Landsat 8 Bands Descriptive Essay

The Landsat 8 Pre-Collection Quality Assessment (QA) band is an important addition to Landsat 8 data files. Each pixel in the QA band contains integers that represent bit-packed combinations of surface, atmosphere, and sensor conditions that can affect the overall usefulness of a given pixel.

The Pre-Collection QA band is available for Landsat 8 OLI/TIRS Pre-Collection data only. The Landsat Collection 1 Quality Band web page provides information for Collection 1 Landsat 4-5 Thematic Mapper (TM), Landsat 7 Enhanced Thematic Mapper Plus (ETM+) and Landsat 8 Operational Land Imager (OLI)-only and Landsat 8 OLI/Thermal Infrared Sensor (OLI/TIRS)- combined data.

The table below displays the bits that are currently populated for Pre-Collection Landsat 8 OLI/TIRS scenes:

What are QA Bits?

The bit-packed information in the QA bands is translation of binary strings. As a simple example, the integer value “1” translates to the binary value “0001.” The binary value “0001” has 4 bits, written right to left as bits 0 (“1”), 1 (“0”), 2 (“0”), and 3 (“0”). Each of the bits 0-3 represents a yes/no indication of a physical value.

Used effectively, QA bits improve the integrity of science investigations by indicating which pixels might be affected by instrument artifacts or subject to cloud contamination. For example, NDVI calculated over pixels containing clouds will show anomalous values. If such pixels were included in a phenology study, the results might not show the true characteristics of seasonal vegetation growth. Cloud contaminated pixels will lower NDVI values, and measures like the timing of ‘green up’ or peak maturity would appear later than they actually occurred. A worse consequence would be that the reported reduction of vegetation growth would be taken as an indicator of environmental change, potentially prompting unnecessary land management policies or practices.

The Landsat QA Tools are no-cost tools that will extract bit-packed information in the Pre-Collection Landsat 8 QA band for easy interpretation.

Landsat 8 Pre-Collection Level-1 QA Band File (BQA.TIF)

Rigorous science applications seeking to optimize the value of pixels used in a study will find QA bits useful as a first level indicator of certain conditions. Otherwise, users are advised that this file contains information that can be easily misinterpreted and it is not recommended for general use.

Landsat 8 Pre-Collection Level-1 data products include a 16-bit QA file (.TIF). Robust image processing software capable of handling 16-bit data is necessary to compute statistics of the number of pixels containing each of the designated bits.

Quality Band (BQA.TIF) displayed as .jpg for reference only
LDCM sample data Path 45 Row 30 Acquired April 23, 2013.

The QA image can be stretched to emphasize the light ("1"s) and dark ("0") pixels for a quick look at general quality conditions. In the Crater Lake, Oregon image above, the lighter pixels are likely to be affected by a quality condition, in this case snow or clouds.

The pixel values in the QA file must be translated to 16-bit binary form to be used effectively. The gray shaded areas in the table below show the bits that are currently being populated in the Level-1 QA Band, and the conditions each describe. None of the currently populated bits are expected to exceed 80% accuracy in their reported assessment at this time.

For the single bits (0, 1, 2, and 3):

  • 0 = No, this condition does not exist
  • 1 = Yes, this condition exists

For the Water bits (4-5)

  • 00 = No, this condition does not exist
  • 10 = Yes, this condition exists

For the Snow/Ice bits (10-11):

  • 00 = No, this condition does not exist
  • 11 = Yes, this condition exists

These conditions are detected by simple algorithms that currenlty indicate yes/no. They were allotted two bits each to leave room for more complex algorithms in the future.

The othe double bits (6-7, 8-9, 12-13, and 14-15), read from left to right, represent levels of confidence that a condition exists:

  • 00 = “Not Determined” = Algorithm did not determine the status of this condition
  • 01 = “No” = Algorithm has low to no confidence that this condition exists (0-33 percent confidence)
  • 10 = “Maybe” = Algorithm has medium confidence that this condition exists (34-66 percent confidence)
  • 11 = “Yes” = Algorithm has high confidence that this condition exists (67-100 percent confidence).

For example, a pixel with a value "56320" translates to the 16-bit binary string "1101 1100 0000 0000." Reading the binary string from right to left and using the table above as an interpretation legend, this pixel is:

  • Bit 0 = 0 = not fill
  • Bit 1 = 0 = not a dropped frame
  • Bit 2 = 0 = not terrain occluded
  • Bit 3 = 0 = unused
  • Bit 4-5 = 01 = not water
  • Bit 6-7 = 00 = unused
  • Bit 8-9 = 00 = unused
  • Bit 10-11 = 01 = snow/ice
  • Bit 12-13 = 10 = not cirrus
  • Bit 14-15 = 11 = cloudy

Certain values occur regularly and can be interpreted without unpacking them into 16-bit strings and using the table above as a reference. Some common pixel values and their meanings are included in the table below.

