I need to estimate Land Surface Temperature (LST) of some areas in the Tropics but most of the available Landsat images for the areas are cloud covered (covered up to more than 40% in some cases). Are there some good methods to remove the cloud cover and sametime retain the accuracy required for LST estimation? Kindly, offer me your thoughts.
Elijah, as a student of Remote Sensing, so far I have never come across any method to remove 'optically thick clouds' from a satellite image and be able to see what is underneath. Of course there are techniques to 'clip/mask' cloudy pixels and replace that portion with a non-cloudy part from another satellite image of a different time period. But I think in case of LST computation it may not be sensible to use this technique. Why don't you try to look for other satellite images from a different sensor like ASTER or MODIS? Although given the fact that your study area is in tropics, it is highly unlikely that you will find 'cloud free' images. Nevertheless, you can give it a try. What is the spatial and temporal resolution of the satellite image you are looking for? Are you doing a time series analysis or a single time step analysis?
You can follow my question page where I had asked a similar question of removing cloud cover from satellite images but in the context of floods. It might be helpful.
https://www.researchgate.net/post/Can_you_help_me_in_analyzing_satellite_imagery_of_September_2014_Indo-Pak_Floods/1
There are different ways to work on cloud removal: here is one link
http://www.int-arch-photogramm-remote-sens-spatial-inf-sci.net/XXXIX-B7/335/2012/isprsarchives-XXXIX-B7-335-2012.pdf
and if you are keen to go further, you can even see through shadow, look for Tobias Schulmann, P. Zawar Reza and Marwan Katurji. I can't remember the exact title of their paper but they published it earlier this year.
Best
Christopher
From the principles of remote sensing, clouds act as reflectors, except in the RADAR. So the energy recorded by the sensor is not the true ground leaving value. In my view, if you succeed to find a tool for correcting the clouds, the obtained result, most probably, will be incorrect. I agree with James, you have to look for cloud free image if possible.
(1) clouds don't always absorb the full sepctrum;
(2) That's where Schulmann et al is new, they explain how they have used the projection of reflected part of the spectrum in shadows;
Certainly it isn't basic remote sensing, but it is feasible, depending on the level of cloud cover and the type of cloud.
Very many thanks to you all for your very insightful leads and comments. I just have to look for other sources. Thanks again.
Dear Njoku, you can check this two articles for your work. Authors are tries to remove the cloud by masking and surface temperature evolution.
https://www.sav.sk/journals/uploads/09181821Ogunbadewa_CGG-42-1_web.pdf
Article Estimation of Land Surface Temperature under Cloudy Skies Us...
Best regards,
- Badges
- Science topic
- Similar topics
- Geoscience
- Geoinformatics
- Geographic Information System