PDF -- Geospatial PDF

Driver short name


Build dependencies

none for write support, Poppler/PoDoFo/PDFium for read support

GDAL supports reading Geospatial PDF documents, by extracting georeferencing information and rasterizing the data. Non-geospatial PDF documents will also be recognized by the driver.

PDF documents can be created from other GDAL raster datasets, and OGR datasources can also optionally be drawn on top of the raster layer (see OGR_* creation options in the below section).

The driver supports reading georeferencing encoded in either of the 2 current existing ways : according to the OGC encoding best practice, or according to the Adobe Supplement to ISO 32000.

Multipage documents are exposed as subdatasets, one subdataset par page of the document.

Driver capabilities

Supports CreateCopy()

This driver supports the GDALDriver::CreateCopy() operation

Supports Georeferencing

This driver supports georeferencing

Supports VirtualIO

This driver supports virtual I/O operations (/vsimem/, etc.)

Vector support

See the PDF vector documentation page


The neatline (for OGC best practice) or the bounding box (Adobe style) will be reported as a NEATLINE metadata item, so that it can be later used as a cutline for the warping algorithm.

XMP metadata can be extracted from the file, and will be stored as XML raw content in the xml:XMP metadata domain.

Additional metadata, such as found in USGS Topo PDF can be extracted from the file, and will be stored as XML raw content in the EMBEDDED_METADATA metadata domain.

Configuration options

Configuration options can be specified in command-line tools using the syntax --config <NAME>=<VALUE> or using functions such as CPLSetConfigOption() (C) or gdal.config_options (Python). The following configuration options are supported:

  • GDAL_PDF_DPI=value: Defaults to 150. To control the dimensions of the raster by specifying the DPI of the rasterization. The driver will make some effort to guess the DPI value either from a specific metadata item contained in some PDF files, or from the raster images inside the PDF (in simple cases).

  • GDAL_PDF_NEATLINE=value: The name of the neatline to select (only available for geospatial PDF, encoded according to OGC Best Practice). This defaults to "Map Layers" for USGS Topo PDF. If not found, the neatline that covers the largest area.

  • PDF_USER_PWD=value: User password for protected PDFs.

  • GDAL_PDF_RENDERING_OPTIONS=[RASTER,VECTOR,TEXT/RASTER,VECTOR/RASTER,TEXT/RASTER/VECTOR,TEXT/VECTOR/TEXT]: a combination of VECTOR, RASTER and TEXT separated by comma, to select whether vector, raster or text features should be rendered. If the option is not specified, all features are rendered (Poppler and PDFium).

  • GDAL_PDF_BANDS=[3/4]: whether the PDF should be rendered as a RGB (3) or RGBA (4) image. The default value will depend on the PDF rendering used (Poppler vs PDFium) and on the content found in the PDF file (if an image with transparency is recognized, then 4 will be used). When 3 bands is selected, a white background is used.

  • GDAL_PDF_LAYERS=value: list of layers (comma separated) to turn ON (or "ALL" to turn all layers ON). The layer names can be obtained by querying the LAYERS metadata domain. When this option is specified, layers not explicitly listed will be turned off (Poppler and PDFium).

  • GDAL_PDF_LAYERS_OFF=value: list of layers (comma separated) to turn OFF. The layer names can be obtained by querying the LAYERS metadata domain (Poppler and PDFium).

  • GDAL_PDF_LAUNDER_LAYER_NAMES=[YES/NO]: (GDAL >= 3.1) Can be set to NO to avoid the layer names reported in the LAYERS metadata domain or as OGR layers for the vector part to be "laundered".

Open Options

Open options can be specified in command-line tools using the syntax -oo <NAME>=<VALUE> or by providing the appropriate arguments to GDALOpenEx() (C) or gdal.OpenEx (Python). Above configuration options are also available as open options.


