GDAL

Main references http://www.gdal.org/ 《GDAL源码剖析与开发指南》Chapter 9 Cheat sheet for GDAL/OGR command-line geodata tools Vector operations Get vector information

ogrinfo -so input.shp layer-name

Or, for all layers

ogrinfo -al -so input.shp

Print vector extent

ogrinfo input.shp layer-name | grep Extent

List vector drivers

ogr2ogr ―formats

Convert between vector formats

ogr2ogr -f “GeoJSON” output.json input.shp

Print count of features with attributes matching a given pattern

ogrinfo input.shp layer-name | grep “Search Pattern” | sort | uniq -c

Read from a zip file

This assumes that archive.zip is in the current directory. This example just extracts the file, but any ogr2ogr operation should work. It’s also possible to write to existing zip files.

ogr2ogr -f ‘GeoJSON’ dest.geojson /vsizip/archive.zip/zipped_dir/in.geojson

Clip vectors by bounding box

ogr2ogr -f “ESRI Shapefile” output.shp input.shp -clipsrc

Clip one vector by another

ogr2ogr -clipsrc clipping_polygon.shp output.shp input.shp

Reproject vector:

ogr2ogr output.shp -t_srs “EPSG:4326” input.shp

Add an index to a shapefile

Add an index on an attribute:

ogrinfo example.shp -sql “CREATE INDEX ON example USING fieldname”

Add a spatial index:

ogrinfo example.shp -sql “CREATE SPATIAL INDEX ON example”

Merge features in a vector file by attribute (“dissolve”)

ogr2ogr -f “ESRI Shapefile” dissolved.shp input.shp -dialect sqlite -sql “select ST_union(Geometry),common_attribute from input GROUP BY common_attribute”

Merge features (“dissolve”) using a buffer to avoid slivers

ogr2ogr -f “ESRI Shapefile” dissolved.shp input.shp -dialect sqlite \

-sql “select ST_union(ST_buffer(Geometry,0.001)),common_attribute from input GROUP BY common_attribute”

Merge vector files:

ogr2ogr merged.shp input1.shp

ogr2ogr -update -append merged.shp input2.shp -nln merged

Extract from a vector file based on query

To extract features with STATENAME ‘New York’,’New Hampshire’, etc. from states.shp

ogr2ogr -where ‘STATENAME like “New%”‘ states_subset.shp states.shp

To extract type ‘pond’ from water.shp

ogr2ogr -where “type = pond” ponds.shp water.shp

Subset & filter all shapefiles in a directory

Assumes that filename and name of layer of interest are the same…

basename -s.shp *.shp | xargs -n1 -I % ogr2ogr %-subset.shp %.shp -sql “SELECT field-one, field-two FROM ‘%’ WHERE field-one=’value-of-interest’”

Extract data from a PostGis database to a GeoJSON file

ogr2ogr -f “GeoJSON” file.geojson PG:”host=localhost dbname=database user=user password=password” \

-sql “SELECT * from table_name”

Extract data from an ESRI REST API

Services that use ESRI maps are sometimes powered by a REST server that can provide data in OGR can consume. Finding the correct end point can be tricky and may take some false starts.

ogr2ogr -f GeoJSON output.geojson “http:/example.com/arcgis/rest/services/SERVICE/LAYER/MapServer/0/query?f=json&returnGeometry=true&etc=…” OGRGeoJSON

Get the difference between two vector files

Given two files that both have an id field, this will produce a vector file with the part of file1.shp that doesn’t intersect with file2.shp :

ogr2ogr diff.shp file1.shp -dialect sqlite \ -sql "SELECT ST_Difference(a.Geometry, b.Geometry) AS Geometry, a.id \ FROM file1 a LEFT JOIN 'file2.shp'.file2 b USING (id) WHERE a.Geometry != b.Geometry"

This assumes that file2.shp and file2.shp are both in the current directory.

Spatial join:

A spatial join transfers properties from one vector layer to another based on a spatial relationship between the features. Fields from the join layer can be aggregated in the output.

Given a set of points (trees.shp) and a set of polygons (parks.shp) in the same directory, create a polygon layer with the geometries from parks.shp and summaries of some columns in trees.shp:

ogr2ogr -f 'ESRI Shapefile' output.shp parks.shp -dialect sqlite \ -sql "SELECT p.Geometry, p.id id, SUM(t.field1) field1_sum, AVG(t.field2) field2_avg FROM parks p, 'trees.shp'.trees t WHERE ST_Contains(p.Geometry, t.Geometry) GROUP BY p.id"

Note that features that from parks.shp that don’t overlap with trees.shp won’t be in the new file.