LandsatLook Quality Image (.jpg)

The Landat 8 Pre-Collection LandsatLook 8-bit Quality Image (.jpg) is available to download when downloading Pre-Collection Landsat 8 data products. This file provides a quick view of the quality of the pixels to determine which scene would work best for each user's application. Only the highest confidence conditions are used to create the LandsatLook Quality image. Similar as stated above, this image may not be useful to all users. (Information on LandsatLook Images)

The table below gives the bits and colors associated with the LandsatLook Quality Image:



Landsat Look "Quality" Image (QA.jpg) displayed as .jpg for reference only
LDCM/Landsat 8 sample data Path 45 Row 30 Acquired April 23, 2013.

NAME

i.landsat8.swlst - Practical split-window algorithm estimating Land Surface Temperature from Landsat 8 OLI/TIRS imagery (Du, Chen; Ren, Huazhong; Qin, Qiming; Meng, Jinjie; Zhao, Shaohua. 2015)

KEYWORDS

imagery, split window, column water vapor, land surface temperature, lst, landsat8

SYNOPSIS

i.landsat8.swlst

i.landsat8.swlst --help

i.landsat8.swlst [-iktcn] [mtl=filename] [prefix=basename] [b10=name] [b11=name] [prefix_bt=basename] [t10=name] [t11=name] [qab=name] [qapixel=pixelvalue[,pixelvalue,...]] [clouds=name] [emissivity=name] [emissivity_out=name] [delta_emissivity=name] [delta_emissivity_out=name] [landcover=name] [emissivity_class=string] lst=namewindow=integer [cwv=name] [--overwrite] [--help] [--verbose] [--quiet] [--ui]

Flags:

-i
Print out model equations, citation
-k
Keep current computational region settings
-t
Time-stamping the output LST (and optional CWV) map
-c
Convert LST output to celsius degrees, apply color table
-n
Set zero digital numbers in b10, b11 to NULL | ToDo: Perform in copy of input input maps!
--overwrite
Allow output files to overwrite existing files
--help
Print usage summary
--verbose
Verbose module output
--quiet
Quiet module output
--ui
Force launching GUI dialog

Parameters:

mtl=filename
Landsat8 metadata file (MTL)
prefix=basename
OLI/TIRS band names prefix
Prefix of Landsat8 OLI/TIRS band names
b10=name
TIRS 10 (10.60 - 11.19 microns)
b11=name
TIRS 11 (11.50 - 12.51 microns)
prefix_bt=basename
Prefix for output at-satellite brightness temperature maps (K)
Prefix for brightness temperature maps (K)
t10=name
Brightness temperature (K) from band 10 | Overrides 'b10'
t11=name
Brightness temperature (K) from band 11 | Overrides 'b11'
qab=name
Landsat 8 Quality Assessment band
qapixel=pixelvalue[,pixelvalue,...]
Quality assessment pixel value for which to build a mask | Source: <http://landsat.usgs.gov/L8QualityAssessmentBand.php>.
Default: 61440
clouds=name
A raster map applied as an inverted MASK | Overrides 'qab'
emissivity=name
Land surface emissivity map | Expert use, overrides retrieving average emissivity from landcover
emissivity_out=name
Name for output emissivity map | For re-use as "emissivity=" input in subsequent trials with different spatial window sizes
delta_emissivity=name
Emissivity difference map for Landsat8 TIRS channels 10 and 11 | Expert use, overrides retrieving delta emissivity from landcover
delta_emissivity_out=name
Name for output delta emissivity map | For re-use as "delta_emissivity=" in subsequent trials with different spatial window sizes
landcover=name
FROM-GLC products covering the Landsat8 scene under processing. Source <http://data.ess.tsinghua.edu.cn/>.
emissivity_class=string
Retrieve average emissivities only for a single land cover class (case sensitive) | Expert use
Options: Cropland, Forest, Grasslands, Shrublands, Wetlands, Waterbodies, Tundra, Impervious, Barren, Snow, Random
lst=name [required]
Name for output Land Surface Temperature map
Default: lst
window=integer [required]
Odd number n sizing an n^2 spatial window for column water vapor retrieval | Increase to reduce spatial discontinuation in the final LST
Default: 7
cwv=name
Name for output Column Water Vapor map | Optional

DESCRIPTION

i.landsat8.swlst is an implementation of a robust and practical Slit-Window (SW) algorithm estimating land surface temperature (LST), from the Thermal Infra-Red Sensor (TIRS) aboard Landsat 8 with an accuracy of better than 1.0 K. [1]

Overview

The components of the algorithm estimating LST values are at-satellite brightness temperature (BT); land surface emissivity (LSE); and the coefficients of the main Split-Window equation (SWC).

