ASTER Surface Reflectance/Radiance VNIR/SWIR Product
Version 2.7, April 2002

This Validated version of the VNIR/SWIR surface leaving radiance and reflectance products provide an estimate of the total radiance leaving the surface including both the reflected solar and sky components for ASTER bands 1-9. This version is a Validated version intended to provide a data product which we have found to be computationally correct and whose accuracy has been estimated (see below). This product is suitable for use in publications given the user understands the accuracy of the product is dependent on the accuracy of Level-1B radiance and atmospheric inputs to the correction procedure. 

These products correct the measured at-sensor radiance for the effects of atmospheric transmission due to gaseous absorption and path radiance due to molecular scattering using atmospheric profile information from NOAA (NCEP) derived assimilation models. The results of the atmospheric correction are only as good as he atmospheric profiles which are used. (Later versions of this product will use atmospheric temperature and water vapor profile information derived from MODIS and aerosol scattering information from MISR, sister instruments to ASTER on the TERRA platform when these atmospheric profile products are validated.) 

Table of Contents
1. Where to get detailed information on this product
2. How to report problems
3. Description of this version
* Product Quality
* Problems and limitations
4. Version history


1. Where to get detailed information on this product

There are several sources of information that users will find useful:

* ASTER Surface Reflectance/Radiance Algorithm Theoretical Basis Document. Describes the algorithm used to create the product. Available at 

http://eospso.gsfc.nasa.gov/eos_homepage/for_scientists/atbd/viewInstrument.php?instrument=24

* ASTER Higher-Level Product User Guide. Describes each of the ASTER Higher Level products in detail, including definitions of each metadata attribute and the science data fields. Available at the ASTER website (see below).


* US ASTER Website. Describes the ASTER instrument, how to obtain data, the various data products that are available, the activities of the US ASTER Science Team, etc.http://asterweb.jpl.nasa.gov/

* Level 1 product information. Because the Surface Reflectance/Radiance product is derived from a Level 1 product, understanding the L1 products is useful. 

Level 1 ATBD available at http://eospso.gsfc.nasa.gov/eos_homepage/for_scientists/atbd/viewInstrument.php?instrument=24

Level 1 User Guide available at http://http://www.science.aster.ersdac.or.jp/en/index.html


2. How to report problems

To report a problem with the product, please send mail to edc@eos.nasa.gov or call EDC DAAC User Services at (605) 594-6116.


3. Description of this version

ASTER VNIR/SWIR Surface reflectance/leaving radiance V2.7

This is the third release of these two products, now considered validated

The units after application of a scaling factor for the surface leaving radiance are Watt/m^2/steradian/micrometer and the reflectance data product is unitless. 

In converting image Data Numbers (DN) to radiance or irradiance there is both an offset and a scale factor. Both have the same value and the surface leaving radiance is found by multiplying the image DN by the appropriate band scale factor and adding the offset (DN*band scale factor + offset). The band scale factors and offsets are listed in the metadata of the product in an array called BANDSCALEFACTORS.

The current (March 2002) [band scale factors, offsets] are;

Band 1: [0.676, -0.676]
Band 2: [0.708, -0.708]
Band 3: [0.862, -0.862]
Band 4: [0.2174, -0.2174]
Band 5: [0.0696, -0.0696]
Band 6: [0.0625, -0.0625]
Band 7: [0.0597, -0.0597]
Band 8: [0.0417, -0.0417]
Band 9: [0.0318, -0.0318]

These are the same band scale factors that are used with the Level 1B VNIR/SWIR image data

Note: If the L1 processing is updated so that the scale factors change, the AST07 and AST09 Version 2.7 product will use the new factors.

Changes
Version 2.5 was released in March 2001 as a Beta Release
Further validation upgraded Version 2.5 to a Provisional Release

Version 2.7 Validated Release

ASTER VNIR/SWIR Surface reflectance/leaving radiance V2.7

This is the third release of these two products, now considered validated

The units after application of a scaling factor for the surface leaving radiance are Watt/m^2/steradian/micrometer and the reflectance data product is unitless. 

In converting image Data Numbers (DN) to radiance or irradiance there is both an offset and a scale factor. Both have the same value and the surface leaving radiance is found by multiplying the image DN by the appropriate band scale factor and adding the offset (DN*band scale factor + offset). The band scale factors and offsets are listed in the metadata of the product in an array called BANDSCALEFACTORS.

The current (March 2002) [band scale factors, offsets] are;

Band 1: [0.676, -0.676]
Band 2: [0.708, -0.708]
Band 3: [0.862, -0.862]
Band 4: [0.2174, -0.2174]
Band 5: [0.0696, -0.0696]
Band 6: [0.0625, -0.0625]
Band 7: [0.0597, -0.0597]
Band 8: [0.0417, -0.0417]
Band 9: [0.0318, -0.0318]

These are the same band scale factors that are used with the Level 1B VNIR/SWIR image data

Note: If the L1 processing is updated so that the scale factors change, the AST07 and AST09 Version 2.7 product will use the new factors.

Changes
Version 2.5 was released in March 2001 as a Beta Release
Further validation upgraded Version 2.5 to a Provisional Release

Version 2.7 Validated Release

Product Quality
The current release of the product is limited to some extent by the lack of aerosol information (as described in the next section). Thus, the quality of this data product will be strongly dependent upon the level of aerosol amount and wavelength. In addition, the quality of the product relies heavily on the accuracy of the Level 1B input radiances. The product still corrects for solar illumination effects, gaseous absorption, and molecular scattering.