Raster operations Get raster information

gdalinfo input.tif

List raster drivers

gdal_translate ―formats

Force creation of world file (requires libgeotiff)

listgeo -tfw mappy.tif

Report PROJ.4 projection info, including bounding box (requires libgeotiff)

listgeo -proj4 mappy.tif

Convert between raster formats

gdal_translate -of “GTiff” input.grd output.tif

Convert 16-bit bands (Int16 or UInt16) to Byte type

(Useful for Landsat 8 imagery…)

gdal_translate -of “GTiff” -co “COMPRESS=LZW” -scale 0 65535 0 255 -ot Byte input_uint16.tif output_byte.tif

You can change ‘0’ and ‘65535’ to your image’s actual min/max values to preserve more color variation or to apply the scaling to other band types - find that number with:

gdalinfo -mm input.tif | grep Min/Max

Convert a directory of raster files of the same format to another raster format

basename -s.img *.img | xargs -n1 -I % gdal_translate -of “GTiff” %.img %.tif

Reproject raster:

gdalwarp -t_srs “EPSG:102003” input.tif output.tif

Be sure to add -r bilinear if reprojecting elevation data to prevent funky banding artifacts.

Georeference an unprojected image with known bounding coordinates: gdal_translate -of GTiff -a_ullr

\

-a_srs EPSG:4269 input.png output.tif

Clip raster by bounding box

gdalwarp -te input.tif clipped_output.tif

Clip raster to SHP / NoData for pixels beyond polygon boundary

gdalwarp -dstnodata -cutline input_polygon.shp input.tif clipped_output.tif

Crop raster dimensions to vector bounding box

gdalwarp -cutline cropper.shp -crop_to_cutline input.tif cropped_output.tif

Merge rasters

gdal_merge.py -o merged.tif input1.tif input2.tif

Alternatively,

gdalwarp input1.tif input2.tif merged.tif

Or, to preserve nodata values:

gdalwarp input1.tif input2.tif merged.tif -srcnodata -dstnodata

Stack grayscale bands into a georeferenced RGB

Where LC81690372014137LGN00 is a Landsat 8 ID and B4, B3 and B2 correspond to R,G,B bands respectively:

gdal_merge.py -co “PHOTOMETRIC=RGB” -separate LC81690372014137LGN00_B{4,3,2}.tif -o LC81690372014137LGN00_rgb.tif

Fix an RGB TIF whose bands don’t know they’re RGB

gdal_merge.py -co “PHOTOMETRIC=RGB” input.tif -o output_rgb.tif

Export a raster for Google Earth

gdal_translate -of KMLSUPEROVERLAY input.tif output.kmz -co FORMAT=JPEG

Raster calculation (map algebra)

Average two rasters:

gdal_calc.py -A input1.tif -B input2.tif ―outfile=output.tif ―calc=”(A+B)/2”

Add two rasters:

gdal_calc.py -A input1.tif -B input2.tif ―outfile=output.tif ―calc=”A+B”

etc.

Create a hillshade from a DEM

gdaldem hillshade -of PNG input.tif hillshade.png

Change light direction:

gdaldem hillshade -of PNG -az 135 input.tif hillshade_az135.png

Use correct vertical scaling in meters if input is projected in degrees

gdaldem hillshade -s 111120 -of PNG input_WGS1984.tif hillshade.png

Apply color ramp to a DEM

First, create a color-ramp.txt file:

(Height, Red, Green, Blue)

Then apply those colors to a DEM:

gdaldem color-relief input.tif color_ramp.txt color-relief.tif

Create slope-shading from a DEM

First, make a slope raster from DEM:

gdaldem slope input.tif slope.tif

Second, create a color-slope.txt file:

(Slope angle, Red, Green, Blue)

0 255 255 255

90 0 0 0

Finally, color the slope raster based on angles in color-slope.txt:

gdaldem color-relief slope.tif color-slope.txt slopeshade.tif

Resample (resize) raster

gdalwarp -ts -r cubic dem.tif resampled_dem.tif

Entering 0 for either width or height guesses based on current dimensions.

Alternatively,

gdal_translate -outsize 10% 10% -r cubic dem.tif resampled_dem.tif

For both of these, -r cubic specifies cubic interpolation: when resampling continuous data (like a DEM), the default nearest neighbor interpolation can result in “stair step” artifacts.

