius.
stdev
Give the cell the population standard deviation of the points that lie within the given radius.

If no points fall within the circle about a raster cell, a secondary algorithm can be used to attempt to provide a value after the standard interpolation is complete. If the window_size option is non-zero, the values of a square of rasters surrounding an empty cell is applied using inverse distance weighting of any non-empty cells. The value provided for window_size is the maximum horizontal or vertical distance that a donor cell may be in order to contribute to the subject cell (A window_size of 1 essentially creates a 3x3 array around the subject cell. A window_size of 2 creates a 5x5 array, and so on.)

Cells that have no value after interpolation are given a value specified by the nodata option.

Default Embedded Stage

This stage is enabled by default

Streamable Stage

This stage supports streaming operations

## Basic Example¶

This pipeline reads the file autzen_trim.las and creates a Geotiff dataset called outputfile.tif. Since output_type isn’t specified, it creates six raster bands (“min”, “max”, “mean”, “idx”, “count” and “stdev”) in the output dataset. The raster cells are 10x10 and the radius used to locate points whose values contribute to the cell value is 14.14.

{
"pipeline":[
"pdal/test/data/las/autzen_trim.las",
{
"resolution": 10,
"filename":"outputfile.tif"
}
]
}


## Options¶

filename
Name of output file. [Required]
resolution
Length of raster cell edges in X/Y units. [Required]
Radius about cell center bounding points to use to calculate a cell value. [Default: resolution * sqrt(2)]
gdaldriver
Name of the GDAL driver to use to write the output. [Default: “GTiff”]
gdalopts

A list of key/value options to pass directly to the GDAL driver. The format is name=value,name=value,… The option may be specified any number of times.

Note

The INTERLEAVE GDAL driver option is not supported. writers.gdal always uses BAND interleaving.

data_type
The data type to use for the output raster (double, float, int32, uint16, etc.). Many GDAL drivers only support a limited set of output data types. The default value depends on the driver.
nodata
The value to use for a raster cell if no data exists in the input data with which to compute an output cell value. [Default: depends on the data_type. -9999 for double, float, int and short, 9999 for unsigned int and unsigned short, 255 for unsigned char and -128 for char]
output_type
A comma separated list of statistics for which to produce raster layers. The supported values are “min”, “max”, “mean”, “idw”, “count”, “stdev” and “all”. The option may be specified more than once. [Default: “all”]
window_size
The maximum distance from a donor cell to a target cell when applying the fallback interpolation method. See the stage description for more information. [Default: 0]
dimension
A dimension name to use for the interpolation. [Default: “Z”]
bounds
The bounds of the data to be written. Points not in bounds are discarded. The format is ([minx, maxx],[miny,maxy]).

Note

The bounds option is required when a pipeline is run in streaming mode.