LAS file format: variable length records

Metadata in a LAS file lives in variable length records (VLRs), which follow the header, and, from LAS 1.4, in extended variable length records (EVLRs), which follow the points and can be larger. Each record is identified by a User ID and a Record ID. The specification owns the User IDs LASF_Spec and LASF_Projection; anyone else should use a User ID of their own, and readers skip records they do not know.

VLR header

54 bytes, then Record Length After Header bytes of data. The header’s Number of Variable Length Records says how many there are.

Offset	Size	Type	Name	Meaning
0	2	uint16	Reserved	0. In LAS 1.0 this was a record signature, `0xAABB`
2	16	char[16]	User ID	Owner of the record, padded with zero bytes
18	2	uint16	Record ID	Record type, meaningful only together with the User ID
20	2	uint16	Record Length After Header	Size of the data, at most 65,535 bytes
22	32	char[32]	Description	Optional text

EVLR header

60 bytes. EVLRs start at the header’s Start of First Extended Variable Length Record and follow one another. A file that must stay readable by older software uses only VLRs.

Offset	Size	Type	Name	Meaning
0	2	uint16	Reserved	0
2	16	char[16]	User ID	As for VLRs
18	2	uint16	Record ID	As for VLRs
20	8	uint64	Record Length After Header	Size of the data
28	32	char[32]	Description	Optional text

Records the specification defines

User ID	Record ID	Record	Status
`LASF_Projection`	2111	OGC math transform WKT	Optional
`LASF_Projection`	2112	OGC coordinate system WKT	The coordinate system when the WKT bit is set
`LASF_Projection`	34735	GeoKeyDirectoryTag	The coordinate system for formats 0 to 5 without WKT; not allowed in 1.5
`LASF_Projection`	34736	GeoDoubleParamsTag	Optional, with 34735
`LASF_Projection`	34737	GeoAsciiParamsTag	Optional, with 34735
`LASF_Spec`	0	Classification lookup	Optional; deprecated in 1.5
`LASF_Spec`	3	Text area description	Optional
`LASF_Spec`	4	Extra bytes	Optional; at most one
`LASF_Spec`	7	Superseded	Marks a record as replaced
`LASF_Spec`	100 to 354	Waveform packet descriptor	Needed for formats 4, 5, 9, 10
`LASF_Spec`	65535	Waveform data packets (EVLR)	Waveform data inside the file

Coordinate system records

Which form is used depends on the point format and the header’s WKT bit (Global Encoding bit 4):

Point formats	WKT bit clear	WKT bit set
0 to 5	GeoTIFF keys	WKT
6 to 10	Not allowed	WKT

WKT records

Record 2112 holds the coordinate system as OGC Well Known Text, ended by a zero byte; record 2111 holds a math transform in the same way. In LAS 1.5 any published WKT version may be used, and the record may be a VLR or an EVLR.

GeoTIFF keys

Record 34735 is an array of uint16 values in the layout of the GeoTIFF GeoKeyDirectoryTag: a header of four values, then four values per key.

Field	Meaning
`wKeyDirectoryVersion`	1
`wKeyRevision`	1
`wMinorRevision`	0
`wNumberOfKeys`	Number of keys that follow
`wKeyID`	A GeoTIFF key, such as 3072 (ProjectedCSTypeGeoKey)
`wTIFFTagLocation`	0: the value is in `wValue_Offset`; 34736: it is a double at that index in record 34736; 34737: it is text at that index in record 34737
`wCount`	Number of characters for text values, otherwise 1
`wValue_Offset`	The value, or the index described above

Record 34736 is an array of float64 values and record 34737 an array of ASCII text, each value ended by a |, which the keys point into.

Classification lookup

256 entries of 16 bytes: a uint8 class number then a 15-character description, naming the classes a file uses. Deprecated in LAS 1.5.

Text area description

Free text describing the data, such as how it was collected or processed.

Extra bytes

Describes the bytes after the standard fields of every point record. The data is an array of 192-byte entries, one per extra attribute, in the order the attributes sit in the record.

Offset	Size	Type	Name	Meaning
0	2	uint8[2]	reserved	0
2	1	uint8	data_type	Type of the attribute; see below
3	1	uint8	options	Bits saying which fields below apply; for data type 0, the attribute’s size in bytes
4	32	char[32]	name	Attribute name
36	4	uint8[4]	unused	0
40	8	any	no_data	Value meaning “no data”
48	16	uint8[16]	deprecated1	0
64	8	any	min	Smallest value
72	16	uint8[16]	deprecated2	0
88	8	any	max	Largest value
96	16	uint8[16]	deprecated3	0
112	8	float64	scale	Multiplier
120	16	uint8[16]	deprecated4	0
136	8	float64	offset	Added after scaling
144	16	uint8[16]	deprecated5	0
160	32	char[32]	description	Description of the attribute

The 8-byte no_data, min and max fields hold the value widened to 8 bytes: floats as float64, unsigned integers as uint64 and signed ones as int64.

Data types

Value	Type	Size
0	Undocumented bytes	Given by `options`
1	uint8	1
2	int8	1
3	uint16	2
4	int16	2
5	uint32	4
6	int32	4
7	uint64	8
8	int64	8
9	float32	4
10	float64	8
11 to 30	Deprecated (two- and three-value arrays of types 1 to 10 in earlier revisions)	–
31 to 255	Reserved	–

Options bits

Bit	Name	Meaning
0	no_data_bit	`no_data` is meaningful
1	min_bit	`min` is meaningful
2	max_bit	`max` is meaningful
3	scale_bit	Multiply each value by `scale`
4	offset_bit	Then add `offset`

Superseded

Software that rewrites a file and wants an existing VLR or EVLR to be ignored changes that record’s Record ID to 7, Superseded.

Waveform packet descriptor

Record IDs 100 to 354 describe how waveforms are stored; record 100 + n − 1 is the descriptor that points with Wave Packet Descriptor Index n use. Each holds 26 bytes:

Offset	Size	Type	Name	Meaning
0	1	uint8	Bits per Sample	Bits in each waveform sample, 2 to 32
1	1	uint8	Waveform Compression Type	0, no compression; other values are reserved
2	4	uint32	Number of Samples	Samples in each waveform
6	4	uint32	Temporal Sample Spacing	Time between samples in picoseconds
10	8	float64	Digitizer Gain	Volts = offset + gain × sample value
18	8	float64	Digitizer Offset	Volts = offset + gain × sample value

The waveform data packets themselves are an EVLR with Record ID 65535 (or a .wdp file beside the LAS file), holding the raw samples that each point’s byte offset and packet size point into.

.LAS File Format – Variable length records