  • DPI=value: Equivalent of GDAL_PDF_DPI configuration option

  • USER_PWD=value: Equivalent of PDF_USER_PWD configuration option

  • PDF_LIB=[POPPLER/PODOFO/PDFIUM]: only available for builds with multiple backends.

  • LAYERS=value: Equivalent of GDAL_PDF_LAYERS configuration option

  • LAYERS_OFF=value: Equivalent of GDAL_PDF_LAYERS_OFF configuration option

  • BANDS=value: Equivalent of GDAL_PDF_BANDS configuration option

  • NEATLINE=value: Equivalent of as GDAL_PDF_NEATLINE configuration option

LAYERS Metadata domain

When GDAL is compiled against Poppler or PDFium, the LAYERS metadata domain can be queried to retrieve layer names that can be turned ON or OFF. This is useful to know which values to specify for the GDAL_PDF_LAYERS or GDAL_PDF_LAYERS_OFF configuration options.

For example :

$ gdalinfo ../autotest/gdrivers/data/adobe_style_geospatial.pdf -mdd LAYERS

Driver: PDF/Geospatial PDF
Files: ../autotest/gdrivers/data/adobe_style_geospatial.pdf
Metadata (LAYERS):

$ gdal_translate ../autotest/gdrivers/data/adobe_style_geospatial.pdf out.tif --config GDAL_PDF_LAYERS_OFF "New_Data_Frame"


The opening of a PDF document (to get the georeferencing) is fast, but at the first access to a raster block, the whole page will be rasterized (with Poppler), which can be a slow operation.

Note: some raster-only PDF files (such as some USGS GeoPDF files), that are regularly tiled are exposed as tiled dataset by the GDAL PDF driver, and can be rendered with any backends.

Only a few of the possible Datums available in the OGC best practice spec have been currently mapped in the driver. Unrecognized datums will be considered as being based on the WGS84 ellipsoid.

For documents that contain several neatlines in a page (insets), the georeferencing will be extracted from the inset that has the largest area (in term of screen points).

Creation Issues

PDF documents can be created from other GDAL raster datasets, that have 1 band (graylevel or with color table), 3 bands (RGB) or 4 bands (RGBA).

Georeferencing information will be written by default according to the ISO32000 specification. It is also possible to write it according to the OGC Best Practice conventions (but limited to a few datum and projection types).

Note: PDF write support does not require linking to any backend.

Creation Options

Creation options can be specified in command-line tools using the syntax -co <NAME>=<VALUE> or by providing the appropriate arguments to GDALCreate() (C) or Driver.Create (Python). The following creation options are available:

  • COMPRESS=[NONE/DEFLATE/JPEG/JPEG2000]: Defaults to DEFLATE. Set the compression to use for raster data.

  • STREAM_COMPRESS=[NONE/DEFLATE]: Defaults to DEFLATE. Set the compression to use for stream objects (vector geometries, JavaScript content).

  • DPI=value: Defaults to 72. Set the DPI to use. May be automatically adjusted to higher value so that page dimension does not exceed the 14400 maximum value (in user units) allowed by Acrobat.

  • WRITE_USERUNIT=[YES/NO]: Whether the UserUnit setting computed from the DPI (UserUnit = DPI / 72.0) should be recorded in the file. When UserUnit is recorded, the raster size in pixels recognized by GDAL on reading remains identical to the source raster. When UserUnit is not recorded, the printed size will depends on the DPI value. If this parameter is not set, but DPI is specified, then it will default to NO (so that the printed size depends on the DPI value). If this parameter is not set and DPI is not specified, then UserUnit will be recorded (so that the raster size in pixels recognized by GDAL on reading remain identical to the source raster).

  • PREDICTOR=[1/2]: Defaults to 1. Only for DEFLATE compression. Might be set to 2 to use horizontal predictor that can make files smaller (but not always!).

  • JPEG_QUALITY=1-100: Defaults to 75. Set the JPEG quality when using JPEG compression. A value of 100 is best quality (least compression), and 1 is worst quality (best compression).

