Fmask:称为Fmask（蒙版功能）的软件用于Landsats 4-8和Sentinel 2图像的自动云，云阴影和雪蒙版

# Fmask The software called Fmask (Function of mask) is used for automated **clouds**, **cloud shadows**, **snow**, and **water** masking for Landsats 4-8 and Sentinel 2 images. If you have any questions, please contact Zhe Zhu (zhe@uconn.edu) and Shi Qiu (shi.qiu@uconn.edu) at Department of Natural Resources and the Environment, University of Connecticut. **IMPORTANT:** This Github page **ONLY** includes the Matlab code for Fmask 4.3. **The Matlab package with GLOBAL AUXILIARY DATA (~1G)** is available at this [Google Drive](https://drive.google.com/drive/folders/1dfQRNASMiFnE4ipFAYVwMn98eK9B9BiQ?usp=sharing), where **autoFmask** is the main function for processing an image. **autoFmaskBacth** can process all Landsats 4-8 and Sentinel-2 images into a folder. Note that Mapping Toolbox in Matlab is required for using the source code. **For MATLAB versions later than 2020b, please delete the geotiffread.m and geotiffinfo.m in the Fmask Matlab package (standalones are not be affected)**. **Fmask 4.3 softwares** (including standalones with Graphical User Interface (GUI) and without GUI) on Windows and Linux (~1G without MCR and ~3G with MCR) are also ready to use now! It can be downloaded at this [Google Drive](https://drive.google.com/drive/folders/1dfQRNASMiFnE4ipFAYVwMn98eK9B9BiQ?usp=sharing). This 4.3 version has substantial better cloud, cloud shadow, and snow detection results for Sentinel 2 and better results (compared to the 3.3 version that is currently being used by USGS as the Collection 1 QA Band) for Landsats 4-8 . This one software can be used for **Landsats 4-8 Collection 1 (or Collection 2) Level 1 product (Digital Numbers)** and **Sentinel 2 Level-1C product (Top Of Atmosphere reflectance)** at the same time. **IMPORTANT:** Majority of the current Collection 1 Landsats 4-8 QA Band provided by USGS are derived form **3.3 Version of Fmask algorithm** based on default parameters (cloud probability is 22.5% and buffer pixel size is 3). For example, (1) The Cloud (bit 4) is based on Fmask cloud mask (0 is not cloud and 1 is cloud in Fmask); (2) The Cloud Confidence (bits 5-6) is based on Fmask cloud probability in which >22.5% is high (11), >12.5% is medium (10), and <12.5% is low (01) with 00 kept for future use; (3) Snow/ice Confidence (bits 9-10) and Cloud Shadow Confidence (bits 7-8) has only low confidence (01) and high confidence (11) which correspond to no and yes respectively in snow/ice and cloud shadow mask provided by Fmask. **IMPORTANT:** When making the accuracy assessment for Fmask, please dilate 3 pixels for cloud shadow, but no dilation for cloud, snow, and water. # 4.3 Version 1) Fixed the bug when GRIDobj reads geotiff with several tiffinfos (see GRIDobj.m). (Shi Qiu 10/15/2020) ----- 4.2 version below ---- (4.2 version can be download at this [Google Drive](https://drive.google.com/drive/folders/1bVwvlGDFOsWnVj5b3MqI5yqRDoi8g935?usp=sharing)) 2) Update Fmask tool for processing Landsat Collection 2 data; and allow the cloud probability thershold in a larger range such as [-100, 100] in the GUI version. (Shi Qiu 4/8/2020) ----- 4.1 version below ---- (4.1 version can be download at this [Google Drive](https://drive.google.com/open?id=1l84t_lbp5Cp5v8L_Rbzk3WQHap7MdFwc) for Matlab code and this [Google Drive](https://drive.google.com/drive/folders/1oVefP9G-TD2vhoCaaKCxQjvAnUlrwB19?usp=sharing) for standalones) 3) The cloud shadow mask over water would not be provided at default settings since this will be less meaningful to use and very time-consuming to process. At the same time, fixed the bug that the auxiliary data may not be used for some Sentinel-2 images, of which the extent in the metadata is defined in [0 360] rather than [-180 180]. (Shi Qiu 3/17/2020) ----- 4.0 version below ---- (4.0 version can be download at this [Google Drive](https://drive.google.com/open?id=1SXBnEBDJ1Kbv7IQ9qIgqloYHZfdP6O1O)) 