Burn vector into raster

gdal_rasterize -b 1 -i -burn -32678 -l layername input.shp input.tif

Extract polygons from raster

gdal_polygonize.py input.tif -f “GeoJSON” output.json

Create contours from DEM

gdal_contour -a elev -i 50 input_dem.tif output_contours.shp

Get values for a specific location in a raster

gdallocationinfo -xml -wgs84 input.tif

Convert GRIB band to .tif

Assumes data for entire globe in WGS84. Be sure to specify band, or you may end up with a nonsense RGB image composed of the first three.

gdal_translate input.grib -a_ullr -180 -90 180 90 -a_srs EPSG:4326 -b 1 output_band1.tif

Other Convert KML points to CSV (simple)

ogr2ogr -f CSV output.csv input.kmz -lco GEOMETRY=AS_XY

Convert KML to CSV (WKT)

First list layers in the KML file

ogrinfo -so input.kml

Convert the desired KML layer to CSV

ogr2ogr -f CSV output.csv input.kml -sql “select *,OGR_GEOM_WKT from some_kml_layer”

CSV points to SHP

Given input.csv :

lon,lat,value

-81,31,13

-80,32,14

-81,33,15

Create a shapefile, using Spatialite functions to generate the point:

ogr2ogr output.shp input.csv -dialect sqlite \

-sql “SELECT MakePoint(CAST(lon as REAL), CAST(lat as REAL), 4326) Geometry, * FROM input”

Note the 4326, which refers to a spatial reference (in this case EPSG:4326 ). Use the correct code for your data.

MODIS operations

First, download relevant .hdf tiles from the MODIS ftp site: ftp://ladsftp.nascom.nasa.gov/ ; use the MODIS sinusoidal grid for reference.

List MODIS Subdatasets in a given HDF (conf. the MODIS products table )

gdalinfo longFileName.hdf | grep SUBDATASET

Make TIFs from each file in list; replace ‘MOD12Q1:Land_Cover_Type_1’ with desired Subdataset name

mkdir output

find . ‘*.hdf’ -exec gdalwarp -of GTiff ‘HDF4_EOS:EOS_GRID:”{}”:MOD12Q1:Land_Cover_Type_1’ output/{}.tif \;

Merge all .tifs in output directory into single file

gdal_merge.py -o output/Merged_Landcover.tif output/*.tif

BASH functions

Size Functions

This size function echos the pixel dimensions of a given file in the format expected by gdalwarp.

function gdal_size() {

SIZE=$(gdalinfo $1 |\

grep ‘Size is ‘ |\

cut -d\ -f3-4 |\

sed ‘s/,//g’)

echo -n “$SIZE”

}

This can be used to easily resample one raster to the dimensions of another:

gdalwarp -ts $(gdal_size bigraster.tif) -r cubicspline smallraster.tif resampled_smallraster.tif

Extent Functions

These extent functions echo the extent of the given file in the order/format expected by gdal_translate -projwin.

(Originally from Linfiniti ).

function gdal_extent() {

if [ -z “$1” ]; then

echo “Missing arguments. Syntax:”

echo “ gdal_extent

“

return

fi

EXTENT=$(gdalinfo $1 |\

grep “Upper Left|Lower Right” |\

sed “s/Upper Left //g;s/Lower Right //g;s/).

//g” |\

tr “\n” “ “ |\

sed ‘s/

$//g’ |\ tr -d “[(,]”)

echo -n “$EXTENT”

}

function ogr_extent() {

if [ -z “$1” ]; then

echo “Missing arguments. Syntax:”

echo “ ogr_extent

“

return

fi

EXTENT=$(ogrinfo -al -so $1 |\

grep Extent |\

sed ‘s/Extent: //g’ |\

sed ‘s/(//g’ |\

sed ‘s/)//g’ |\

sed ‘s/ - /, /g’)

EXTENT= echo $EXTENT | awk -F ',' '{print $1 " " $4 " " $3 " " $2}'

echo -n “$EXTENT”

}

function ogr_layer_extent() {

if [ -z “$2” ]; then

echo “Missing arguments. Syntax:”

echo “ ogr_extent

“

return

fi

EXTENT=$(ogrinfo -so $1 $2 |\

grep Extent |\

sed ‘s/Extent: //g’ |\

sed ‘s/(//g’ |\

sed ‘s/)//g’ |\

sed ‘s/ - /, /g’)

EXTENT= echo $EXTENT | awk -F ',' '{print $1 " " $4 " " $3 " " $2}'

echo -n “$EXTENT”

}

Extents can be passed directly into a gdal_translate command like so:

gdal_translate -projwin $(ogr_extent boundingbox.shp) input.tif clipped_output.tif

or

gdal_translate -projwin $(gdal_extent target_crop.tif) input.tif clipped_output.tif

This can be a useful way to quickly crop one raster to the same extent as another. Add these to your ~/.bash_profile file for easy terminal access.