  • JPEG2000_DRIVER=[JP2KAK/JP2ECW/JP2OpenJPEG/JPEG2000]: Set the JPEG2000 driver to use. If not specified, it will be searched in the previous list.

  • TILED=[YES/NO]: Defaults to NO. By default monoblock files are created. This option can be used to force creation of tiled PDF files.

  • BLOCKXSIZE=value: Defaults to 256. Sets tile width.

  • BLOCKYSIZE=value: Defaults to 256. Set tile height.

  • CLIPPING_EXTENT=xmin,ymin,xmax,ymax: Set the clipping extent for the main source dataset and for the optional extra rasters. The coordinates are expressed in the units of the SRS of the dataset. If not specified, the clipping extent is set to the extent of the main source dataset.

  • LAYER_NAME=value: Name for layer where the raster is placed. If specified, the raster will be be placed into a layer that can be toggled/un-toggled in the "Layer tree" of the PDF reader.

  • EXTRA_RASTERS=<dataset_ids>: Comma separated list of georeferenced rasters to insert into the page. Those rasters are displayed on top of the main source raster. They must be georeferenced in the same projection, and they will be clipped to CLIPPING_EXTENT if it is specified (otherwise to the extent of the main source raster).

  • EXTRA_RASTERS_LAYER_NAME=<dataset_names>: Comma separated list of name for each raster specified in EXTRA_RASTERS. If specified, each extra raster will be be placed into a layer, named with the specified value, that can be toggled/un-toggled in the "Layer tree" of the PDF reader. If not specified, all the extra rasters will be placed in the default layer.

  • EXTRA_STREAM=<content>: A PDF content stream to draw after the imagery, typically to add some text. It may refer to any of the 14 standard PDF Type 1 fonts (omitting hyphens), as /FTimesRoman, /FTimesBold, /FHelvetica, /FCourierOblique, ... , in which case the required resource dictionary will be inserted.

  • EXTRA_IMAGES=<image_file_name,x,y,scale[,link=some_url]> (possibly repeated): A list of (ungeoreferenced) images to insert into the page as extra content. This is useful to insert logos, legends, etc... x and y are in user units from the lower left corner of the page, and the anchor point is the lower left pixel of the image. scale is a magnifying ratio (use 1 if unsure). If link=some_url is specified, the image will be selectable and its selection will cause a web browser to be opened on the specified URL.

  • EXTRA_LAYER_NAME=value: Name for layer where the extra content specified with EXTRA_STREAM or EXTRA_IMAGES is placed. If specified, the extra content will be be placed into a layer that can be toggled/un-toggled in the "Layer tree" of the PDF reader.

  • MARGIN=value: Margin around image in user units.

  • LEFT_MARGIN=value: Margin left of image in user units.

  • RIGHT_MARGIN=value: Margin right of image in user units.

  • TOP_MARGIN=value: Margin above image in user units.

  • BOTTOM_MARGIN=value: Margin below image in user units.

  • GEO_ENCODING=[NONE/ISO32000/OGC_BP/BOTH]: Defaults to ISO32000. Set the Geo encoding method to use.

  • NEATLINE=<polygon_definition_in_wkt>: Set the NEATLINE to use.

  • XMP=[NONE/<xml_xmp_content>]: By default, if the source dataset has data in the 'xml:XMP' metadata domain, this data will be copied to the output PDF, unless this option is set to NONE. The XMP xml string can also be directly set to this option.

  • WRITE_INFO=[YES/NO]: Defaults to YES. By default, the AUTHOR, CREATOR, CREATION_DATE, KEYWORDS, PRODUCER, SUBJECT and TITLE information will be written into the PDF Info block from the corresponding metadata item from the source dataset, or if not set, from the corresponding creation option. If this option is set to NO, no information will be written.

  • AUTHOR=value: Author metadata to write into the PDF Info block.