LSEs are derived from an established look-up table linking the FROM-GLC classification scheme to average emissivities. The NDVI and the FVC are not computed each time an LST estimation is requested. Read [0] for details.

The SWC depend on each pixel's column water vapor (CWV). CWV values are retrieved based on a modified Split-Window Covariance-Variance Matrix Ratio method (MSWCVMR) [1, 2]. Note, the spatial discontinuity found in the images of the retrieved CWV, is attributed to the data gap in the images caused by stray light outside of the FOV of the TIRS instrument [2]. In addition, the size of the spatial window querying for CWV values in adjacent pixels, is a key parameter of the MSWCVMR method. It influences accuracy and performance.

At-satellite brightness temperatures are derived from the TIRS channels 10 and 11. Prior to any processing, these are filtered for clouds and their quantized digital numbers converted to at-satellite temperature values.

Hence, to produce an LST map, the algorithm requires at minimum:

  • TIRS bands 10 and 11
  • the acquisition's metadata file (MTL)
  • a Finer Resolution Observation & Monitoring of Global Land Cover (FROM-GLC) product

Details

A new refinement of the generalized split-window algorithm proposed by Wan (2014) [19] is added with a quadratic term of the difference amongst the brightness temperatures (Ti, Tj) of the adjacent thermal infrared channels, which can be expressed as (equation 2 in [0])

where:

  • and are Top of Atmosphere brightness temperatures measured in channels (~11.0 microns) and (~12.0 µm), respectively
  • from http://landsat.usgs.gov/band_designations_landsat_satellites.php:
    • Band 10, Thermal Infrared (TIRS) 1, 10.60-11.19, 100*(30)
    • Band 11, Thermal Infrared (TIRS) 2, 11.50-12.51, 100*(30)
  • e is the average emissivity of the two channels (i.e., )
  • De is the channel emissivity difference (i.e., )
  • (k = 0, 1, ... 7) are the algorithm coefficients derived from a simulated dataset.

In the above equations,

  • (k = 0, 1...6) and (k = 1, 2, 3, 4) are the algorithm coefficients;
  • is the Column Water Vapor;
  • and are the average emissivity and emissivity difference of two adjacent thermal channels, respectively, which are similar to Equation (2);
  • and (k = 0 and 1) is related to the influence of the atmospheric transmittance and emissivity, i.e., .

Comparing to other split-window algorithms

From the paper:

Note that the algorithm (Equation (6a)) proposed by Jimenez-Munoz et al. added column water vapor (CWV) directly to estimate LST. Rozenstein et al. used CWV to estimate the atmospheric transmittance (, ) and optimize retrieval accuracy explicitly. Therefore, if the atmospheric CWV is unknown or cannot be obtained successfully, neither of the two algorithms in Equations (6a) and (6b) will work. By contrast, although the current algorithm also needs CWV to determine the coefficients, it still works for unknown CWVs because the coefficients are obtained regardless of the CWV, as shown in Table 1 [0].

NOTES

Cloud Masking

The first important step of the algorithm is cloud screening. The module offers two ways to achieve this:

  1. use of the Quality Assessment band and some user-defined QA pixel value
  2. use an external cloud map as an inverted MASK

Calibration of TIRS channels 10, 11

Conversion to Spectral Radiance

Conversion of Digital Numbers to TOA Radiance. OLI and TIRS band data can be converted to TOA spectral radiance using the radiance rescaling factors provided in the metadata file:

where:

  • = TOA spectral radiance (Watts/( m2 * srad * microns))
  • = Band-specific multiplicative rescaling factor from the metadata (RADIANCE_MULT_BAND_x, where x is the band number)
  • = Band-specific additive rescaling factor from the metadata (RADIANCE_ADD_BAND_x, where x is the band number)
  • = Quantized and calibrated standard product pixel values (DN)

Conversion to at-Satellite Temperature

Conversion to At-Satellite Brightness Temperature TIRS band data can be converted from spectral radiance to brightness temperature using the thermal constants provided in the metadata file:

where:

  • = At-satellite brightness temperature (K)
  • = TOA spectral radiance (Watts/(m^2 * srad * microns)), below 'DUMMY_RADIANCE'
  • = Band-specific thermal conversion constant from the metadata (K1_CONSTANT_BAND_x, where x is the band number, 10 or 11)
  • = Band-specific thermal conversion constant from the metadata (K2_CONSTANT_BAND_x, where x is the band number, 10 or 11)

Land Surface Emissivity

Determination of LSEs (overview of Section 3.2)

  1. The FROM-GLC (30m) contains 10 types of land covers (cropland, forest, grassland, shrubland, wetland, waterbody, tundra, impervious, barren land and snow-ice).

