tmconvert.pdf

2
Converting Landsat TM/ETM Digital Numbers to radiance and temperature values These steps convert digital numbers in the thermal bands of Landsat TM/ETM (Bands 5, 6 and 7) to radiance and temperature values. 1) Get .cpf (calibration parameter file) for your specific Landsat image from the EROS data center: http://edcwww.cr.usgs.gov/ 2) Convert the Landsat DN's to at-satellite radiance. This is easy, because the DN/radiance relationship is linear. However, you need to know the upper and lower bounds of radiance to scale the 255 values between. You get this radiance range from the .cpf files (I have included one for an example). Each cpf file covers a particular time period of ETM acquisition (usually a couple of months), so be sure your scene has the right cpf file. I just open the cpf files in Excel. Around line 368 are the 'group scaling parameters' which shows what the upper/lower bounds for radiance are in each band. Example (not to be used for your scene!): GROUP = SCALING_PARAMETERS GROUP = SCALING_PARAMETERS_LOW B1L_Lmin_Lmax = (-6.2,293.7) B2L_Lmin_Lmax = (-6.4,300.9) B3L_Lmin_Lmax = (-5.0,234.4) B4L_Lmin_Lmax = (-5.1,241.1) B5L_Lmin_Lmax = (-1.0,47.57) B6L_Lmin_Lmax = (0.0,17.04) B7L_Lmin_Lmax = (-0.35,16.54) B8L_Lmin_Lmax = (-4.7,243.1) END_GROUP = SCALING_PARAMETERS_LOW GROUP = SCALING_PARAMETERS_HIGH B1H_Lmin_Lmax = (-6.2,191.6) B2H_Lmin_Lmax = (-6.4,196.5) B3H_Lmin_Lmax = (-5.0,152.9) B4H_Lmin_Lmax = (-5.1,157.4) B5H_Lmin_Lmax = (-1.0,31.06) B6H_Lmin_Lmax = (3.2,12.65) B7H_Lmin_Lmax = (-0.35,10.80) B8H_Lmin_Lmax = (-4.7,158.3) END_GROUP = SCALING_PARAMETERS_HIGH So for Band 5, if the scaling parameter is 'LOW' then DN value 0 is equivalent to a radiance of -1, and 255 equals 47.57. Radiance is in units of watts per meter squared per steradian per micrometer.

Transcript of tmconvert.pdf

  • Converting Landsat TM/ETM Digital Numbers to radiance and temperature values These steps convert digital numbers in the thermal bands of Landsat TM/ETM (Bands 5, 6 and 7) to radiance and temperature values. 1) Get .cpf (calibration parameter file) for your specific Landsat image from the EROS data center: http://edcwww.cr.usgs.gov/ 2) Convert the Landsat DN's to at-satellite radiance. This is easy, because the DN/radiance relationship is linear. However, you need to know the upper and lower bounds of radiance to scale the 255 values between. You get this radiance range from the .cpf files (I have included one for an example). Each cpf file covers a particular time period of ETM acquisition (usually a couple of months), so be sure your scene has the right cpf file. I just open the cpf files in Excel. Around line 368 are the 'group scaling parameters' which shows what the upper/lower bounds for radiance are in each band. Example (not to be used for your scene!):

    GROUP = SCALING_PARAMETERS GROUP = SCALING_PARAMETERS_LOW B1L_Lmin_Lmax = (-6.2,293.7) B2L_Lmin_Lmax = (-6.4,300.9) B3L_Lmin_Lmax = (-5.0,234.4) B4L_Lmin_Lmax = (-5.1,241.1) B5L_Lmin_Lmax = (-1.0,47.57) B6L_Lmin_Lmax = (0.0,17.04) B7L_Lmin_Lmax = (-0.35,16.54) B8L_Lmin_Lmax = (-4.7,243.1) END_GROUP = SCALING_PARAMETERS_LOW GROUP = SCALING_PARAMETERS_HIGH B1H_Lmin_Lmax = (-6.2,191.6) B2H_Lmin_Lmax = (-6.4,196.5) B3H_Lmin_Lmax = (-5.0,152.9) B4H_Lmin_Lmax = (-5.1,157.4) B5H_Lmin_Lmax = (-1.0,31.06) B6H_Lmin_Lmax = (3.2,12.65) B7H_Lmin_Lmax = (-0.35,10.80) B8H_Lmin_Lmax = (-4.7,158.3) END_GROUP = SCALING_PARAMETERS_HIGH So for Band 5, if the scaling parameter is 'LOW' then DN value 0 is equivalent to a radiance of -1, and 255 equals 47.57. Radiance is in units of watts per meter squared per steradian per micrometer.

  • Now, you say, how do I know if the scaling parameter is 'Low' gain or 'High' gain. Well, you have to go to the EOS data gateway, find your scene in their inventory, and click on the 'Granule Attributes' for that particular scene - a pain, I know. It might be in the metafile, too. 3) I like to convert micrometers to meters in the radiance units, so multiply by 106 (just my preference) 4) If surface temperatures are what you're after (and I imagine they are), then you have to convert this at-satellite radiance to actual ground radiance, and then use the Planck function to convert ground radiance to ground temperature. Converting the at-satellite radiance (Rsatellite) to actual ground radiance (Rground) requires calculating the effects of surface emissivity (), atmospheric transmissivity (), upwelling radiance (Ru), and reflected radiance (Rr):

    rusatellite

    ground RRR

    R

    You can find emissivity of substances in Salisbury and D'Aria (1984). Atmospheric transmissivity and upwelling radiance can be estimated with atmospheric modeling programs like MODTRAN 4. Reflected radiance (probably not significant if using ETM Band 6, but will play a role in Band 5 and 7) is trickier - some use nearby pixels to garner an estimate (I'm not the person to ask). If you don't make this correction, your resulting temperature will only be a very rough approximation of surface temperature (although the effectiveness of this correction is arguable). Ground radiance in these bandwidths can then be converted to ground temperature using the inverse Planck function:

    )1]/ln([ 512

    Rcc

    T

    where T = brightness temperature, c1 = 3.742 x 10-16 W m2, c2 = 0.0144 m K, = band central wavelength, and R = spectral emittance (radiance). Make sure your units are all right. By the way, Band 61 and 62, as you probably know, use exactly the same detector I believe (same wavelength and bandwidth) but the gain is set differently (61 is set to 'low' gain, 62 is set to 'high' gain). Any questions feel free to contact me. Matt Patrick [email protected] Alaska Volcano Observatory