  • CREATOR=value: Creator metadata to write into the PDF Info block.

  • CREATION_DATE=value: Create date metadata to write into the PDF Info block. The format of the value must be D:YYYYMMDDHHmmSSOHH'mm' (e.g. D:20121122132447+02'00' for 22 nov 2012 13:24:47 GMT+02) (see PDF Reference, version 1.7, page 160)

  • KEYWORDS=value: Keywords metadata to write into the PDF Info block.

  • PRODUCER=value: Producer metadata to write into the PDF Info block.

  • SUBJECT=value: Subject metadata to write into the PDF Info block.

  • TITLE=value: Title metadata to write into the PDF Info block.

  • OGR_DATASOURCE=value: Name of the OGR datasource to display on top of the raster layer.

  • OGR_DISPLAY_FIELD=value: Name of the field (matching the name of a field from the OGR layer definition) to use to build the label of features that appear in the "Model Tree" UI component of a well-known PDF viewer. For example, if the OGR layer has a field called "ID", this can be used as the value for that option : features in the "Model Tree" will be labelled from their value for the "ID" field. If not specified, sequential generic labels will be used ("feature1", "feature2", etc... ).

  • OGR_DISPLAY_LAYER_NAMES=value: Comma separated list of names to display for the OGR layers in the "Model Tree". This option is useful to provide custom names, instead of OGR layer name that are used when this option is not specified. When specified, the number of names should be the same as the number of OGR layers in the datasource (and in the order they appear when listed by ogrinfo for example).

  • OGR_WRITE_ATTRIBUTES=[YES/NO]: Defaults to YES. Whether to write attributes of OGR features.

  • OFF_LAYERS=value: Comma separated list of layer names that should be initially hidden. By default, all layers are visible. The layer names can come from LAYER_NAME (main raster layer name), EXTRA_RASTERS_LAYER_NAME, EXTRA_LAYER_NAME and OGR_DISPLAY_LAYER_NAMES.

  • EXCLUSIVE_LAYERS=value: Comma separated list of layer names, such that only one of those layers can be visible at a time. This is the behavior of radio-buttons in a graphical user interface. The layer names can come from LAYER_NAME (main raster layer name), EXTRA_RASTERS_LAYER_NAME, EXTRA_LAYER_NAME and OGR_DISPLAY_LAYER_NAMES.

  • JAVASCRIPT=<script>: Javascript content to run at document opening. See Acrobat(R) JavaScript Scripting Reference.

  • JAVASCRIPT_FILE=<filename>: Name of Javascript file to embed and run at document opening. See Acrobat(R) JavaScript Scripting Reference.

  • COMPOSITION_FILE=<xml_filename>: (GDAL >= 3.0) See below paragraph "Creation of PDF file from a XML composition file"

Update of existing files

Existing PDF files (created or not with GDAL) can be opened in update mode in order to set or update the following elements :

  • Geotransform and associated projection (with SetGeoTransform() and SetProjection())

  • GCPs (with SetGCPs())

  • Neatline (with SetMetadataItem("NEATLINE", polygon_definition_in_wkt))

  • Content of Info object (with SetMetadataItem(key, value) where key is one of AUTHOR, CREATOR, CREATION_DATE, KEYWORDS, PRODUCER, SUBJECT and TITLE)

  • xml:XMP metadata (with SetMetadata(md, "xml:XMP"))

For geotransform or GCPs, the Geo encoding method used by default is ISO32000. OGC_BP can be selected by setting the GDAL_PDF_GEO_ENCODING configuration option to OGC_BP.

Updated elements are written at the end of the file, following the incremental update method described in the PDF specification.

Creation of PDF file from a XML composition file (GDAL >= 3.0)

A PDF file can be generate from a XML file that describes the composition of the PDF:

  • number of pages

  • layer tree, with visibility state, exclusion groups

  • definition or 0, 1 or several georeferenced areas per page

  • page content made of rasters, vectors or labels

The GDALCreate() API must be used with width = height = bands = 0 and datatype = GDT_Unknown and COMPOSITION_FILE must be the single creation option.

