Super typhoon Haiyan makes landfall
Super typhoon Haiyan
Subscribe here to receive special images

SPOT-VEGETATION

About the VGT products

1. What are the different VEGETATION products?

The standard VEGETATION products are:

  • VGT P products (physical)
  • VGT S1 products (daily Maximum Value Composite synthesis)
  • VGT S10 products (10-day MVC synthesis)

 

The 10-day syntheses are made for day 1 – 10, 11 – 20, and 21 – end of the month.

 

2. What are the VGT P products?

VGT P (P=physical) products are adapted for scientific applications requiring highly accurate physical measurements. The data are corrected for system errors (registration errors in the different channels, calibration of all detectors along the line-array for each spectral band) and are re-sampled to predefined geographic projections chosen by the user. The pixel brightness count is the ground area's apparent reflectance as observed at the top-of-atmosphere (TOA). Auxiliary data supplied with the products allow users to process the original reflectance values using their own algorithms.

Each image pixel represents a ground area of approximately 1 x 1 km2. The image products cover all or a part of a VEGETATION segment.

 

3. What are VGT S products?

VGT S (S=synthesis) products are MVC or Maximum Value Composite Syntheses, which means that reflectances are selected on the maximum TOA NDVI value. The pixel brightness count is the ground area's reflectance (corrected for atmospheric effects); pixel values over sea and ocean are set to 0. A map of computed normalised NDVI values is also supplied with the product.

Each pixel of a VGT S product represents a ground area of approximately 1 x 1 km2. The global coverage of a synthesis extends from 75°N to 56°S. VGT S products can be ordered at the Product Distribution Facility as S1 or S10 products in a Plate Carrée projection for user-defined Regions of Interests (RoIs).

 

4. What are VGT S1 products?

VGT S1 products (daily synthesis) are composed of the 'best' ground reflectance measurements of all segments received during one day for almost the entire Earth’s surface. This is done for each of the images covering the same geographical area. At higher latitudes, multiple observations during a day are available, so for these areas the best available observations are selected. The products comprise all spectral band reflectances, the NDVI, and auxiliary data on image acquisition parameters.

 

5. What are VGT S10 products?

The VGT-S10 are global, 10-day composite images composed from the 'best available' SPOT-VGT observations over a 'dekad' (i.e., from day 1 – 10, 11 – 20, and 21 – end of the month). The products provide data from all spectral bands (SWIR, NIR, RED, BLUE), the NDVI, and auxiliary data on image acquisition parameters.

 

 

6. What are the different data layers in the VEGETATION product?

An extensive description on the data layers within the SPOT-VGT products can be found in the Section “Product Data Access and Description” of the SPOT-VGT Products User Manual.

 

7. What is the Logical Volume descriptor (0001_LOG.TXT)?

The Logical Volume Descriptor file contains the following information:

  • The product (spectral band or auxiliary data reference)
  • Projection information (type, unit, geodetic parameters, resolution, etc.)
  • Geotie points (upper left, lower left, upper right, and lower right coordinates)
  • Reference to geometric correction PCI file
  • Reference to radiometric correction PCI file
  • Segment start and end date
  • Cloud and snow/ice detection algorithm version
  • Production date

 

8. What is the relation between the digital number (DN) and real VEGETATION reflectance and NDVI values (Physical Values, PV)?

 

The conversion of SPOT-VGT image pixel values (DN) to physical values (PV) is done via:

 

PV = a * DN + b

 

For reflectance values, the values for a and b are 0.0005 and 0, respectively, while for the NDVI product these values are 0.004 and -0.1. With a valid DN range of [0, 255] this gives a physical NDVI range of [-0.1, 0.92].

 

9. How must I read the 8-bit status map?

In the SPOT-VGT P, S1, and S10 files, the Status Map contains a quality state indicator per pixel, consisting of an observation indicator (clear, cloud, ice, shadow, undefined), a land/sea flag, and a radiometric quality indicator. The Table below explains the Status Map pixel quality indicators.

 

The Table below can also be found in the “Status Map” Section of the SPOT-VGT PUM.

Explanation of the pixel quality indicators in the Status Map.

