I already calculated the surface temperature using a Landsat 8 imagery through the built-in calibration tool in ENVI. Having computed the statistics for the resultant image, I noticed that the statistics are different between the two thermal bands based on which the temperature map has been created. More particularly, the min, max, mean temperature are quite different. I am wondering on which thermal band I should base my analysis and how these two actually compare to each other?
Band 10 is better in case of calculating the actual evapotranspiration as it give high gain where Band 11 give the low gain.
http://ac.els-cdn.com/S0034425715302650/1-s2.0-S0034425715302650-main.pdf?_tid=5136699a-f664-11e5-97ab-00000aacb361&acdnat=1459334822_c3b429445f057b915ae0439c3572c2de
In addition to the link above, check this article as well.
Article Retrieval of the reflectance and land surface temperature of...
You could also try to follow a different approach by using both thermal bands and trying in this way to remove unwanted terms/effects. An approach used commonly on water, but adapted also for land, is to use Split Window Algorithms (SWA).
See for example these two papers:
http://www.mdpi.com/2072-4292/7/1/647
http://www.mdpi.com/2072-4292/6/10/9829
Regards,
Giulio
For some satellites with two thermal bands, as Giulio mentioned SWA is the easiest and more accurate algorithm for LST estimation.
Thanks everyone for your valuable inputs. To update, apparently the USGS itself recommends not to use the band 11 at the moment and only benefit from band 10. This is due to inconsistencies reported in band 11. So, I guess in the meantime it would be safest to base LST analysis upon band 10.
I am attaching the link of USGS website saying so.
http://landsat.usgs.gov/Landsat8_Using_Product.php
Dear
Mahyar Masoudi,
We have used Landsat TM5 data from band-6 (10.4 to 12.5 micrometers) and following steps were used: (Ref: Estimation of Landsat TM Surface Temperature Using ERDASImagine Spatial Modeler by KoKo Lwin, SIS Tutorial Series. Copyright © 2009 by Division of Spatial Information Science)
Step1. Conversion of the Digital Number (DN) to
Spectral Radiance (L)
L = LMIN + (LMAX - LMIN) * DN / 255
Where
L = Spectral radiance
LMIN = 1.238 (Spectral radiance of DN value 1)
LMAX = 15.600 (Spectral radiance of DN value 255)
DN = Digital Number
Step2. Conversion of Spectral Radiance to
Temperature in Kelvin
Where
K1= Calibration Constant 1 (607.76)
K2 = Calibration Constant 2 (1260.56)
TB = Surface Temperature
Step3. Conversion of Kelvin to Celsius
TB = TB -273
However we have still not attempted using Landsat 8 Tm data from bands 10 and 11.
Dear Manik H Kalubarme,
Thanks a lot for your detailed reply. However, what you detailed only gives the at-satellite Brightness Temperature, which is of course a necessary medium step towards calculation of LST.
I am attaching the link to a paper detailing the procedure to calculate the LST from a satellite imagery. I hope you also find it useful.
http://www.sciencedirect.com/science/article/pii/S1001074207600412
You can find answers for your question in the following article - Derivation of Land Surface Temperature for Landsat-8 TIRS Using a Split Window Algorithm
Offer Rozenstein, Zhihao Qin, Yevgeny Derimian and Arnon Karnieli.
link - http://www.mdpi.com/1424-8220/14/4/5768
Regards,
Zbigniew
follow this article
https://www.researchgate.net/publication/301797360_Analyzing_land_surface_temperature_distribution_in_response_to_land_useland_cover_change_using_split_window_algorithm_and_spectral_radiance_model_in_Sundarban_Biosphere_Reserve_India
Article Analyzing land surface temperature distribution in response ...
you can estimate LST for any city using the below link
http://rslab.gr/downloads_LandsatLST.html
- Badges
- Science topic
- Similar topics
- Engineering
- Modeling