4) Fixed the bug that the cloud shadows in Sentinel-2 imagery would be projected along a wrong direction when solar azimuth angle > 180 degrees. (Shi Qiu 01/19/2019) 5) Integrated Cloud Displacement Index (CDI) into this Fmask 4.0 for better seperating clouds from bright surfaces especail for Sentinel-2. The CDI was specially designed to separate clouds from bright surfaces based on the view angle parallax of the three near infrared bands (band 7, 8 and 8a) ([Frantz et al., 2018](https://doi.org/10.1016/j.rse.2018.04.046)). (Shi Qiu and Zhe Zhu 06/03/2018) 6) Revised the method to identify the potential false positive cloud pixels. (Shi Qiu and Zhe Zhu 05/23/2018) 7) Restricted the height of the clouds located in the scene boundary into the predicted cloud height derived from its neighboring clouds. (Shi Qiu 04/05/2018) 8) Removed the overlap between the predicted cloud shadow and the potential cloud shadow layer for cloud shadow detection. (Shi Qiu and Zhe Zhu 03/29/2018) 9) Fixed the bug that the reading blue band using GRIDobj may lead to Nan value for Landsat images. (Shi Qiu 03/26/2018) 10) Improved the computational efficiency specially for cloud shadow matching procedure. (Zhe Zhu and Shi Qiu 03/24/2018) 11) Released Fmask 4.0 beta version. (Shi Qiu, Zhe Zhu, and Binbin He 03/22/2018) Please cite the following papers: **paper 1: Qiu S., et al., Fmask 4.0: Improved cloud and cloud shadow detection in Landsats 4-8 and Sentinel-2 imagery, Remote Sensing of Environment, (2019), [doi.org/10.1016/j.rse.2019.05.024](https://doi.org/10.1016/j.rse.2019.05.024) (paper for Fmask 4.0).** **paper 2: Zhu, Z. and Woodcock, C. E., 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 (2014) [doi:10.1016/j.rse.2014.12.014](https://doi:10.1016/j.rse.2014.12.014) (paper for Fmask version 3.2).** **paper 3: Zhu, Z. and Woodcock, C. E., Object-based cloud and cloud shadow detection in Landsat imagery, Remote Sensing of Environment (2012), [doi:10.1016/j.rse.2011.10.028](https://doi:10.1016/j.rse.2011.10.028) (paper for Fmask version 1.6).** **paper 4: Qiu S., et al. Improving Fmask cloud and cloud shadow detection in mountainous area for Landsats 4–8 images, Remote Sensing of Environment (2017), [doi.org/10.1016/j.rse.2017.07.002](https://doi.org/10.1016/j.rse.2017.07.002) (paper for Mountainous Fmask ([MFmask](https://github.com/qsly09/MFmask)), that has been integrated into this Fmask 4.0).** **paper 5: Qiu, S., et al., Making Landsat Time Series Consistent: Evaluating and Improving Landsat Analysis Ready Data, Remote Sensing (2019), [doi.org/10.3390/rs11010051](https://doi.org/10.3390/rs11010051) (First paper introducing Fmask 4.0 for improving LTS consistency).** **The training data** for the Fmask 4.0 are availble at the following link: https://landsat.usgs.gov/landsat-8-cloud-cover-assessment-validation-data **The validation data** for the Fmask 4.0 are comming soon. After running Fmask 4.0, there will be an image called XXXFmask.tif. The image values are presenting the following classes: 0 => clear land pixel 1 => clear water pixel 2 => cloud shadow 3 => snow 4 => cloud 255 => no observation # 3.3 Version Updates (since 3.2): 1) Bug fixed in cloud and cloud shadow matching algorithm. 2) Bug fixed in building 3D cloud objects for small clouds (radium <= 3) for versions where TIRS band is used. The 3.3 version of **Matlab code** for **Landsats 4-8 in which Landsat 8 has valid TIRS band** can be downloaded at this [link] (https://www.dropbox.com/sh/riruwk721zbl0he/AAAe_ccQiNS7_wHNC3HadOqRa?dl=0) The 3.3 version of **Windows stand alone software** for **Landsats 4-8 in which Landsat 8 has valid TIRS band** can be downloaded at this [link] (https://www.dropbox.com/sh/ylzub1uzosqidwy/AAC3zmk4M3DSbSoS2OLhR5r9a?dl=0) (provided by Sean Griffin segriffin@gmail.com) The 3.3 version of **Matlab code** for **Landsats 4-8 in which Landsat 8 does not have valid TIRS band** (zeor v