pkcomposite(1) program to mosaic and composite geo-referenced images


pkcomposite -i input [-i input] -o output [options] [advanced options]


pkcomposite can be used to {mosaic} and {composite} multiple (georeferenced) raster datasets. A mosaic can merge images with different geographical extents into a single larger image. Compositing resolves the overlapping pixels according to some rule (e.g, the median of all overlapping pixels). This utility is complementary to GDAL, which currently does not support a composite step. Input datasets can have different bounding boxes and spatial resolutions.

Example: Calculate the maximum NDVI composite of two multispectral input images (e.g., red is band 0 and near infrared is band 1)

pkcomposite -i input1.tif -i input2.tif -o output.tif -cr maxndvi -cb 0 -cb 1

Example: Calculate the minimum nadir composite of two input images, where the forth band (b=3) contains the view zenith angle

pkcomposite -i input1.tif -i input2.tif -o minzenith.tif -cr minband -cb 3

Example: Calculate the minimum of two input images in all bands

pkcomposite -i input1.tif -i input2.tif -o minimum.tif -cr minallbands


-i filename, --input filename
Input image file(s). If input contains multiple images, a multi-band output is created
-o filename, --output filename
Output image file
-b band, --band band
band index(es) to crop (leave empty if all bands must be retained)
-dx xres, --dx xres
Output resolution in x (in meter) (empty: keep original resolution)
-dy yres, --dy yres
Output resolution in y (in meter) (empty: keep original resolution)
-e vector, --extent vector
get boundary from extent from polygons in vector file
-cut, --crop_to_cutline
Crop the extent of the target dataset to the extent of the cutline
-eo options, --eo options
Special extent options controlling rasterization: ATTRIBUTE|CHUNKYSIZE|ALL_TOUCHED|BURN_VALUE_FROM|MERGE_ALG, e.g., -eo ATTRIBUTE=fieldname
-m mask, --mask mask
Use the first band of the specified file as a validity mask (0 is nodata)
-msknodata value, --msknodata value
Mask value not to consider for composite
-mskband value, --mskband value
Mask band to read (0 indexed)
-ulx ULX, --ulx ULX
Upper left x value bounding box
-uly ULY, --uly ULY
Upper left y value bounding box
-lrx LRX, --lrx LRX
Lower right x value bounding box
-lry LRY, --lry LRY
Lower right y value bounding box
-cr rule, --crule rule
Composite rule (overwrite, maxndvi, maxband, minband, mean, mode (only for byte images), median, sum
-cb band, --cb band
band index used for the composite rule (e.g., for ndvi, use --cband=0 --cband=1 with 0 and 1 indices for red and nir band respectively
-srcnodata value, --srcnodata value
invalid value for input image
-bndnodata band, --bndnodata band
Bands in input image to check if pixel is valid (used for srcnodata, min and max options)
-min value, --min value
flag values smaller or equal to this value as invalid.
-max value, --max value
flag values larger or equal to this value as invalid.
-dstnodata value, --dstnodata value
nodata value to put in output image if not valid or out of bounds.
-r resampling_method, --resampling-method resampling_method
Resampling method (near: nearest neighbor, bilinear: bi-linear interpolation).
-ot type, --otype type
Data type for output image ({Byte / Int16 / UInt16 / UInt32 / Int32 / Float32 / Float64 / CInt16 / CInt32 / CFloat32 / CFloat64}). Empty string: inherit type from input image
-of GDALformat, --oformat GDALformat
Output image format (see also gdal_translate(1)). Empty string: inherit from input image
Creation option for output file. Multiple options can be specified.
-a_srs EPSG:number, --a_srs EPSG:number
Override the spatial reference for the output file (leave blank to copy from input file, use epsg:3035 to use European projection and force to European grid)
-v, --verbose

Advanced options

-file, --file
write number of observations (1) or sequence nr of selected file (2) for each pixels as additional layer in composite. Default: 0
-w weight, --weight weight
Weights (type: short) for the composite, use one weight for each input file in same order as input files are provided). Use value 1 for equal weights.
-c name, --class name
classes for multi-band output image: each band represents the number of observations for one specific class. Use value 0 for no multi-band output image.
-ct filename, --ct filename
colour table in ASCII format having 5 columns: id R G B ALFA (0: transparent, 255: solid)
-align, --align
Align output bounding box to first input image
-scale value, --scale value
Scale value output=scale*input+offset
-off value, --offset value
Offset value output=scale*input+offset
-d description, --description description
Set image description


Create a composit from two input images. If images overlap, keep only last image (default rule)

pkcomposite -i input1.tif -i input2.tif -o output.tif

Create a composit from two input images. Values of 255 in band 1 (starting from 0) are masked as invalid. Typically used when second band of input image is a cloud mask

pkcomposite -i input1.tif -i input2.tif -srcnodata 255 -bndnodata 1 -dstnodata 0 -o output.tif

Create a maximum NDVI (normalized difference vegetation index) composit. Values of 255 in band 0 are masked as invalid and flagged as 0 if no other valid coverage. Typically used for (e.g., MODIS) images where red and near infrared spectral bands are stored in bands 0 and 1 respectively. In this particular case, a value of 255 in the first input band indicates a nodata value (e.g., cloud mask is coded within the data values).

pkcomposite -i input1.tif -i input2.tif -cr maxndvi -rb 0 -rb 1 -srcnodata 255 -bndnodata 0 -dstnodata 0 -o output.tif

Create a composite image using weighted mean: output=(3/4*input1+6/4*input2+3/4*input2)/3.0

pkcomposite -i input1.tif -i input2.tif -i input3.tif -o output.tif -cr mean -w 0.75 -w 1.5 -w 0.75

Create a median composit of all GTiff images found in current directory that cover (at least part of) the image coverage.tif. Values smaller or equal to 0 are set as nodata 0 (default value for -dstnodata)

pkcomposite -i large.tif $(for IMAGE in *.tif;do pkinfo -i $IMAGE --cover $(pkinfo -i coverage.tif -bb);done) -cr median -min 0 -o output.tif


Q1. First question

A1. For individual invalid value(s) in input image, use -srcnodata

Usage: use unique value for each invalid bands set in --bndnodata or use a single value that will be applied to all invalid bands


pkcomposite -i input1.tif -i input2.tif -o output.tif -srcnodata 0 -srcnodata 255 -bndnodata 0 -bndnodata 1

will consider 0 in band 0 and 255 in band 1 of input images as no value

pkcomposite -i input1.tif -i input2.tif -o output.tif -srcnodata 0 -bndnodata 0 -bndnodata 1

will consider 0 in both bands 0 and 1 of input images as no value

For range(s) of invalid values in input images: use -min (--min) and -max (--max) Usage: use unique range set for each invalid bands set in -bndnodata


pkcomposite -i input1.tif -i input2.tif -o output.tif -min 0 -max 200 -min 0 -max 2 -bndnodata 0 -bndnodata 1

will consider all negative values in band 0 and 1 of input images as invalid. Values larger or equal to 200 in band 0 will be invalid, as well as values larger or equal to 2 in band 1

Q2. If I take the mean value as composit rule for multi-band input images, will the output image contain the mean value of overlapping images in each band?

A2. Yes