  2. Deriving emissivities for each land cover class by using different combinations of three BRDF kernel models (geometrical, volumetric and specular models)

  3. Vegetation and ground emissivity spectra for the BRDF models selected from the MODIS University of California, Santa Barbara (UCSB) Emissivity Library

  4. Estimating FVC (to obtain emissivity of land cover with temporal variation) from NDVI based on Carlson (1997) and Sobrino (2001)

  5. Finally, establishing the average emissivity Look-Up table

Column Water Vapor

Retrieving atmospheric CWV from Landsat8 TIRS data based on the modified split-window covariance and variance ratio (MSWCVR).

Algorithm Coefficients (overview of Section 3.1)

  1. The CWV is divided into 5 sub-ranges with an overlap of 0.5 g/cm2 between 2 adjacent sub-ranges: [0.0, 2.5], [2.0, 3.5], [3.0, 4.5], [4.0, 5.5] and [5.0, 6.3] g/cm2.

  2. The CWV is retrieved from a modified split-window covariance and variance ratio method.

  3. However, given the somewhat unsuccessful CWV retrieval, a group of coefficients for the entire CWV range is calculated to ensure the spatial continuity of the LST product.

Modified Split-Window Covariance-Variance Method

With a vital assumption that the atmosphere is unchanged over the neighboring pixels, the MSWCVR method relates the atmospheric CWV to the ratio of the upward transmittances in two thermal infrared bands, whereas the transmittance ratio can be calculated based on the TOA brightness temperatures of the two bands. Considering N adjacent pixels, the CWV in the MSWCVR method is estimated as:

  • (3a)

where:

  • ~

In Equation (3a):

  • , and are coefficients obtained from simulated data;
  • is the band effective atmospheric transmittance;
  • is the number of adjacent pixels (excluding water and cloud pixels) in a spatial window of size (i.e., );
  • and are top of atmosphere brightness temperatures (K) of bands and for the th pixel;
  • and are the mean (or median -- not implemented yet) brightness temperatures of the pixels for the two bands.

TIRS channels are originally of 100m spatial resolution. However, bands 10 and 11 are resampled, via a cubic convolution filter, to 30m. Consequently, an appropriately sized spatial window is required for a meaningful CWV estimation attempt. The spatial window should be composed by a number of pixels stretching over an area that accounts for several adjacent 100m-sized pixels. Note, while the CWV estimation accuracy increases with larger windows (up to a certain level), the performance (speed) of the module decreases greatly.

The regression coefficients:

  • = -9.674
  • = 0.653
  • = 9.087

where obtained by:

  • 946 cloud-free TIGR atmospheric profiles,
  • the new high accurate atmospheric radiative transfer model MODTRAN 5.2
  • simulating the band effective atmospheric transmittance Model analysis indicated that this method will obtain a CWV RMSE of about 0.5 g/cm^2.

The algorithm will not cause significant uncertainty to the final LST retrieval with known CWV, but it will lead some error to the LST result for the cases without input CWV. To reduce this effect, the authors are trying to find more representative profiles to optimize the current algorithm.

Details about the columnw water vapor retrieval can be found in:

Ren, H.; Du, C.; Qin, Q.; Liu, R.; Meng, J.; Li, J. Atmospheric water vapor retrieval from landsat 8 and its validation. In Proceedings of the IEEE International Geosciene and Remote Sensing Symposium (IGARSS), Quebec, QC, Canada, July 2014; pp. 3045--3048.

Split-Window Algorithm

The algorithm removes the atmospheric effect through differential atmospheric absorption in the two adjacent thermal infrared channels centered at about 11 and 12 microns. The linear or non-linear combination of the brightness temperatures is finally applied for LST estimation based on the equation:

To reduce the influence of the CWV error on the LST, for a CWV within the overlap of two adjacent CWV sub-ranges, we first use the coefficients from the two adjacent CWV sub-ranges to calculate the two initial temperatures and then use the average of the initial temperatures as the pixel LST.