Bit (LSB to MSB)

Description

Value

Key

0-2

Cloud/Ice Snow/Shadow Flag

000

001

010

011

100

Clear

Shadow

Undefined

Cloud

Ice

3

Land/Sea

0

1

Sea

Land

4

Radiometric quality SWIR flag

0

1

Bad

Good

5

Radiometric quality NIR flag

0

1

Bad

Good

6

Radiometric quality RED flag

0

1

Bad

Good

7

Radiometric quality BLUE flag

0

1

Bad

Good

Note: Because of using a bi-cubic interpolation to improve the geometrical accuracy, it is not straightforward to label a pixel as 'bad'. The pixel value is not a 'pure' instrument pixel value but a weighted average of the 16 nearest pixels (4 × 4). The radiometric pixel quality is considered as bad when ≥20% of its value is produced by 'bad' pixels at instrument level. 

Bit 3: land=1 and water=0, values are computed from the "Digital Chart of the World"
Bit 2: ice/snow=1, no ice/snow=0, values computed from reflectance thresholds.

 

10. What is the maximum pixel size for the raw data?

The maximum pixel size (in Ground Sampling Distance, GSD) for raw data across track is 1.8 km at the swath edges. Satellite attitude oscillations, the Earth's relief, and the Earth's oblateness make that this is an approximate value. However, note that pixels in VEGETATION products are projected and interpolated, resulting in a constant pixel resolution of 1 km.

 

11. How is the coordinate localisation in a VEGETATION pixel?

The coordinate localisation in VEGETATION pixels is given for the pixel centre, see the Figure below.

 

12. What is the strange 'shadow' effect at the land borders of the VEGETATION MVC synthesis?

The land/sea mask applied to the syntheses is slightly over-dimensioned for land masses in order to cope with localisation inaccuracies. As a result, a 7-8 pixel-wide border of sea pixels surrounds the land masses. The selection of cloudy pixels in this sea border results in the above mentioned 'shadow' effect.

 

13. What are the vertical stripes in the SWIR band?

The vertical stripes in the SWIR band are caused by blind or aberrant detectors:

  • The SWIR sensor consists of 6 bricks of 300 detectors in line. At every brick junction there is a blind detector, so in-between the 6 bricks there are 5 blind 'built-in' detectors.
  • The SWIR detectors are sensible to solar proton fluxes. When a proton hits a detector, it is disturbed or destroyed, depending on the proton power or the number of the previous shocks it received.

In the image processing chain, these stripes are made invisible by interpolating the radiometry of the neighbouring detectors. The interpolation can only be done when the position of the blind or aberrant detector is known. The positions of the 'built-in' blind detectors are well known. For the others, the Image Quality Centre for Vegetation images (QIV) required a calibration campaign every other week to update the list of the blind or aberrant detectors. This list was used to interpolate bad stripes. An algorithm used at the CTIV analysed the incoming VEGETATION data to perform this detection without the updated list. However, this automatic detection may miss some slightly deviant detectors, for these detectors the list is used.

The interpolated MIR pixels are flagged as 'bad' in the status map. Depending on the image projection, one to three bad stripes can be found in the MIR band.
On previously computed VEGETATION products, the unknown blind and aberrant detectors gave undetected bad data, i.e., vertical stripes.

 

14. Which method is used for cloud and ice/snow detection?

In the previous VGTdata collection (Collection 2), the cloud and snow/ice detection changed from V1 to V2 on 11 May 2001. This change resulted in a larger amount of detected clouds. However, omission errors (detecting cloudy pixels as ‘clear’) were still present in the detection. In the Collection 3 dataset this was further improved by adding a number of rules to the existing V2 cloud and snow/ice detection schemes in the processing chain.

 

Additional rules were derived from Quesney (2008) and Berthelot (2004) for cloud and snow/ice detection, respectively. These are referred to as CLOUD V3 and SNOW/ICE V3, whereas the C2 algorithms are both referred to as V2.

 

The order of detection is the following:

  • First, both the V2 methods are applied (CLOUD and SNOW/ICE). Outputs are a cloud_V2 and snow_V2 flag. After this step, the cloud shadow detection is performed.
  • Next, the SNOW/ICE V3 detection is done, which creates a temporary output snow_V3 flag, as well as a flag for band saturation.
  • Then, the V3 cloud detection is added, which creates a temporary output cloud_V3 flag. No check is done in the following cases:
    • If CLOUD V2 returned a “cloud” or “uncertain” flag, and
    • If SNOW/ICE V3 resulted in a snow_V3 flag.
  • If a temporary cloud flag exists, it is first applied. Only afterwards, a temporary snow flag is applied. Then, the cloud shadow is applied for the cloud_V3 flags.