The XML schema against which the composition file must validate is pdfcomposition.xsd

Example on how to use the API:

char** papszOptions = CSLSetNameValue(nullptr, "COMPOSITION_FILE", "the.xml");
GDALDataset* ds = GDALCreate("the.pdf", 0, 0, 0, GDT_Unknown, papszOptions);
// return a non-null (fake) dataset in case of success, nullptr otherwise.

A sample Python script gdal_create_pdf.py is also available. Starting with GDAL 3.2, the gdal_create utility can also be used.

Example of a composition XML file:


    <LayerTree displayOnlyOnVisiblePages="true">
        <Layer id="l1" name="Satellite imagery"/>
        <Layer id="l2" name="OSM data">
            <Layer id="l2.1" name="Roads" initiallyVisible="false"/>
            <Layer id="l2.2" name="Buildings" mutuallyExclusiveGroupId="group1">
                <Layer id="l2.2.text" name="Buildings name"/>
            <Layer id="l2.3" name="Cadastral parcels" mutuallyExclusiveGroupId="group1"/>

    <Page id="page_1">
        <Georeferencing id="georeferenced">
            <SRS dataAxisToSRSAxisMapping="2,1">EPSG:4326</SRS>
            <BoundingBox x1="1" y1="1" x2="9" y2="14"/>
            <BoundingPolygon>POLYGON((1 1,9 1,9 14,1 14,1 1))</BoundingPolygon>
            <ControlPoint x="1"  y="1"  GeoY="48"  GeoX="2"/>
            <ControlPoint x="1"  y="14" GeoY="49"  GeoX="2"/>
            <ControlPoint x="9"  y="1"  GeoY="49"  GeoX="3"/>
            <ControlPoint x="9"  y="14" GeoY="48"  GeoX="3"/>

            <IfLayerOn layerId="l1">
                <!-- image drawn, and stretched to (x1,y1)->(x2,y2), without reading its georeferencing -->
                <Raster dataset="satellite.png" x1="1" y1="1" x2="9" y2="14"/>
            <IfLayerOn layerId="l2">
                <IfLayerOn layerId="l2.1">
                    <Raster dataset="roads.jpg" x1="1" y1="1" x2="9" y2="14"/>
                    <!-- vector drawn with coordinates in PDF coordinate space -->
                    <Vector dataset="roads_pdf_units.shp" layer="roads_pdf_units" visible="false">
                        <LogicalStructure displayLayerName="Roads" fieldToDisplay="road_name"/>>
                <IfLayerOn layerId="l2.2">
                    <!-- image drawn by taking into account its georeferencing -->
                    <Raster dataset="buildings.tif" georeferencingId="georeferenced"/>
                    <IfLayerOn layerId="l2.2.text">
                        <!-- vector drawn by taking into account its georeferenced coordinates -->
                        <VectorLabel dataset="labels.shp" layer="labels" georeferencingId="georeferenced">
                <IfLayerOn layerId="l2.3">
                    <PDF dataset="parcels.pdf">
                        <Blending function="Normal" opacity="0.7"/>

    <Page id="page_2">

        <OutlineItem name="turn only layer 'Satellite imagery' on, and switch to fullscreen" italic="true" bold="true">
                <SetAllLayersStateAction visible="false"/>
                <SetLayerStateAction visible="true" layerId="l1"/>
                <JavascriptAction>app.fs.isFullScreen = true;</JavascriptAction>
        <OutlineItem name="Page 1" pageId="page_1">
            <OutlineItem name="Important feature !">
                    <GotoPageAction pageId="page_1" x1="1" y1="2" x2="3" y2="4"/>
        <OutlineItem name="Page 2" pageId="page_2"/>


Build dependencies

For read support, GDAL must be built against one of the following libraries :

  • Poppler (GPL-licensed)

  • PoDoFo (LGPL-licensed)

  • PDFium (New BSD-licensed, supported since GDAL 2.1.0)

Note: it is also possible to build against a combination of several of the above libraries. PDFium will be used in priority over Poppler, itself used in priority over PoDoFo.