For example, the LST pixel with a CWV of 2.1 g/cm2 is estimated by using the coefficients of [0.0, 2.5] and [2.0, 3.5]. This process initially reduces the delta-LSTinc and improves the spatial continuity of the LST product.

EXAMPLES

At minimum, the module requires the following in order to derive a land surface temperature map:

  1. The Landsat8 scene's acquisition metadata (MTL file)

  2. Bands 10, 11 and QA

  3. A FROM-GLC product for the same Path and Row as the Landsat scene to be processed

The shortest call for processing a complete Landsat8 scene normally is:

where:

  • the name of the MTL metadata file (normally with a extension)

  • the prefix of the band names imported in GRASS GIS' data base

  • the name of the FROM-GLC map that covers the extent of the Landsat8 scene under processing

  • the flag will set zero digital number values, which may represent NoData in the original bands, to NULL. This option is probably unnecessary for smaller regions in which there are no NoData pixels present.

The pixel value 61440 is selected automatically to build a cloud mask. At the moment, only a single pixel value may be requested from the Quality Assessment band. For details, refer to [http://landsat.usgs.gov/L8QualityAssessmentBand.php USGS' webpage for Landsat8 Quality Assessment Band]

is an important option. It defines the size of the spatial window querying for column water vapor values. Small window sizes introduce a spatial discontinuation effect in the final LST image. Larger window sizes lead to more accurate results, at the cost of performance. However, too large window sizes should be avoided as they would include large variations of land and atmospheric conditions. In [2] it is stated:

A small window size n (N = n * n, see equation (1a)) cannot ensure a high correlation between two bands' temperatures due to the instrument noise. In contrast, the size cannot be too large because the variations in the surface and atmospheric conditions become larger as the size increases.

An example instructing a spatial window of size 9^2 is:

In order to restrict the processing in to the currently set computational region, the flag can be used:

The Landsat8 scene's time and date of acquisition may be applied to the LST (and to the optionally requested CWV) map via the flag.

The output land surface temperature map maybe be delivered in Celsius degrees (units and appropriate color table) via the flag:

A user defined map for clouds, instead of relying on the Quality Assessment band, can be used via the option:

Using the option, the in-between at-satellite brightness temperature maps may be saved for re-use in sub-sequent trials via the and options. Using the and options, will skip the conversion from digital numbers for bands B10 and B11. Alternatively, any existing at-satellite brightness temperature maps maybe used via the options. For example using the option instead of :

or using both and :

A faster run is achieved by using existing maps for all in-between processing steps: at-satellite temperatures, cloud and emissivity maps.

Expert users may need to request for a fixed average surface emissivity, in order to perform the algorithm for a single land cover class (one from the classes defined in the FROM-GLC classification scheme) via the option. Consequently, cannot be used at the same time with the option.

A transparent run-through of what kind of and how the module performs its computations, may be requested via the use of both the and flags:

The above will print out a description for each individual processing step, as well as the actual mathematical epxressions applied via GRASS GIS' module.

Example figures

TODO

  • Go through Submitting Python
  • Proper command history tracking.
  • Deduplicate code where applicable
  • Test compiling in other systems
  • Improve documentation

REFERENCES

  • [0] Du, Chen; Ren, Huazhong; Qin, Qiming; Meng, Jinjie; Zhao, Shaohua. 2015. "A Practical Split-Window Algorithm for Estimating Land Surface Temperature from Landsat 8 Data." Remote Sens. 7, no. 1: 647-665. http://www.mdpi.com/2072-4292/7/1/647/htm

  • [1] Huazhong Ren, Chen Du, Qiming Qin, Rongyuan Liu, Jinjie Meng, and Jing Li. "Atmospheric Water Vapor Retrieval from Landsat 8 and Its Validation." 3045--3048. IEEE, 2014.

  • [2] Ren, H., Du, C., Liu, R., Qin, Q., Yan, G., Li, Z. L., & Meng, J. (2015). Atmospheric water vapor retrieval from Landsat 8 thermal infrared images. Journal of Geophysical Research: Atmospheres, 120(5), 1723-1738.

SEE ALSO

i.emissivity

AUTHORS

Nikos Alexandris

SOURCE CODE

Available at: i.landsat8.swlst source code (history)

Main index | Imagery index | Topics index | Keywords index | Graphical index | Full index

© 2003-2018 GRASS Development Team, GRASS GIS 7.4.1svn Reference Manual

Comments

Leave a Reply

Your email address will not be published. Required fields are marked *