This algorithm was developed by Dr. Zhe Zhu at Texas Tech University.
Email address of primary contact: firstname.lastname@example.org.
The following computer codes have been made available (Current version 3.3) to implement the FMASK
How to use
(1) The Matlab version
(2) The Windows Standalone version
Any kinds of Landsat TM, ETM+, and OLI/TIR TIF images with a MTL.txt header in the same folder
There will be an image called XXXFmask that can be opened by ENVI. The image values are presenting the following classes are as follows: 0 ⇒ clear land pixel; 1 ⇒ clear water pixel; 2 ⇒ cloud shadow; 3 ⇒ snow; 4 ⇒ cloud; 255 ⇒ no observation
Descriptions of the code by the author:
The software called Fmask (Function of mask) is used for automated clouds, cloud shadows, and snow masking for Landsat TM/ETM+/OLI/TIRS images. For the most recent 3.3 version, there are two methods to run the Fmask algorithm. 1) If you have Matlab installed, you can run the Fmask Matlab code easily in any kinds of OS system (Windows, Linux, Unix, Mac); 2) If you do not have Matlab installed, you can still run the Fmask algorihtm in Windows by using standalone software provided. You need to have 64-bits machine with 8G memory to run the Fmask software, and it takes around 0.5 to 5 minutes to run one Landsat image, and the total processing time is mostly determined by the percent of cloud cover in the image. The Fmask algorithm has been well validated and the cloud detection has a overall accuracy of 96%. However,the 4% error can be still problematic for many applications. Using the extra temporal dimension of the satellite data can further improve the Fmask results.
Zhu, Z. and C.E. Woodcock. 2012. Object-based cloud and cloud shadow detection in Landsat imagery. Remote Sensing of Environment, 118, 83-94.
Zhu, Z., S. Wang, C.E. Woodcock. 2015. Improvement and expansion of the Fmask algorithm: Cloud, cloud shadow, and snow detection for Landsats 4–7, 8, and Sentinel 2 images, Remote Sensing of Environment, 159, 269-277.