For regions of small terrain relief, the expected uncertainties will be less than 0.01 reflectance units for surface reflectance less than 0.2. At higher reflectance, the uncertainty is expected to be 5% of the reflectance. For the current version of the software, these uncertainties will be true for cases where the aerosol scattering optical depth is less than 0.05. For larger aerosol loadings, the uncertainty will increase by approximately 0.005 in reflectance for each additional 0.05 aerosol optical depth increment at low reflectance (<0.20). At higher reflectance, each additional 0.05 optical depth increment will cause approximately an additional 5% uncertainty. These uncertainties have been derived for aerosol types with low absorption, thus, regions dominated by anthropogenic aerosols will have larger uncertainties. Because scattering optical depths decrease with wavelength, results will typically be more accurate at longer wavelengths where scattering effects are smaller.

Terrain relief also degrades the accuracy of this product. This effect is due to the following reasons. The first is that the algorithm itself breaks down due to an assumption of homogeneity in regions of large changes in terrain relief. In addition, there is additional uncertainty due to the resolution of the digital elevation model that is available. In most cases, it is expected that digital elevation models with spatial resolution better than 100 m will be available. In these cases, there will be no added uncertainty due to the elevation model. In other cases, the input model will revert to a 1-km model. There are two effects in this case. The first is that the poorer accuracy of the input elevation at the 15 and 30 m spatial scale will increase the uncertainty of the results. This effect will again be more important in regions of strong relief. The second effect is that the 1-km resolution will create a noticeable artifact in the corrected data when the data moves from a pixel within one 1-km grid cell to another and this problem will be more visible in areas of strong relief.

Bands that are more affected by atmospheric absorption will have poorer accuracies due to uncertainties in the correction for these effects as well as uncertainties in the input parameters needed to correct for these effects. Bands where these effects are most important are Bands 6-9. All bands except band 1 have significant absorption effects, but these effects in bands 2-5 are either readily characterized or small enough that uncertainties in both inputs and treatment of the absorption will be minimal. For regions of high humidity the uncertainties in this band can be more than 20% over bright surfaces. The effect is smaller over dark objects.

Problems and limitations
The primary limitation to this product is the lack of aerosol correction. This has been done so as to create a product corrected for solar zenith angle changes and gaseous absorption, but not to overcorrect for the unknown aerosol inputs. Thus, the current data product will always overestimate the surface reflectance and surface radiance.

A second limitation is the lack of an adjacency correction. For pixels with a surface reflectance much smaller than the overall reflectance of the region surrounding it, the retrieved reflectance will be larger than the actual value and for bright surfaces the reflectance will be underestimated. The effect is also larger at larger scattering optical depths.

The current release of the product is limited to some extent by the lack of aerosol information (as described in the next section). Thus, the quality of this data product will be strongly dependent upon the level of aerosol amount and wavelength. In addition, the quality of the product relies heavily on the accuracy of the Level 1B input radiances. The product still corrects for solar illumination effects, gaseous absorption, and molecular scattering.

For regions of small terrain relief, the expected uncertainties will be less than 0.01 reflectance units for surface reflectance less than 0.2. At higher reflectance, the uncertainty is expected to be 5% of the reflectance. For the current version of the software, these uncertainties will be true for cases where the aerosol scattering optical depth is less than 0.05. For larger aerosol loadings, the uncertainty will increase by approximately 0.005 in reflectance for each additional 0.05 aerosol optical depth increment at low reflectance (<0.20). At higher reflectance, each additional 0.05 optical depth increment will cause approximately an additional 5% uncertainty. These uncertainties have been derived for aerosol types with low absorption, thus, regions dominated by anthropogenic aerosols will have larger uncertainties. Because scattering optical depths decrease with wavelength, results will typically be more accurate at longer wavelengths where scattering effects are smaller.

Terrain relief also degrades the accuracy of this product. This effect is due to the following reasons. The first is that the algorithm itself breaks down due to an assumption of homogeneity in regions of large changes in terrain relief. In addition, there is additional uncertainty due to the resolution of the digital elevation model that is available. In most cases, it is expected that digital elevation models with spatial resolution better than 100 m will be available. In these cases, there will be no added uncertainty due to the elevation model. In other cases, the input model will revert to a 1-km model. There are two effects in this case. The first is that the poorer accuracy of the input elevation at the 15 and 30 m spatial scale will increase the uncertainty of the results. This effect will again be more important in regions of strong relief. The second effect is that the 1-km resolution will create a noticeable artifact in the corrected data when the data moves from a pixel within one 1-km grid cell to another and this problem will be more visible in areas of strong relief.

Bands that are more affected by atmospheric absorption will have poorer accuracies due to uncertainties in the correction for these effects as well as uncertainties in the input parameters needed to correct for these effects. Bands where these effects are most important are Bands 6-9. All bands except band 1 have significant absorption effects, but these effects in bands 2-5 are either readily characterized or small enough that uncertainties in both inputs and treatment of the absorption will be minimal. For regions of high humidity the uncertainties in this band can be more than 20% over bright surfaces. The effect is smaller over dark objects.

Problems and limitations
The primary limitation to this product is the lack of aerosol correction. This has been done so as to create a product corrected for solar zenith angle changes and gaseous absorption, but not to overcorrect for the unknown aerosol inputs. Thus, the current data product will always overestimate the surface reflectance and surface radiance.

A second limitation is the lack of an adjacency correction. For pixels with a surface reflectance much smaller than the overall reflectance of the region surrounding it, the retrieved reflectance will be larger than the actual value and for bright surfaces the reflectance will be underestimated. The effect is also larger at larger scattering optical depths.

4. Version History

The designation for Version 2.5 was upgraded to Provisional following further validation.
Version 2.7 upgraded the release designation for NCEP based processing to Validated.
Future releases will produce validated products using MODIS and MISR atmospheric inputs.