 

The result is that more clouds are detected. Visual inspection  has shown that there is a large improvement. A full validation of the cloud and snow/ice masking is planned.

 

Information on the cloud and snow/ice detection version number can be found in the 0001_LOG.TXT file.

 

15. Where can I find appropriate HDF software?

The SPOT-VEGETATION HDF4 data format can be opened using several software programmes including ENVI, ARCGIS, HDFView, etc.

Using ENVI:
1. In version 5.2, select File
è Open As è SPOT Vegetation

2. In version 4.8, select File è Open External File è SPOT  Vegetation.

 

Using QGIS (version 2.16 Nodebø):

Layer à Add Layer à Add Raster Layer (or click the ‘Add Raster Layer’ button on the left side of the main screen). Because SPOT-VGT data are in HDF4 format, the geolocation information cannot be displayed, due to the fact that QGIS is not able to properly read the Coordinate Reference System (CRS) information embedded in the HDF4 files.  

 

Using HDFView:

File à Open à select HDF4 file. Please note that in HDFView data analysis and modification is very limited and usage of other packages for these purposes is recommended (see below).

Other possibilities:

 

16. What are the VGT2 geometric performances?

Using a GCP database, the SPOT-VGT geometrical performance significantly improved compared to the initial post-launch performance, with a decrease of the absolute location RMS error from 725 m before March 1999 to 300 m in 2000 and further improving to 190 m afterwards. Likewise, the 95% absolute location error improved from 1380 m before March 1999 to 375 m after 2000. The multi-temporal registration error decreased from 885 m RMS (March 1999) to 220 m. The Table below shows the RMS and 95% absolute geolocation errors for VGT1 and VGT2.

 

RMS and 95% absolute location errors for VGT1 and VGT2.

Absolute Location Error

Multi-Temporal registration

Instrument

RMS

MAX(95%)

RMS

MAX(95%)

VGT1

190 m

375 m

220 m

450 m

VGT2

170 m

345 m

155 m

320 m

 

More details on the geolocation validation methodology and results can be found in Sylvander et al. (2003):

 

Sylvander, S., I. Albert-Grousset, and P. Henry, (2003), Geometrical performance of the VEGETATION products, in: Geoscience and Remote Sensing Symposium, 2003. IGARSS'03. Proceedings. 2003 IEEE International (Vol. 1, 573 – 575).

 

17. What are the causes of the BAD SWIR lines in SPOT-VGT data?

The problem with the SWIR band on VGT1 was related to the used sensor technology. Due to cosmic radiation after a certain time some “pixels” on the sensor became aberrant. From that moment onwards the specific “pixel” became an idle line in SWIR band image. It was already known before launch that throughout time VGT2 would manifest the same aberrant lines as VGT1 and repararing the SWIR band was not an option.

 

18. What is the relation between the digital number and angular, atmospheric, and HRVIR data layers (MVC synthesis)?

Angles (SZA, SAA, VZA, and VAA)
Angle value = a * DN + b

 

VZA and SZA coefficient a = 0.5
VAA and SAA coefficient a = 1.5
VZA and SZA coefficient b = 0
VAA and SAA coefficient b = 0 

Atmospheric data layers
Physical quantity = a * DN + b

Water vapour coefficient a = 0.04
Water vapour coefficient b = 0
Unit = [g cm-2]

Ozone coefficient a = 0.004
Ozone coefficient b = 0
Unit = [atm.cm] (Dobson Units)

Aerosol coefficient a = 0.004
Aerosol coefficient b = 0 

HRVIR data layers
Localisation information = a * DN + b

1BHRVIR latitude coefficient a = 0.000001
1BHRVIR latitude coefficient b = 0
1BHRVIR longitude coefficient a = 0.000001
1BHRVIR longitude coefficient b = 0

 

About the VGT instrument

 

1. What are specific characteristics of VEGETATION images? The Table below gives an overview of the spectral and dynamic surface reflectance ranges for the VGT1 and VGT2 instruments.