Unix build

The relevant configure options are --with-poppler, --with-podofo, --with-podofo-lib and --with-podofo-extra-lib-for-test.

Starting with GDAL 2.1.0, --with-pdfium, --with-pdfium-lib, --with-pdfium-extra-lib-for-test and --enable-pdf-plugin are also available.


libpoppler itself must have been configured with -DENABLE_UNSTABLE_API_ABI_HEADERS=ON so that the xpdf C++ headers are available. Note: the poppler C++ API isn't stable, so the driver compilation may fail with too old or too recent poppler versions.


As a partial alternative, the PDF driver can be compiled against libpodofo to avoid the libpoppler dependency. This is sufficient to get the georeferencing and vector information. However, for getting the imagery, the pdftoppm utility that comes with the poppler distribution must be available in the system PATH. A temporary file will be generated in a directory determined by the following configuration options : CPL_TMPDIR, TMPDIR or TEMP (in that order). If none are defined, the current directory will be used. Successfully tested versions are libpodofo 0.8.4, 0.9.1 and 0.9.3. Important note: using PoDoFo 0.9.0 is strongly discouraged, as it could cause crashes in GDAL due to a bug in PoDoFo.


Using PDFium as a backend allows access to raster, vector, georeferencing and other metadata. The PDFium backend has also support for arbitrary overviews, for fast zoom-out.

Only GDAL builds against static builds of PDFium have been tested. Building PDFium can be challenging, and particular builds must be used to work properly with GDAL.

With GDAL >= 3.9

The scripts in the https://github.com/rouault/pdfium_build_gdal_3_9 repository must be used to build a patched version of PDFium.

With GDAL = 3.8

The scripts in the https://github.com/rouault/pdfium_build_gdal_3_8 repository must be used to build a patched version of PDFium.

With GDAL 3.7

The scripts in the https://github.com/rouault/pdfium_build_gdal_3_7 repository must be used to build a patched version of PDFium.

With GDAL 3.5 and 3.6

The scripts in the https://github.com/rouault/pdfium_build_gdal_3_5 repository must be used to build a patched version of PDFium.

With GDAL 3.4

The scripts in the https://github.com/rouault/pdfium_build_gdal_3_4 repository must be used to build a patched version of PDFium.

With GDAL 3.2 and 3.3

The scripts in the https://github.com/rouault/pdfium_build_gdal_3_2 repository must be used to build a patched version of PDFium.

With GDAL 3.1.x

The scripts in the https://github.com/rouault/pdfium_build_gdal_3_1 repository must be used to build a patched version of PDFium.


  • Create a PDF from 2 rasters (main_raster and another_raster), such that main_raster is initially displayed, and they are exclusively displayed :

    gdal_translate -of PDF main_raster.tif my.pdf -co LAYER_NAME=main_raster
                   -co EXTRA_RASTERS=another_raster.tif -co EXTRA_RASTERS_LAYER_NAME=another_raster
                   -co OFF_LAYERS=another_raster -co EXCLUSIVE_LAYERS=main_raster,another_raster
  • Create of PDF with some JavaScript :

    gdal_translate -of PDF my.tif my.pdf -co JAVASCRIPT_FILE=script.js

    where script.js is :

    button = app.alert({cMsg: 'This file was generated by GDAL. Do you want to visit its website ?', cTitle: 'Question', nIcon:2, nType:2});
    if (button == 4) app.launchURL('http://gdal.org/');

See also

PDF vector documentation page

Specifications :

Libraries :

Samples :