Spectral ranges (Full Width at Half Maximum, FWHM) and centre wavelengths (in parentheses) for VGT1 and VGT2 (2nd and 3rd columns), and the dynamic surface reflectance range per band (rightmost column).

Spectral band

VGT1

[µm]

VGT2

[µm]

Surface reflectance range [-]

BLUE (B0)

0.437 - 0.480 (0.459)

0.439 - 0.476 (0.458)

0.0 – 0.5

RED (B2)

0.615 - 0.700 (0.658)

0.616 - 0.690 (0.653)

0.0 – 0.5

NIR (B3)

0.773 - 0.894 (0.834)

0.783 - 0.892 (0.838)

0.0 – 0.7

SWIR (MIR)

1.603 - 1.695 (1.649)

1.584 - 1.685 (1.635)

0.0 – 0.6

 

Consistency between VGT1 and VGT2

What are the radiometric differences between VGT1 and VGT2?

A consistency analysis for the new Collection 3 between VGT1 and VGT2 revealed that reflectance differences are generally < 1%. For NDVI, VGT2 has larger values than VGT1, but overall these differences are small. More information on the consistency analysis can be found in the SPOT-VGT Products User Manual.

 

Data scaling, policy, known issues, and user contact

 

Scaling from DN to PV and No Data values

The Table below gives an overview of the a and b parameters to calculate the SPOT-VGT band reflectance, angles, and NDVI Physical Values (using the formula PV = a * DN + b), as well as the No Data values for each data layer.

 

Data layer

a

b

No Data

B0

0.0005

0.0

-1.0

B2

0.0005

0.0

-1.0

B3

0.0005

0.0

-1.0

NDVI

0.004

-0.1

255

SM

1.0

0.0

2

SZA and VZA

0.5

0.0

255

SAA and VAA

1.5

0.0

255

 

Data policy

Are SPOT-VGT data freely available?

Yes, all SPOT-VGT data are freely available from the VITO Earth Observation Product Distribution Portal: http://www.vito-eodata.be/. Registration to this portal is required. See the instruction movie at the portal for more information on the registration and data ordering process.

 

Further, SPOT-VGT can be explored without the need for downloading data through the PROBA-V Mission Exploitation Platform: https://proba-v-mep.esa.int/.

 

Known segment issues

Although several major issues have been addressed during the reprocessing campaign for Collection 3, some minor issues were discovered due to payload, archiving, and missing data issues. The tables below summarize these issues.

 

Data missing due to payload issues

 

Sensor

Date start

Date end

Comment

Impact on

VGT2

22/02/2012

23/02/2012

No programming on-board due to power constraints

P, S1, S10

VGT2

07/03/2008

11/03/2008

Technical problems on the VGT2 payload

P, S1, S10

VGT2

06/09/2007

12/09/2007

Technical problems on the VGT2 payload

P, S1, S10

 

 

 

 

Data missing due to archiving issues

 

Sensor

Date start

Date end

Comment

Impact on

VGT2

27/08/2007

27/08/2007

Segment 178 missing, could not be retrieved

P

VGT2

30/01/2004

30/01/2004

Segment 234 missing, could not be retrieved

P

VGT2

22/01/2004

22/01/2004

Segment 231 missing, could not be retrieved

P

VGT2

22/01/2004

22/01/2004

Segment 221 missing, could not be retrieved

P

VGT1

21/04/2000

21/04/2000

Segment 245 missing, could not be retrieved

P

VGT1

28/12/1999

28/12/1999

Segment 241 missing, could not be retrieved

P

VGT1

19/06/1998

19/06/1998

Segment 050 missing, could not be retrieved

P

VGT1

16/05/1998

16/05/1998

Segment 029 missing, could not be retrieved

P

 

Missing data

 

Sensor

Date start

Date end

Comment

Impact on

VGT1

11/08/1999

11/08/1999

Segment 247 not used due to solar eclipse

S1, S10

VGT2

29/03/2006

29/03/2006

Segments 047 and 049 not used due to solar eclipse

S1, S10

VGT2

03/10/2005

03/10/2005

Segments 126 and 127 not used due to solar eclipse

S1, S10

 

User contact

SPOT-VGT users can address technical and scientific questions to the VITO Remote Sensing Helpdesk: helpdeskticket@vgt.vito.be.

 

Please allow for a maximum of 2 working days upon receiving an answer.